Komunitas PDF

GAMBARAN DERAJAT KLINIS TERHADAP FAKTOR RESIKO PADA PASIEN
REAKSI HIPERSENSITIVITAS DI RUMAH SAKIT SANGLAH DENPASAR PERIODE

2012-2013
I Gede Aswin Arinata1, I Ketut Suardamana2

Program Studi Pendidikan Dokter, Fakultas Kedokteran Universitas Udayana1
Bagian Allergi dan Imunologi Penyakit Dalam RSUP Sanglah2
ABSTRAK
Derajat klinis dari reaksi hipersensitivitas dibedakan menjadi derajat ringan, derajat sedang, dan
derajat berat. Secara epidemiologi dikatakan prevalensi reaksi hipersensitivitas 1-2% dari
populasi penduduk dunia. Faktor risiko yang berkaitan dengan beratnya derajat klinis antara lain
usia tua, jenis kelamin, riwayat alergi/asma, dan allergen. Tujuan penelitian ini adalah untuk
mengetahui gambaran derajat klinis terhadap faktor resiko pada pasien reaksi hipersensitivitas di
rumah sakit sanglah denpasar tahun 2012 hingga 2013. Penelitian ini menggunakan rancangan
studi deskriptif cross sectional. Data diperoleh secara sekunder melalui rekam medis yang
tercatat di Bagian Allergy dan Imunology Penyakit Dalam RSUP Sanglah. Analisis data
dilakukan secara deskriptif. Pada penelitian ini didapatkan 144 sampel. Jumlah jenis kelamin
yang didapatkan berimbang antara laki-laki dengan perempuan yaitu 73 pasien (50,7%) dan 71
pasien (49,3%). Derajat klinis berat didapatkan paling banyak pada laki-laki yaitu 24 pasien
(16,7%). Usia rata-rata sampel adalah 39,5 tahun. Derajat klinis berat ditemukan paling banyak
pada usia 41-60 tahun yaitu 16 pasien (11,1%). Jumlah pasien yang memiliki riwayat
allergy/asma lebih banyak memiliki derajat klinis berat yaitu 34 pasien (23,6%). Obat paling
banyak menyebabkan reaksi hipersensitivitas yaitu sebanyak 114 pasien (79,2%) dengan derajat
klinis berat yaitu 33 pasien (22,9%). Pasien dengan allergen makanan paling banyak mengalami
derajat klinis sedang yaitu sebanyak 12 pasien (8,3%). Allergen lain terjadi pada 7 pasien (4,9%)
dengan derajat klinis berat sebanyak 5 pasien (3,5%).
Kata Kunci: Reaksi Hipersensitivitas, faktor risiko, derajat klinis
1
DESCRIPTION OF CLINICAL DEGREE OF THE RISK FACTORS IN PATIENTS WITH
HYPERSENSITIVITY REACTIONS AT SANGLAH HOSPITAL, DENPASAR, IN 2012-2013
ABSTRACT
Clinical degree of hypersensitivity reactions can be divided into mild, moderate, and severe
degree. In epidemiology said the prevalence of hypersensitivity reactions 1-2 % of the world
population. Risk factors associated with the severity of clinical degrees include older age, sex,
history of allergy/asthma, and allergen. The purpose of this study was to describe the clinical
degree of the risk factors in patients with hypersensitivity reactions in Sanglah hospital in 2012
to 2013. This study used a descriptive cross-sectional study design. Data obtained secondarily
through medical records listed in Allergy and Imunology division of Sanglah. The data were
analyzed descriptively. There are 144 samples in this study. The number of acquired gender
balanced between men with women that 73 patients (50.7%) and 71 patients (49.3%). Severe
clinical degrees obtained at most in men is 24 patients (16.7%). The average age of the sample
was 39.5 years. Severe clinical degrees found at most in the 41-60 years age is 16 patients
(11.1%). The number of patients who have a history of allergy/asthma have more severe clinical
degrees to 34 patients (23.6%). Drugs most likely to cause hypersensitivity reactions as many as
114 patients (79.2%) with severe clinical degree that 33 patients (22.9%). Patients with food
allergens most experienced with moderate clinical degrees were as many as 12 patients (8.3%).
Other Allergen occurred in 7 patients (4.9%) with severe clinical degrees by 5 patients (3.5%).
Keyword : Hipersensitivity reactions, risk factors, clinical degree
PENDAHULUAN
Reaksi hipersensitivitas akut adalah reaksi angioedema/edema periorbita. Derajat
alergi pasca paparan yang melibatkan sedang yang melibatkan sistem respirasi,
sistem kulit/mukosa dan jaringan bawah kardiovaskuler, gastrointestinal seperti sesak
kulit. Sedangkan anaphylaksis merupakan nafas, stridor, mengi, mual, muntah, pusing
reaksi alergi sistemik akut yang melibatkan (pre syncope), rasa tidak enak di
dua sistem organ atau lebih (sistem tenggorokan dan dada serta nyeri perut.
kulit/mukosa dan jaringan bawah kulit, Derajat berat sudah terjadi hipoksia,
sistem respirasi, sistem kardiovaskular, hipotensi, syok dan manifestasi neurologis
sistem gastrointestinal).1 seperti sianosis (SpO2 90%), hipotensi
Derajat klinis dari reaksi (SBP<90 mmHg pada dewasa), kolaps,
hipersensitivitas dibedakan oleh Brown, penurunan kesadaran dan inkontinensia.
menjadi derajat ringan yang hanya Reaksi dengan derajat ringan dikenal
melibatkan kulit dan jaringan dibawah sebagai reaksi hipersensitivitas akut,
kulit seperti eritema generalisata, urtikaria,
2
sedangkan untuk derajat sedang dan berat terjangkau adalah pasien yang menderita
merupakan gambaran klinis anaphylaksis.1 reaksi hipersensitivitas di RSUP Sanglah
Secara epidemiologi dikatakan Denpasar dari tahun 2012 sampai dengan
prevalensi reaksi hipersensitivitas 1-2% dari 2013. Sampel dipilih berdasarkan kriteria
populasi penduduk dunia.2 Di Denmark inklusi dan eksklusi. Kriteria inklusi yang
dikatakan terdapat 3,2 kasus reaksi dipakai adalah pasien dengan diagnosis
hipersensitivitas derajat berat per 100.000 reaksi hipersensitivitas yang memiliki
orang/tahun, di Jerman dikatakan terdapat catatan lengkap rekam medisnya dan berusia
9,8 kasus reaksi hipersensivitas per 100.000 12 tahun keatas. Kriteria eksklusi adalah
orang/tahun. Di Eropa berkisar dari 1-3 pasien dengan penyulit seperti hipertensi,
kasus per 10.000 penduduk.3,4 gagal jantung kongestif, penyakit jantung
Faktor risiko yang berkaitan dengan koroner, penyakit paru obstruktif kronik,
beratnya derajat klinis antara lain usia tua, penyakit ginjal kronis, diabetes mellitus,
jenis kelamin, allergen (obat, makanan, Human Immunodeficiency Virus
sengatan serangga) riwayat alergi dan (HIV)/Acquired Immune Deficiency
asma, dan adanya penyulit atau penyakit Syndrome (AIDS), sepsis, syok sepsis serta
sistemik.5,6 yang tidak memiliki catatan lengkap rekam
Tujuan penelitian ini adalah untuk medisnya.
mengetahui gambaran derajat klinis terhadap Data yang digunakan adalah jenis
faktor resiko pada pasien reaksi kelamin, umur, riwayat alergi/asma dan
hipersensitivitas di rumah sakit sanglah allergen sesuai yang tercatat pada rekam
denpasar tahun 2012 hingga 2013 medis penderita. Derajat klinis dilihat
berdasarkan klinis dan diagnosis yang
METODE PENELITIAN tercatat pada rekam medis. Derajat klinis
Penelitian ini menggunakan rancangan studi ringan dengan manifestasi pada kulit dan
deskriptif cross sectional. Data diperoleh jaringan subkutis seperti Generalized
secara sekunder melalui rekam medis yang erythema, urticaria, angioedema, dan
tercatat di Bagian Allergy dan Imunology pruritus tanpa rash. Derajat sedang dengan
Penyakit Dalam RSUP Sanglah. tambahan manifestasi pada saluran nafas,
Populasi target adalah pasien yang sistem kardiovaskular atau sistem saluran
menderita reaksi hipersensitivitas. Populasi pencernaan. Derajat berat terdapat tambahan
3
manifestasi seperti hypoxia, hypotension, Derajat klinis berat paling banyak pada
Shock, dan kelainan neurology. Data yang kelompok umur 41-60 tahun yaitu sebanyak
diperoleh disajikan dalam bentuk tabel dan 16 pasien (11,1%), derajat klinis sedang
dianalisis secara deskriptif. paling banyak pada kelompok umur 21-40
yaitu 24 pasien (16,7%) dan derajat klinis
HASIL ringan paling banyak pada kelompok umur
Pada penelitian ini didapatkan 144 sampel 41-60 tahun yaitu sebanyak 21 pasien
yang sesuai dengan criteria inklusi dan (14,6%). Sedangkan derajat klinis berat
eksklusi. Sampel terdiri dari 73 laki-laki paling sedikit pada kelompok umur ≤21
(50,7%) dan 71 perempuan (49,3%). Usia yaitu sebanyak 5 pasien (3,5%) (Tabel 3).
rata-rata sampel adalah 39,5 tahun (Tabel 1). Pasien yang memiliki riwayat allergy/asma
Berdasarkan derajat klinis, paling banyak mengalami derajat klinis
didapatkan derajat klinis ringan 53 pasien berat yaitu sebanyak 34 pasien (23,6%).
(36,8% ), derajat klinis sedang 47 pasien Sedangkan yang tidak memiliki riwayat
(32,6%) dan derajat klinis berat 44 pasien allergy/asma paling banyak mengalami
(30,6%). Kemudian dilakukan tabulasi derajat klinis ringan yaitu sebanyak 35 orang
silang antara faktor resiko dengan derajat (24,3%) (Tabel 4). Obat-obatan paling
klinis reaksi hipersensitivitas. banyak menyebabkan reaksi
Terdapat beberapa faktor resiko yang hipersensitivitas yaitu sebanyak 114 pasien
memberatkan klinis dari reaksi (79,2%) dengan derajat klinis berat yaitu 33
hipersensitivitas antara lain jenis kelamin, pasien (22,9%). Pasien dengan allergen
usia tua, riwayat allergy/asma, dan allergen makanan paling banyak mengalami derajat
penyebab terjadinya reaksi hipersensitivitas. klinis sedang yaitu sebanyak 12 pasien
Antara jenis kelamin dengan derajat klinis (8,3%). Allergen lain seperti sengatan tawon
didapatkan bahwa laki-laki lebih banyak hanya terjadi pada 7 pasien (4,9%) dengan
mengalami derajat klinis ringan yaitu 32 derajat klinis berat sebanyak 5 pasien
pasien (22,2%) Sedangkan perempuan (3,5%). (Tabel 5)
sebanyak 30 orang (20,8%) (Tabel 2).
4
Tabel 1. Karakteristik Sampel berdasarkan jenis kelamin dan umur pada penderita reaksi
hipersensitivitas di RSUP Sanglah tahun 2012-2013
Karakteristik Responden Frekuensi (Penderita) Persentase (%)
Jenis Kelamin
- Laki-laki 73 50,7%
- Perempuan 71 49,3%
Umur
- <21 18 12,5%
- 21-40 54 37,5%
- 41-60 52 36,1%
- >60 20 13,9%
Tabel 2.Tabulasi Silang Jenis Kelamin Dengan Derajat Klinis Reaksi Hipersensitivitas pada
penderita reaksi hipersensitivitas di RSUP Sanglah tahun 2012-2013
Derajat Klinis
Variabel Total
Ringan Sedang Berat
Jenis Laki-laki 32 (22,2%) 17 (11,8%) 24 (16,7%) 73 (100%)
Kelamin Perempuan 21 (14,6%) 30 (20,8%) 20 (13,9%) 71 (100%)
Total 53(36,8%) 47(32,6%) 44(30,6%) 144(100%)
Tabel 3. Tabulasi Silang Umur Dengan Derajat Klinis Reaksi Hipersensitivitas pada penderita
reaksi hipersensitivitas di RSUP Sanglah tahun 2012-2013
Derajat Klinis
Variabel Total
Ringan Sedang Berat
Umur <21 tahun 9 (6,3%) 4 (2,8%) 5 (3,5%) 18 (12.5%)
21-40 tahun 15 (10,4%) 24 (16,7%) 15 (10,4%) 54 (37,5%)
41-60 Tahun 21 (14,6%) 15 (10,4%) 16 (11,1%) 52 (36,1%)
>60 Tahun 8 (5,6%) 4 (2,8%) 8 (5,6%) 20 (13,9%)
Total 53 (36,8%) 47 (32,6%) 44 (30,6%) 144 (100%
5
Tabel 4. Tabulasi Silang Riwayat Allergy/Asma Dengan Derajat Klinis Reaksi Hipersensitivitas
pada penderita reaksi hipersensitivitas di RSUP Sanglah tahun 2012-2013
Derajat Klinis
Variabel Total
Ringan Sedang Berat
Riwayat Ada 18 (12,5%) 23 (16,0%) 34 (23,6%) 75 (52,1%)
Allergy/Asma Tidak 35 (24,3%) 24 (16,7%) 10 (6,9%) 69 (47,9%)
Total 53(36,8%) 47(32,6%) 44(30,6%) 144(100%)
Tabel 5. Tabulasi Silang Allergen Dengan Derajat Klinis Reaksi Hipersensitivitas pada penderita
reaksi hipersensitivitas di RSUP Sanglah tahun 2012-2013
Derajat Klinis
Variabel Total
Ringan Sedang Berat
Obat 47 (32,6%) 34 (23,6%) 33 (22,9%) 114 (79,2%)
Allergen Makanan 5 (3,5%) 12 (8,3%) 6 (4,2%) 23 (16,0%)
Lain-lain 1 (0,7%) 1 (0,7%) 5 (3,5%) 7 (4,9%)
Total 53(36,8%) 47(32,6%) 44(30,6%) 144(100%)
PEMBAHASAN hipersensitivitas serupa dengan penelitian

Pada penelitian ini, selama tahun 2012- yang kami dapatkan.8 Pada penelitian ini,
2013, terdapat 144 pasien reaksi derajat klinis berat didapatkan paling banyak
hipersensitivitas di RSUP Sanglah. Data dari pada laki-laki yaitu 24 pasien (16,7%).
penelitian Gupta dkk pada tahun 2003 Sedangkan, Caleb dan venu pada tahun 2009
terdapat 13.250 pasien dengan reaksi mendapatkan lebih banyak berjenis kelamin
hipersensitivitas dari tahun 1992 sampai perempuan yang mengalami derajat klinis
2000.7 Dari jenis kelamin didapatkan jumlah berat.9 Kemungkinan seseorang mengalami
yang berimbang antara laki-laki dengan derajat klinis yang lebih berat, secara
perempuan yaitu 73 pasien (50,7%) dengan fisiologis kurang dipengaruhi oleh jenis
71 pasien (49,3%). Hasil penelitian Kemp kelamin.10,11
dkk pada tahun 2002 menunjukkan hasil Derajat klinis berat ditemukan paling
yang berimbang antara jumlah laki-laki dan banyak pada usia 41-60 tahun. Jumlah
perempuan yang mengalami reaksi pasien yang mengalami derajat klinis sedang
6
dan berat cenderung lebih besar pada usia sebagian besar rute masuknya obat melalui
tua dibandingkan usia muda. Berdasarkan jalur oral (82,4%).17 Pada penelitian ini,
penelitian Tang dkk tahun 2009, kejadian kami tidak melakukan analisis rute
reaksi hipersensitivitas dengan derajat klinis masuknya obat yang menyebabkan
sedang maupun berat lebih meningkat perbedaan derajat klinis reaksi
dengan pertambahan usia. Pertambahan usia hiersensitivitas.
menyebabkan perubahan fisiologis pada
tubuh yaitu dengan menurunnya daya PENUTUP
12
kompensasi dan daya adaptasi. Menurut Simpulan
Khan dan Lieberman tahun 2009, reaksi Pada penelitian ini didapatkan
hipersensitivitas pada usia tua dapat 1. Karakteristik pasien reaksi
memberikan gejala yang lebih berat. Hal hipersensitivitas di RSUP Sanglah
tersebut berkaitan dengan penurunan fungsi menurut jenis kelamin berjumlah hampir
organ pada usia lanjut.13,14 sama yaitu laki-laki 73 pasien (50,7%)
Jumlah pasien yang memiliki riwayat dan perempuan 71 pasien (49,3%).
allergy/asma tiga kali lebih banyak memiliki Sedangkan jumlah penderita dalam
derajat klinis berat dibandingkan dengan kelompok umur terbanyak di rentang
yang tidak memiliki riwayat allergy/asma. umur 21-40 tahun yaitu sebanyak 54
Sedangkan pada derajat klinis ringan lebih pasien (37,5%) dan paling sedikit pada
banyak pada pasien yang tidak memliki rentang umur kurang dari 21 tahun yaitu
riwayat allergy/asma. Hal ini dilaporkan sebanyak 18 pasien (12,5%)
serupa oleh Andrea dkk tahun 2008.15 2. Dilihat dari faktor resiko, pada faktor
Allergen merupakan penyebab resiko jenis kelamin paling banyak
terjadinya reaksi hipersensitivitas dapat mengalami derajat klinis berat adalah
berupa obat, makanan dan lain-lain seperti laki-laki yaitu 24 pasien(16,7%), pada
sengatan serangga. Pada penelitian ini, kategori umur paling banyak mengalami
reaksi hipersensitivitas derajat ringan hingga derajat klinis berat pada umur 41-60
berat disebabkan paling banyak oleh obat tahun yaitu 16 pasien (11,1%), pasien
diikuti makanan dan lain-lain. Hal serupa dengan faktor resiko riwayat
dilaporkan suryana dkk tahun 2008.16 Pada allergy/asma memiliki derajat klinis
penelitian oleh Aziz dkk tahun 2008, berat paling banyak yaitu 34 pasien
7
(23,6%), dan obat-obatan menjadi faktor 7. Gupta R, Sheik A, Strachan D, Anderson
resiko allergen yang paling banyak H. Increasing hospital admissions for
systemic allergic disorders in England:
terdapat pasien dengan derajat klinis
Analysis of nation admissions data. Br
berat yaitu 33 pasien (22,9%) Med J. 327:1142-3. 2003
Saran 8. Kemp SF, Lockey RF. Anaphylaksis : A
Perlu adanya penelitian lebih lanjut review of causes and mechanism. J
dengan rancangan penelitian analitik allergy Clinical Immunology. 110:341-8.
2002
untuk mengetahui pengaruh faktor risiko
9. Kelly Caleb, Gangur Venu. Sex
tersebut dengan derajat klinis reaksi
Disparity in Food Allergy : Evidence
hipersensitivitas. From The Pubmed Database. Journal of
Allergy. Hindawi Publishing
DAFTAR PUSTAKA Corporation. 2009
1. Brown SGA. Clinical features and 10. Abbas AK, Litchman AH, Pillai S.
severity grading of anaphylaksis. J Immediate Hypersensitivity. In: Abbas
Allergy Clinical Immunology.113:371-6. AK, editor. Cellular and molecular
2004 immunology. 6th ed. Philadelphia:
2. Lieberman P. Epidemiology of Saunders Elsevier: p.447-53. 2007
anaphylaxis. Curr Opin Allergy Clin 11. Karnen GB, Iris R. Reaksi
Immunol. 8(4):316-20. 2008 hipersensitivitas. In: Karnen GB, editor.
3. Bresser H, Sandner CH, Rakoski J. Imunologi dasar. 8th ed. Jakarta: Balai
Anaphylactic emergencies in Munich in Penerbit FKUI: p.379-81. 2009
1992 (abstract). J Allergy Clin Immunol. 12. Tang ML, Osbone N, Allen K.
95:368. 2000 Epidemiology of anaphylaksis. Curr
4. Mertes PM, Laxenaire MC, Alla F. Opin Clinical Immunology.9:351-6.
Anaphylactic and anaphylactoid 2009
reactions occurring during anesthesia in 13. Khan BQ, Lieberman P. Anaphylaksis in
France in 1999-2000. Anesthesiology. the elderly. Future Medicine. 4:377-87.
99(3):536-45. 2003 2009
5. Thong BYH, Tan TC. Epidemiology and 14. Freitas AA Magalhaes PD. A review and
risk factors for drug allergy. British Appraisal of the DNA damage Theory of
Journal of Clinical Pharmacology. Aging. Mutation Research. 728:12-22.
71:684-700. 2010 2011
6. Krause SC. Anaphylaksis. 2009. Akses: 15. Andrea J et al. Clinical and Genetic Risk
17 November 2014. Available at : Factors of Self-reported Penicillin
http//www.emedicine.com
8
Allergy. Journal of Allergy and Clinical
Immunology. 122:152-8. 2008
16. Suryana K, Suardamana K, Imbawan
IGNE. Karakteristik pasien dengan
reaksi hipersensitivitas/anaphylaksis
yang dirawat di RSUP Sanglah
Denpasar. 2008
17. Aziz S, Bernadette A. Hospital
admissions for acute anaphylaksis: time
trend study. BMJ. 320:1441-6. 2000
9
http://jurnal.fk.unand.ac.id 90
Artikel Penelitian
Karakteristik Klinis dan Patologis Karsinoma Nasofaring di

Bagian THT-KL RSUP Dr.M.Djamil Padang
1 2 3
Shofi Faiza , Sukri Rahman , Aswiyanti Asri
Abstrak
Karsinoma nasofaring banyak terjadi di Cina dan Asia Tenggara, termasuk di Indonesia, sering didiagnosis
pada keadaan lanjut dan memiliki prognosis yang buruk. Tujuan penelitian ini adalah untuk mengevaluasi
epidemiologi, karakteristik klinis, dan tipe histopatologi pada pasien karsinoma nasofaring di Bagian THT-KL.
Metodologi penelitian ini adalah deskriptif dengan menggunakan data rekam medik di RSUP Dr. M. Djamil Padang
selama Juni 2010 sampai Juli 2013 dan data hasil pemeriksaan histopatologi sebagai konfirmasi. Didapatkan
sebanyak 44 kasus yang lengkap pada periode tersebut, yang mana 52,27% penderita adalah laki-laki dan 47,22%
perempuan, perbandingan laki-laki dan perempuan adalah 1,2 : 1. Sebaran umur penderita dari 17 sampai 75 tahun
dengan insiden puncak pada umur 41- 65 tahun. Gejala klinis terdiri atas massa di leher 93,17%, diikuti dengan
obstruksi nasal 79,55%, dan gangguan pendengaran 79,55% sedangkan tanda klinis terdiri atas pembesaran
kelenjar getah bening leher 90,91%, diikuti dengan tuli 79,55%, cranial nerve palsy dan perluasan kelenjar getah
bening ke fossa supraklavikula masing-masing 15,8%. Sebagian besar pasien berada pada stadium IV 83,16%,
dengan derajat tumor terbanyak T4N2M0 15,91%. Tipe histopatologi yang terbanyak adalah nonkeratinizing
carcinoma, undifferentiated type 75%, diikuti keratinizing SCC 13,64%, dan nonkeratinizing carcinoma -
differentiated type 11,36%.
Kata kunci: karsinoma nasofaring, karakteristik klinis, histopatologi, padang
Abstract
Nasopharyngeal carcinoma is more frequent in China and Southeast Asia, including Indonesia, commonly with
advance stages at diagnosis and has a poor prognosis. The objective of this study was to evaluate epidemiology,
clinical characteristic and histopathology types of patients with nasopharyngeal carcinoma in the department of
Otorhinolaryngology - Head and Neck Surgery.This is a descriptive study that used data from medical record of Dr. M.
Djamil General Hospital in Padang during June 2010 to July 2013 and histopathology examination as confirmation.
The result demonstrated 44 cases found on that period, of which 52,27% was male and 47,72% was female, hence the
male and female ratio was 1,2 : 1. The age-range from 17 to 75 years old with incidence peak between 41 - 65 years
old. Clinical symptoms were neck mass 93,17%, followed by nasal obstruction 79,55 %, and audiological complaints
79,55% while clinical sign were cervical lymphadenopathy 90,91%, followed by hearing loss 79,55%, cranial nerve
palsy and lymphadenopathy metastases to fossa supraclavicular each subject 15,8%. Most of patients were classified
as stage IV 83,16%, with T4N2M0 15,91%. The histopathology type were nonkeratinizing carcinoma, undifferentiated
type had percentage 75%, followed by keratinizing SCC 13,64%, and nonkeratinizing carcinoma - differentiated type
11,36%.
Keywords: nasopharyngeal carcinoma, clinical characteristic, histopathology types, padang
Affiliasi penulis: 1. Pendidikan Dokter FK UNAND Fakultas Korespondensi: Sukri Rahman, Bagian THT-KL FK UNAND/
Kedokteran Universitas Andalas Padang, 2. Bagian THT-KL FK RSUP.Dr.M. Djamil Padang, sukri_rahman@yahoo.com
UNAND/ RSUP Dr. M.Djamil Padang, 3. Bagian Patologi Anatomi FK
UNAND
Jurnal Kesehatan Andalas. 2016; 5(1)

PENDAHULUAN tahun dan menurun setelah umur 60 tahun. Angka

Keganasan yang dalam istilah medis disebut kejadian KNF pada anak bervariasi antara 1-5 % dari
kanker merupakan salah satu kasus kematian utama seluruh kejadian kanker pada anak. Pria lebih banyak
di dunia, termasuk di negara berkembang. Kanker daripada wanita, yaitu 3 : 1.6,7 Sementara penyebab
juga merupakan hal yang paling dicemaskan oleh KNF bersifat multifaktorial, dikaitkan dengan adanya
masyarakat saat ini. Dalam lingkupan medis, kanker interaksi antara infeksi kronik oncogenic gamma
menjadi kajian menarik karena masih banyak yang herpesvirus Epstein-Barr virus yang mana virus
belum terungkap tentang penyakit ini. Berbagai teori Epstein-Barr telah menginfeksi lebih dari 95% populasi
telah dimunculkan untuk mekanisme terjadinya dunia. Selain itu faktor lingkungan dan faktor genetik,
3
kanker, begitu juga dengan pengobatan serta juga terlibat dalan proses multistep karsinogenik.
prognosis yang belum memuaskan. Berdasarkan data Data epidemiologi menyebutkan bahwa ras
WHO, di dunia terdapat 13% kematian disebabkan Mongoloid memiliki angka kejadian yang tinggi untuk
oleh kanker dan ada 100 jenis kanker yang bisa menderita karsinoma nasofaring. Masyarakat
menyerang tubuh manusia. Sekitar 70% kematian oleh Indonesia yang sebagian besar termasuk dalam ras
kanker berasal dari populasi negara dengan Mongoloid serta memiliki kebiasaan mengonsumsi
pendapatan rendah dan menengah. Di Indonesia, ikan asin yang merupakan salah satu dari 9 bahan
prevalensi kanker adalah 4,3 per 1.000 penduduk dan makan pokok masyarakat Indonesia. Ikan asin
merupakan penyebab kematian nomor 7 (5.7%) memiliki kandungan nitrosamin yang merupakan salah
setelah stroke, tubekulosis, hipertensi, cedera, satu faktor pencetus kanker ini. Nitrosamin juga diteliti
perinatal, dan diabetes mellitus. 1,2 terkandung dalam beberapa jenis makanan yang
Salah satu masalah kanker yang sulit dideteksi diawetkan, seperti daging olahan. Nitrosamin
dini adalah Karsinoma Nasofaring (KNF). KNF merupakan mediator utama yang dapat mengaktifkan
merupakan kanker yang mempunyai keunikan dan virus Epstein-Barr yang memicu mekanisme kanker.
berbeda dari tumor ganas di daerah kepala dan leher Berdasarkan studi genetik juga terdapat adanya
lainnya dalam hal epidemiologi, spektrum gambaran hubungan antara alel Human Leukocyt Antigen (HLA)
histopatologi, karakteristik klinik dan sifat biologi. Hal kelas I dan II dalam populasi dan risiko KNF.8,6
ini terlihat dari kejadian KNF yang bersifat endemik di Gejala dan tanda klinis yang sering ditemu-kan
Asia seperti Cina Selatan, Asia Tenggara, Jepang, pada KNF diantaranya epistaksis, obstruksi hidung,
dan Timur Tengah. Insiden KNF tertinggi di dunia tinnitus serta tuli, sefalgia, gejala saraf kranial,
dijumpai pada penduduk daratan Cina bagian selatan, diplopia, pembesaran KGB leher dan gejala
khususnya suku Kanton di provinsi Guang Dong dan metastasis jauh dengan lokasi tersering adalah ke
daerah Guangxi dengan angka mencapai lebih dari 50 tulang, paru-paru, hati dan sering juga terjadi
per 100.000 penduduk pertahun. Sementara insiden metastasis pada banyak organ sekaligus.9
KNF di dunia tergolong jarang, yaitu 2% dari seluruh Klasifikasi WHO tahun 1978 membagi KNF
karsinoma sel squamous kepala dan leher, dengan menjadi squamous cell carcinoma (WHO tipe 1),
insiden 0.5 sampai 2 per 100.000 di Amerika nonkeratinizing carcinoma (WHO tipe 2) dan
3,4 undifferentiated carcinoma (WHO tipe 3). Klasifikasi
Serikat. Berdasarkan data registrasi kanker berbasis
rumah sakit di RS Kanker Dharmais tahun 2003-2007 yang saat ini digunakan adalah WHO tahun 1991
didapatkan bahwa KNF berada diperingkat ketiga yang membagi tumor ganas ini menjadi squamous cell
setelah karsinoma mamae dan serviks, sementara carcinoma (keratinizing SCC), nonkeratinizing
untuk kanker terbanyak pada pria, KNF berada carcinoma yang terdiri atas differentiated dan
diperingkat pertama dan pada wanita berada undifferentiated, dan basaloid SCC. Batasan antara
5 subtipe ini terkadang tidak jelas, bahkan beberapa
diperingkat kelima.
KNF dapat mengenai berbagai umur, tersering peneliti melaporkan bahwa SCC dan nonkeratinizing
umur 40-60 tahun. Mulai meningkat setelah umur 20 carcinoma sebenarnya adalah varian dari satu

kelompok tumor yang homogen. Berdasarkan laporan dua tahun pasien KNF antara stadium II, III, dan IV
15
dari berbagai negara, KNF subtipe undifferentiated yang dilakukan terapi kemoradiasi. Kegagalan
carcinoma (termasuk nonkeratinizing carcinoma) pengobatan ini disebabkan oleh perluasan
adalah subtipe yang terbanyak ditemukan yaitu, Hong lokoregional (40-80% pasien) dan kekambuhan (15-
16
Kong (99%), Singapore (83%), Tunisia (92%), Jepang 50% pasien).
(87%) dan Amerika Serikat (75%).10,11 Terlihat bahwa karsinoma nasofaring
Penelitian yang dilakukan Piasiska (2010) di merupakan kanker kepala dan leher yang paling
Medan tahun 2009 didapatkan subtipe terbanyak sering ditemukan pada masyarakat kita dengan faktor
adalah WHO - 3 (Undifferentiated subtype ) sebesar risiko yang tinggi untuk menderita penyakit ini baik dari
12
51,63%. Data dari bagian Patologi Anatomi Fakultas segi ras, geografis dan kebiasaan.Penderita
Kedokteran Universitas Andalas tahun 2006-2008 karsinoma nasofaring sering lambat terdeteksi karena
didapatkan sebanyak 45 kasus KNF di Sumatera faktor lokasi tumor, gejala yang tidak khas dan biopsi
Barat, dengan subtipe terbanyak adalah subtipe WHO- yang sulit sehingga menyebabkan buruknya prognosis
2 (Nonkeratinizing Carcinoma) dan WHO-3 dari penyakit ini.
13
(Undifferentiated subtype).
Penanggulangan KNF sampai saat ini masih METODE
merupakan suatu masalah yang cukup sulit. Hal ini Penelitian yang dilakukan bersifat deskriptif
karena etiologinya yang masih belum pasti. Selain itu dengan menggunakan data rekam medik di RSUP Dr.
letak nasofaring yang cukup tersembunyi sehingga M. Djamil Padang dari Oktober sampai Desember
sulit untuk mendeteksinya. Gejala dini dari penyakit ini 2013. Sampel penelitian adalah data semua pasien
sering tidak jelas dan tidak khas sehingga sering dengan diagnosis akhir karsinoma nasofaring yang
diabaikan. Sebagian besar gejala klinis baru dirawat di Bangsal THT-KL RSUP Dr. M. Djamil
bermanifestasi setelah tumor bermetastasis ke Padang dari Juni 2010 sampai Juli 2013 berupa
kelenjar getah bening (KGB) leher. Mutlak dilakukan identitas pasien, status rawat inap / jalan pasien, hasil
biopsi histopatologis sebagai konfirmasi diagnosis pemeriksaan klinis, histopatologi dan diagnostik. Data
kanker ini. Pemeriksaan biopsi nasofaring sering yang diperoleh diolah dan dikelompokkan secara
ditemukan hasil yang negatif karena letak tumor yang manual dan komputerisasi dalam bentuk tabel
tersembunyi mempersulit pengambilan dan distribusi frekuensi dan diagram batang.
penanganan oleh dokter.14
Pada stadium awal, pengobatan standar untuk HASIL
KNF adalah radioterapi karena sifatnya yang
Ada 44 kasus KNF yang diteliti, yang terdiri dari
radiosensitif. Pembedahan tidak berperan besar untuk
52,27 % penderita adalah laki-laki dan 47,72% adalah
mengobati karsinoma nasofaring tetapi dapat
perempuan (Tabel 1). Penderita terbanyak ditemukan
membantu dalam mengangkat sisa nodul di leher
pada dewasa tua dengan kisaran umur 41-65 tahun
setelah diradiasi. Sementara kemoterapi memberikan
sebesar 68,18%, diikuti oleh dewasa muda dengan
harapan dalam meningkatkan kontrol tumor dan
kisaran umur 21- < 41 tahun sebesar 24,99%,
survival pada KNF stadium lanjut.14
kemudian remaja dengan kisaran umur 13- < 21 tahun
Penelitian yang dilakukan Kurniawan (2011)
sebesar 4,54%, dan kisaran umur yang paling sedikit
mengenai angka harapan hidup dua tahun dari 56
ditemukan pada manula dengan umur > 65 tahun
kasus didapatkan pasien KNF dengan kemoradiasi
sebesar 2,27%.
secara keseluruhan sebesar 60%. Pada bulan ke 24,
Bayi, balita dan anak-anak tidak ditemukan
pasien stadium II memiliki angka harapan hidup diatas
dalam sampel. Sesuai dengan distribusi umur menurut
80%, pasien stadium IV sebesar 60%, dan pasien
jenis kelamin laki-laki terbanyak ditemukan pada
stadium III hanya sebesar 40%. Tidak terdapat
dewasa tua dengan kisaran umur 41-65 tahun sebesar
perbedaan yang bermakna pada angka harapan hidup

40,91%. Perempuan ditemukan terbanyak pada Tanda klinis terbanyak yang ditemukan pada
dewasa tua dengan kisaran umur 41-65 tahun sebesar penderita KNF adalah pembesaran KGB leher sebesar
27,27% (Tabel 1) 93,17% tuli sebesar 79,54%. Diikuti dengan tuli
sebesar 79,54%, cranial nerve palsy dan perluasan
Tabel 1. Distribusi frekuensi penderita karsinoma KGB leher ke fossa supraklavikula masing-masing
nasofaring menurut umur dan jenis kelamin sebesar 29,55%. Dermatomiositis tidak ditemukan
Umur Laki -laki Perempuan Total pada penelitian ini.(Tabel3)
(tahun) f % f % f %
0 - <1 - - - - - -
Tabel 3. Distribusi penderita karsinoma nasofaring
1 - <5 - - - - - -
5 - <13 - - - - - - menurut tanda klinis
13 - <21 - - 2 4,54 % 2 4,54 % Tanda klinis f %
21 - <41 5 11,36 % 6 13,63 % 11 24,99 % Pembesaran KGB leher 40 90.91 %
41 - 65 18 40,91 % 12 27,27 % 30 68,18 %
Perluasan KGB leher ke fossa 13 29.55 %
>65 0 0.00 % 1 2,27 % 1 2,27 %
supraklavikula
Total 23 52,27 % 21 47,72 % 44 100 %
Cranial nerve palsy 13 29.55 %
Tuli 35 79.55 %
Gejala klinis terbanyak ditemukan adalah
Dermatomiositis 0 0.00 %
massa di leher sebesar 90,91%. Obstruksi hidung dan
gangguan pendengaran ditemukan sebesar 79,55%, Pada Gambar 1 terlihat penderita umumnya
epistaksis sebesar 68,18%, sakit kepala sebesar datang pada stadium IV sebesar 75%. Diikuti
63,64%, dan tinnitus sebesar 56,82%. Gejala klinis penderita yang datang pada stadium II ditemukan
berupa diplopia sebesar 38,64%, otorea sebesar sebanyak 13,64%. Penderita yang datang pada
31,82%, dan ptosis sebesar 22,73%. Gejala klinis stadium III sebesar 11,36% dan tidak ada pasien yang
berupa rinore, otalgia, deviasi lidah, trismus, peng- datang pada stadium I. Pada penelitian ini ditemukan
lihatan kabur, dan eksoftalmus masing-masing juga sebesar 11,36% pasien yang berada pada
ditemukan lebih kecil dari 20%.(Tabel 2) stadium IV C yang mengalami metastasis jauh ke
organ seperti paru-paru, tulang dan ginjal (Gambar 1).
Tabel 2. Distribusi penderita karsinoma nasofaring
menurut gejala klinis
40,00%
Gejala klinis f % 31,82% 31,82%
Massa leher 40 90.91 % 30,00%
Hidung :
Epistaksis 30 68.18 % 20,00% 13,64%
11,36% 11,36%
Obstruksi Hidung 35 79.55 %
Rinore 7 15.91 % 10,00%
0,00%
Telinga : 0,00%
Gangguan 79.55 % I II III IV A IV B IV C
pendengaran 35
Tinnitus 25 56.82 % Stadium tumor (%)
Otalgia 1 2.27 %
Otorea 14 31.82 %
Saraf kranial : Gambar 1. Distribusi penderita karsinoma nasofaring
Paresis Wajah 14 31.82 % menurut stadium tumor
Deviasi lidah 2 4.55 %
Trismus 4 9.09 %
Mata : Hasil penelitian ditemukan tipe histopatologi
Diplopia 17 38.64 %
terbanyak adalah Nonkeratinizing carcinoma-
Pengliharan kabur 5 11.36 %
Ptosis 10 22.73% undifferentiated type, sebesar 75%.Keratinizing
Eksoftalmus 2 4.55 %
Squamous Cell Carcinoma ditemukan sebanyak
Nyeri Kepala 28 63.64 %
13,64% dan Nonkeratinizing carcinoma-differentiated
type sebesar 11,36%. Sedangkan Basaloid Squamous

Cell Carcinoma tidak ditemukan dari seluruh pende- perluasan tumor ke kelenjar getah bening (KGB). KNF
rita.(Gambar 2). biasanya menyebar melalui limfe ke KGB leher. Aliran
limfatik dari Fossa Rossenmuller mengalir ke Nodus
75,00% Rouvier sampai ke ruang retrofaringeal dan berlanjut

80,00%
70,00% ke KGB leher atas dalam. Hal ini menjelaskan bahwa
60,00% massa di leher sering menjadi gejala klinis dari KNF
50,00%
yang membuat pasien berobat ke dokter. Kanker ini
40,00%
akan menyebar melalui aliran darah ke daerah yang
30,00%
21
20,00% 13,64% jauh seperti tulang, paru dan hati. Obstruksi hidung
11,36%
10,00%
0,00%
dikarenakan massa tumor yang telah menginvasi
0,00% rongga hidung atau sinus paranasal. Massa tumor
Keratinizing SCC Nonkeratinizing Nonkeratinizing Basaloid SCC
Carcinoma - Carcinoma -
differentiated undifferentiated
yang awalnya tumbuh di daerah Fossa Rossenmuller
type type
biasanya akan menginfiltrasi daerah tuba eustachius
Tipe Histopatologi, n=44 di dekatnya dan menyebabkan penyumbatan tuba dan
bermanifestasi klinis terhadap menurunnya
Gambar 2. Distribusi penderita karsinoma nasofaring
pendengaran. Kesulitan dalam menelan biasanya
menurut tipe histopatologi
dikarenakan massa tumor yang telah menginvasi
daerah orofaring dan menekan saraf daerah
PEMBAHASAN
tenggorok. Sakit kepala biasanya diakibatkan oleh
Perbandingan penderita laki-laki dengan
gejala neurologis yang diikuti gejala mata pada KNF
perempuan adalah 1,2 : 1. Penelitian Yenita dan Asri
akibat inflasi tumor ke daerah otak, mata dan juga
(2010) di Sumatera Barat mendapatkan perbandingan
gejala psikologis yang diakibatkan tumor. 21
13
2 : 1. Di Amerika Serikat didapatkan angka
Pada pemeriksaan klinis, pasien yang
perbandingan 2 : 1.17 KNF selalu lebih tinggi
ditemukan dengan keadaan pembesaran KGB leher
ditemukan pada laki-laki. Belum ditemukan secara
sebesar 90,91%. Tanda klinis lainnya adalah tuli
pasti penyebab tingginya penderita laki-laki daripada
sebesar 79,55%, diikuti dengan pembesaran KGB
perempuan pada kanker ini. Laki-laki mungkin
leher ke fossa supraklavikula dan cranial nerve palsy
cenderung lebih sering terpapar zat-zat karsinogen di
yang masing-masing sebesar 29,55 %. Persentase ini
lingkungan kerjanya dibanding perempuan sehingga
berbeda dengan data di RS Pamela Youde Nethersole
lebih berisiko untuk menderita kanker.
Western Hospital, Hong Kong dari tahun 1994 – 2001,
Pada penelitian ini secara keseluruhan
yaitu pembesaran KGB leher unilateral merupakan
didapatkan kisaran umur antara 17 – 75 tahun.
terbanyak sebesar 72%, diikuti dengan pembesaran
Penderita berumur dibawah 20 tahun tergolong rendah
KGB leher bilateral sebesar 35%, kemudian perluasan
yaitu 4,54%. Di Departemen IKA RSCM, KNF berada
KGB leher ke fossa supraklavikula sebesar 12%,
pada urutan ke-10 dan berkontribusi 2% (24 dari 1194
cranial nerve palsy 10%, tuli sebesar 3%, dan derma-
18
pasien) seluruh kanker pada anak tahun 2004-2009.
tomiositis sebesar 1%. Untuk daerah yang sangat
KNF pada anak diduga berkaitan erat dengan peran
endemik dapat ditemukan tanda klinis dermatomiositis
infeksi EBV. Peningkatan titer antibodi immunoglobulin
pada pasien KNF, hal ini berkaitan dengan tingginya
G dan A (IgG dan IgA) terhadap antigen kapsid viral 11
risiko pasien dermatomiositis untuk menderita KNF.
19
ditemukaan pada pasien KNF. Adanya faktor genetik,
Dalam penelitian ini tidak ditemukan dermatomiositis.
lingkungan dan kebiasaan hidup, dan infeksi berperan
Studi terakhir membuktikan rendahnya
sebagai kontributor dominan terhadap kejadian KNF
kesadaran terhadap gejala dan tanda awal KNF pada
20
ini masih memerlukan penelitian lanjut.
tenaga medis di Indonesia. Tersembunyinya lokasi
Massa di leher merupakan gejala klinis
tumor dan diperlukannya pemeriksaan khusus di
terbanyak yang ditemukan pada penderita sebesar
Rumah Sakit yang memiliki fasilitas lengkap juga
90,91%. Massa di leher menunjukkan telah terjadinya

21
mempersulit deteksi awal dari kanker ini. Pada konsumsi rokok dan alkohol, lebih dari 75%
dasarnya, pemeriksaan dan biopsi tumor nasofaring merupakan karsinoma sel skuamosa. Undifferentiated
perlu dilakukan dengan pemeriksaan nasoendoskopi. type merupakan yang terbanyak dalam prevalensi
Pengetahuan tentang tanda dan gejala yang paling KNF di Asia Tenggara dan beberapa daerah dengan
sering timbul membantu klinisi untuk lebih cepat insiden KNF yang tinggi, dan sangat berhubungan
mendiagnosis tumor yang ada, sehingga dapat dengan infeksi EBV. Sedangkan keratinizing SCC
22
meningkatkan pilihan pengobatan. Adanya dapat juga dihubungkan dengan EBV di daerah
keterlambatan diagnosis terkait dengan ketidak- endemik, dan banyak terdapat di daerah non-endemik
sadaran pasien terhadap KNF dan kecenderungan seperti Amerika, yang sering dihubungkan dengan
23
pasien mencari obat tradisional sebagai pengobatan perokok dan pecandu alcohol.
21
pertama.
Pada penelitian ini didapatkan penderita datang KESIMPULAN
dengan stadium lanjut sebesar 86,36%, yang terdiri Perbandingan penderita laki-laki dan
atas stadium IV dan III, diikuti stadium awal yaitu perempuan adalah 1,2 : 1, dengan sebaran umur
stadium II sebesar 13,64%, sedangkan stadium I tidak mulai dari 17 sampai 75 tahun. Penderita terbanyak
ditemukan. Hal ini memperlihatkan bahwa penderita ditemukan pada kelompok umur 41 sampai 65 tahun.
biasanya telah datang dalam keadaan stadium lanjut Pembesaran kelenjar getah bening leher merupakan
dengan gambaran penderita terbanyak berupa massa tanda dan gejala klinis terbanyak yang ditemukan
tumor yang telah menginvasi intrakranial dengan pada seluruh penderita.
melibatkan nervus kranial, fossa infratemporal, Mayoritas penderita datang pada stadium lanjut
hipofaring, orbit, atau ruang mastikator yang dapat (stadium IV dan III), diikuti stadium II ,stadium I tidak
dilihat pada hasil pemeriksaan CT Scan. Adanya ditemukan, terdapat pasien yang mengalami
pembesaran kelenjar getah bening leher bilateral di metastasis jauh ke paru-paru, tulang, dan ginjal.
atas fossa supraklavikula dengan ukuran kurang lebih Tipe histopatologis terbanyak adalah bentuk
6 cm dan tidak terdapat metastasis berdasarkan foto nonkeratinizing carcinoma – undifferentiated type,
thorax. diikuti keratinizing, nonkeratinizing carcinoma –
Hasil pemeriksaan tipe histopatologi yang juga differentiated type, sedangkan basaloid SCC tidak
sebagai konfirmasi diagnosis, jenis KNF yang paling ditemukan.
banyak ditemukan adalah nonkeratinizing carcinoma-
undifferentiated type sebesar 72,73%. Untuk DAFTAR PUSTAKA
keratinizing SCC sebesar 13,64% dan nonkeratinizing 1. WHO. 10 facts about cancer. 2013 (diunduh 5 Mei
carcinoma-differentiated type sebesar 11,36%. 2013). Tersedia dari: URL: HYPERLINK
Penelitian Yenita dan Asri (2010) terhadap kasus KNF http://www.who.int/features/factfiles/cancer/en/inde
di Sumatera Barat berdasarkan WHO 1978 mendapat- x.html
kan WHO-3 sebagai tipe terbanyak sebesar 68%. 2. Riskesdas. Laporan nasional 2007 badan
WHO-3 merupakan undifferentiated carcinoma, yang penelitian dan pengembangan kesehatan
mana karsinoma anaplastik masuk kedalam tipe ini. Departemen Kesehatan RI; 2007
Sementara dalam klasifikasi WHO 1991, tipe WHO-3 3. Zhou XJ, Cui AA, Kajdacsy BH, Ye J, Wang PN,
masuk ke dalam jenis nonkeratinizing carcinoma- Rao. The progress on genetic analysis of
undifferentiated type.Walaupun terdapat perbedaan nasopharyngeal carcinoma. Comparative and
angka, terlihat bahwa nonkeratinizing carcinoma- Functional Genomics; 2007.
undifferentiated type merupakan tipe histopatologi 4. Chang ET, Adami HO. The enigmatic epidemiology
13
yang terbanyak ditemukan pada KNF. of nasopharyngeal carcinoma. Cancer
Tumor rongga mulut, tumor rongga hidung dan Epidemiology Biomarkers and Prevention.
tumor laring yang diyakini berhubungan dengan 2006;15(10):1765–77.

5. RKSD. Laporan kerja subbagian registrasi kanker management. London: Springer-Verlag; 2010.
bagian penelitian dan pengembangan RSKD. 15. Kurniawan R. Angka harapan hidup dua tahun
Pusat Kanker Nasional Departemen Kesehatan penderita karsinoma nasofaring pada berbagai
Republik Indonesia. Jakarta; 2010 (diunduh 5 Mei stadium yang dilakukan terapi kemoradiasi
2013). Tersedia dari: URL: HYPERLINK (skripsi). Semarang: Fakultas Kedokteran
http://indonesianjournalofcancer.org/ Universitas Diponegoro; 2011.
6. Chan ATC, Teo PML, Johnson PJ. 16. Farias TP, et al. Prognostic factors and outcome
Nasopharyngeal carcinoma. Dalam: B.Brockstein for nasopharyngeal carcinoma. Arch ortholaryngol
Head and Neck Cancer. New York: Kluwer head neck surg. 2003;129:794-9.
Academic. hlm.275-91. 17. Globocan. CA. Cancer J Clin. 2011;61:69-90.
7. Haleema Saeed AZ, Mehreen A, Rabia H, Amber Tersedia dari: URL: HYPERLINK http://globocan.
D. Pediatric Nasopharyngeal carcinoma: a review iarc.fr/Pages/fact_sheets_cancer.aspx
of 27 cases over 10 years at Shaukat Khanum 18. Amelia C, Novie, et al. Karsinoma nasofaring pada
Memorial Cancer Hospital and Research Center, anak: karakteristik, tata laksana, dan prognosis.
Pakistan. As Pac J Cancer Prevent. 2009;10:917- Dalam: Sari Pediatri. 2011;13(1).
20. 19. Brays F, Haugen M, Moger TA, Tretli S, Aalen OO,
8. Roezin, Averdi, Syafril A. Karsinoma nasofaring. Grotmol T. Age-incidence curves of naso-
Dalam: Soepardi EA, Iskandar N, editor pharyngeal carcinoma worldwide: bimodality in
(penyunting). Buku Ajar Ilmu Kesehatan Telinga low-risk populations and aetiologic implications.
Hidung Tenggorokan Kepala & Leher. Edisi Ke-6. Dalam: Cancer Epidemiol. Biomark; 2008. hlm.
Jakarta: Balai Penerbit FKUI; 2006. 2356-64.
9. Desen W. Tumor di kepala dan leher. Dalam: Buku 20. Sultan .I, Casanova M, Ferrari A, Rihani R,
Ajar Onkologi Klinis. Edisi ke-2. Jakarta: Balai Rodriguez-Galindo C. Differential features of
Penerbit FKUI; 2008. hlm.263-78. nasopharyngeal carcinoma in children and adults :
10. Kumar V, Cotran RS, Robbins SL. Robbins basic a SEER study. Pediatr. Blood Cancer. 2010; 55:
pathology. Edisi ke-7. Philadelphia: Elsevier 279-81.
Saunders; 2003. 21. Alan Soo-Beng Khoo, Kin-Choo Pua, Diagnosis
11. WHO. Pathology and genetics of head and neck and Clinical Evaluation of Nasopharyngeal
tumours. IARC Press.Lyon. 2005. Carcinoma. In Nasopharyngeal Carsinoma Keys
12. Piasiska H. Profil penderita karsinoma nasofaring for Translational Medicine and Biology. Springer,
di laboratorium patologi anatomi kota Medan tahun 2013.
2009 (tesis). Medan: Fakultas Kedokteran 22. Marlinda Adham, et al, Nasopharyngeal carcinoma
Universitas Sumatera Utara; 2010. in Indonesia: epidemiology, incidence, signs, and
13. Yenita, Asri A. Studi retrospektif karsinoma symptoms at presentation. Chinese Journal of
nasofaring di Sumatera Barat : reevaluasi subtipe Cancer. 2012;31(4).
histopatologi berdasarkan klasifikasi WHO. 23. Nicholls J, Niedobitek G. Histopathological
Padang: Fakultas Kedokteran Universitas Andalas; diagnosis of nasopharyngeal carcinoma: looking
2010. beyong the blue book. Dalam: Nasopharyngeal
14. Lu Jiade J, Jay S, Cooper, Anne, WM, Lee. Carsinoma Keys for Translational Medicine and
Nasopharyngeal cancer: multidiciplinary Biology. Springer; 2013.

Proceeding of International Conference on Electrical Engineering, Computer Science and Informatics (EECSI 2014), Yogyakarta, Indonesia, 20-21 August 2014
Design of Personalized Asthma Management System

With Data Mining Methods
Cut Fiarni1
Department of Information System,
InstitutTeknologiHarapanBangsa
DipatiUkur 80-84, Bandung, Indonesia
cutfiarni@ithb.ac.id
Abstract— Asthma is a chronic lung disease that inflames and lead a better asthma management for asthma patient. But, the
narrows the airways. Asthma is a multifactorial chronic illness, problem is because of the vast and complex trigger factors
there is no uniformity of triggers factors of asthma attack on for asthma attacks, to there will be huge amounts of data that
asthma patients with different degree of asthma severity. The need to be processed and analyzed. Data mining provides the
better understanding of the trigger factors could lead a better methodology and technology to extract these data into
asthma management for asthma patient. in this paper, we knowledge. One major advantage of data mining over a
proposed a design of personalized asthma management system. in traditional statistical approach is its ability to deal directly with
this system we used pattern sequential mining, clustering and heterogeneous data fields, which a usually contained in
classification to predicting the patient's asthma attack based on medical data sets [5]. In medical research, data mining begins
extracted rules from data mining. the main methodology is to
with the hypothesis and results are adjusted accordingly. With
identify the most important trigger factors based on the
clustering and classification rules to be extracted. The
the use of data mining methods, useful patterns of information
contribution to this research area is to analyze the suitable data can be found in this data, that will later be used for further
mining technique and algorithm for the proposed system. research and report evaluation [6].
The purpose of this study is to explore the applicability of
Keywords—Asthma; data mining; personalized management data mining technique in the efforts of personalized asthma
system. management and asthma severe attack prevention with
particular emphasis to build a model that could help to extract
I. INTRODUCTION pattern of asthma induced factors. With this objective, data
Asthma is a chronic lung disease that inflames and narrows mining technique was employed to classify patient daily
the airways. It characterized by airway swollen and hyper- activity records on the basis of the values of attributes asthma
responsiveness, which causes airflow limitation. Asthma is a attack, the the trigger factors and asthma attack degree of
major cause of chronic morbidity and mortality throughout the severity.
world and there is evidence that its prevalence has increased In this proposed Personalized Asthma Management
considerably over the past 20 years, especially in children [1]. System, daily activities, external and internal trigger factors of
In Indonesia, asthma prevalence is 4.5% per mile, with the asthma attack are collected, then two data mining methods
highest prevalence is in Sulawesi Tengah (7,8%), then in Nusa clustering and classification are adopted for predicting trigger
Tenggara Timur (7,3%), DI Yogyakarta (6,9%), and Sulawesi factors on asthma attack in different degree of severity. The
Selatan (6,7%) [2]. first methodology objective is to extract the significant
Chronic asthmatic sufferers need to be constantly observed information of trigger factors of asthma attacks by clustering
to prevent sudden attacks. Of all asthma patients, 50% have user's daily activity over some period of time. Then system will
symptoms on a daily basis and almost all patients report build a classifier by rules that extracted from the clustering
limitations to daily activities. A severe asthma attack may method. With this proposed system, the patients hopefully
require emergency care, and it can cause death. Despite the could gain better understanding of their trigger factors of
high quality of available medications and treatment regimens asthma attack, that could lead to better asthma management
that are being simplified on a regular basis, asthma is still not and a better quality of life. The rest of this paper is organized
sufficiently controlled in many cases [3]. Sometime asthma as follows. In section 2, we review several related studies. In
symptoms are mild and go away on their own or after minimal section 3 we explained the propose system architecture in
treatment with an asthma medicine. Other times, symptoms detail.. Finally, we summarize our research and list some future
continue to get worse [4]. It is important to treat work in section 4.
symptoms of early staged, in order to prevent the symptoms
from worsening and causing a severe asthma attack. II. RELATED WORK
Asthma is a multifactorial chronic illness, there is no Definition Data mining or is the non-trivial extraction of
uniformity of triggers factors of asthma attack. each patient has implicit, previously unknown and potentially useful
different trigger factors that lead to different degree of asthma information from the data. This encompasses a number of
severity. The better understanding of the trigger factors could technical approaches, such as clustering, data summarization,
120
classification, finding dependency networks, analyzing 1. User profile: gender, age, BMI
changes, and detecting anomalies [6]. Today, data mining is
one of the best technique to gain knowledge from huge 2. InternalFactors : Exercise, alergic food.
Infections(respiratory infections --like a cold, the flu,
amounts of data that has been used by many organizations.
Especially in healthcare, which their transaction data are really or sinusitis), Stress and emotion, Food Allergies, Diet
with low in vitamin and high fat
complex and sometime need to be recorded on very long period
of time, consequently it is resulting a voluminously dataset that 3. External : Dust Mites and Dust Control, animals
is vital to gain better decision on diagnosis or treatment. Data exposure , Environmental Tobacco Smoke, poor
mining brings a set of tools and techniques that can be applied Moisture and Ventilation, Air Pollution, Household
to discover hidden patterns that provide healthcare Products and Air Fresheners with strong odor and
professionals an additional source of knowledge for making weather condition.
important decisions.There are vast potential for data mining
applications in healthcare. Generally, these can be grouped as B. System Architecture
the evaluation of treatment effectiveness; management of
In this research, in order to increase asthma management,
health care; customer relationship management; and detection
we proposed a personalized asthma management system to
of fraud and abuse. In this paper we focus data mining for
records all significant information about patient regarding to its
Healthcare management. The main focus of data mining for
asthma condition, over some period of time. Then we will
healthcare management are to aid healthcare management on
adopt data mining technique discover important information of
the better way to identify and track chronic disease states and
patient asthma attacks with asthma datasets. The knowledge
high-risk patients, design appropriate interventions, and reduce
that extracted from data mining process will be used by system
the number of hospital admissions and claims.
to give recommendations and restriction regarding to personal
There are already many studies which focus on the etiology trigger factor of asthma attack. The knowledge also can be
and diagnosis of asthma. O'Leary study emphasized patients’ used to predict the possibility of asthma attack and its degree
DNA and found the allergen of asthma [7]. Jan et al. proposed severity from patient activity and condition. The architecture of
a system named as Blue Angle Asthma Guardian System, the proposed system illustrate on Figure 1.
which is designed for children asthma monitoring [8].
O’Sullivan et al. propose to incorporate formalized external
expert knowledge in building a prediction model for asthma
exacerbation severity for pediatric patients in the emergency
department [9]. Tseng.proposed an integrated bio-signal data
mining system with two data mining methods for predicting
asthma attacks [10]. Meanwhile, Kudyba stated that strong
disease management programs such as asthma management,
depend on data mining technique[11].
III. PERSONALIZED ASTHMA MANAGEMENT SYSTEM WITH K-

MEAN AND RULE BASED CLASSIFICATION
In this section, we will explain the problem definition,
system architecture and data mining methods for the proposed
system. This section will also explain how well asthma
management system and data mining are integrated and also
describes the datasets undertaken for theproposed system.
A. Problem Definition
Fig. 1. Architecture Of Personalized Asthma Management System
Asthma is a multifactorial chronic illness, there are no
uniformity of triggers factors of asthma attack. each patients As shown on Fig 1, the source data for data mining proces
has different trigger factors that lead to different degree of are come from four major database, with categorize as
asthma severity. Triggers are things that may bother the explained on problem definition. All data from all database
airways, making it hard to breathe or cause asthma signs or would preprocessed before its could be extracted to gain
symptoms. Ideally, any asthma patient should know their pattern and knowledge, to minimize error and integrity of data.
trigger factors, so they can avoid their triggers whenever
possible. But the problem is, triggers factors of asthma attack
can be different from person to person. The triggers varies from C. Data Mining Methods
gender, age, BMI, environmental and living condition, This paper proposed personalized asthma management
variations in lifestyle, allergic factors, daily activity and system with three data mining methods as illustrate on Fig.1.
exercise, eating habits, hormonal factors, and stress condition. this following part will explained in detail how data mining
techniques integrated to gather the predictive trigger factors of
On this proposed system we divided the trigger factors of asthma attack with its degree of severity.
asthma attack into three major categories:
121
The first phase of this research is to build a personalized 1. Define the number of clusters to be founded by k-
asthma monitoring application. The application will record all means, which in this case is 3;
suspected asthma trigger factors with its degree of asthma 2. k-means starts by initializing a number of
severity daily. A dataset that will be used for data mining prototypes equal to number of desired clusters,
processes would be generated from data warehouse of 3. then the system will assigning each point to its
personalized asthma monitoring application. The database closest center, then moving each prototype to the
would contain the following 4 major variables of particular mean of its assigned points.
interest to Asthma management: gender, age, body mass index 4. Reapeat step three (3) until it converges to a
(BMI), allergic status, the number of times a patient exercises solution.
per week, stress level, onset of an asthma attack and its
possible excurbations. Decision support system is connected
The popular heuristics for solving the k-means problem
with the database management [12]. All trigger factors become
target variable on the data mining process. It measured by a is based on a simple iterative scheme for finding a locally
minimal solution. K-means algorithm works optimally with
dichotomous variable indicating severity degree of the asthma
attack. For this kind of integrated dataset, we adopted categorical and numeric data so that this is the best for binary
dataclustering. It is simple and fairly fast [15], results are easy
sequential pattern mining to discover high risk prefix
sequential patterns of asthma attacks and to segmenting the to interpret and it can work under a variety of conditions hence
it stand as the standard algorithm for clustering.
data, based on the frequency and severity of asthma attack.
The final step on data mining process is tobuilt a classifier
The next step in data mining process is to build clusters
for each cluster with the extracted patterns. Once the rules are
with the extracted patterns. In more detail, clustering the
learned from a clustering, they can be used for predicting the
trigger factors that have the same asthma status and the
class type of previously unseen data, with classification
degree of severity to help discovering asthma trigger rules.
technique. Classification is the most commonly applied data
Clustering uses to make groups of objects, eachgroups has
mining technique, which employs a set of pre-classified
similar characteristics but has different characters from the
examples to develop a model that can classify the population of
objects in another group. Ussually the suitable number of
records at large. This approach frequently employs decision
clusters is not clearly shown. One difficulty in this proposed
tree or neural network-based classification algorithms. The data
data mining process is that we don’t have any prior
classification process involves learning and classification. In
knowledge about the structure of the data, or its labels,
Learning the training data are analyzed by classification
because clustering is considered to be an unsupervised
algorithm. In classification test data are used to estimate the
learning problem [13]. The data flow of trigger factors of
accuracy of the classification rules. If the accuracy is
asthma attack with cluster analysis is illustrate on Figure 2.
acceptable the rules can be applied to the new data tuples [16].
We used Classification trees to build decision rules in a
hierarchical fashion to make a classification of severity degree
of asthma attack as illustrate on Figure 3.
Fig. 1.Data Flow of Trigger Factor of Asthma Attack with cluster analysis
In this proposed system, data are clustered into three groups

according to their asthma attack severity and we adopt K
means algorithm for that purpose. K-means algorithm works Fig. 2.Decision Tree analysis
optimally with categorical and numeric data so that this is the
best for binary data clustering. It is simple and fairly fast A node in the decision tree represents a decision rule. A
[14],results are easy to interpret and it can work under a variety leaf in the decision tree represents one possible cluster as has
of conditions hence it stand as the standard algorithm for been extracted from clustering methods in the previous step.
clustering. The fundamental idea is to find K average or mean The diagram below illustrates how the reduced data can
values, about which the data can be clustered. The k-means provide one or more node statistics used in the decision tree. A
algorithm is a simple iterative method to partitiona given decision tree will give requisite model which provides
dataset into a user specified number of clusters K. K-means is sufficient insights into the asthma management problem
initialized from some random or approximate solution. The The classification tree is used to group the severity of
algorithm works as follows: asthma attack into four major groups. We used classification as
122
has been done by Nagori [17], which are mild intermittent, Clustering techniques used for segmenting severity degree of
mild persistent, moderate persistent and severe persistent.. The asthma attack. Then we used classification tree for predicting
decision tree algorithm used for the proposed system based on the likeliness of asthma attacks. This study will help asthma
following steps: patients to have a better understanding of her/his asthma trigger
factors, that could lead to improvements on their daily activity.
1. Check for basic symptoms of asthma attack It also to identify those trigger factors and taking appropriate
2. For each symptom find normalization of medical action at the right time. The future work of this
information gain and find the best attribute from research is to build the prototype system and generated data for
external and internal factor from the highest the mining proposed. We will also explore the most suitable
normalizad of information gain. data mining algorithm comparingseveral algorithms and find
the best precision and recall.
3. Create a decision node that split into 4 groups
The severity group based on various symptoms when REFERENCES
patient having asthma attack such as wheezing, coughing,
shortness of breathing, chest tightness etc. Table 1 shows
[1] Global Initiative for asthma management and prevention, A pocket
exactness of the four groups for asthma severity attack that will guide for physicians and nurses, GINA, 2012, pp 5-7
be used on classification trees on the proposed system. [2] Riset Kesehatan Dasar ,Badan Penelitian dan Pengembangan Kesehatan,
Kementrian Kesehatan RI, 2013
TABLE I. GROUPS OF ASTHMA SEVERITY ATTACK [3] Gruffydd-Jones K. Measuring pulmonary function in practice.
Practitioner 2002; 246: 445–449. Medline
Asthma Symptoms Of Airflow Night Time Peak [4] Global Initiative for Asthma (GINA). Global strategy for asthma
Attack Obstruction Expiratory management and prevention. Updated 2011. Cape Town: University of
Severity Symptoms Flow* Cape Town Lung Institute; 2011
Degree
[5] Cios, K.J: Medical data mining and Knowledge Discovery. IEEE
Severe • Continuous frequent >60% of engineering in medicine and biology 7. (2000) pp 15-16
Persistent personal [6] Milovic Boris and Milan Milovic, Prediction and Decision Making in
• Limited physical activity best Healthcare Using Data Mining. International Journal of Public Health
Science (IJPHS) Vol 1, No.2,December 2012,pp.69-78
Moderate • More than once a day More than Between
Persistent once a week 60% to 80% [7] O'Leary, M. Koolpiruck, D. Balachandran, W. Emberlin, J. Lewis, R.
• Attack affect daily of personal “The role of electrostatic charge accumulated by respirable sized
activity best allergens with regard to thunderstorm asthma,” Industry Applications
Conference, 2005.
Mild • More than once a week At least twice More than
[8] R. L. Jan, J. Y. Wang, M. C. Huang, S. M. Tseng, H. J. Su, L.F. Liu.
Persistent but less than once a day a month 80% of
personal “An Internet-based interactive telemonitoring system for improving
best childhood asthma outcomes in Taiwan.” Telemedicine and e-Health,
2006, pp. 1-28.
Mild • Less than once a week Less than More than [9] Dympna O’Sullivan, William Elazmeh, Szymon Wilk, Ken Farion, Stan
Intermittent twice a month 80% of Matwin, Wojtek Michalowski, and Morvarid Sehatkar, “Using
personal Secondary Knowledge to Support Decision Tree Classification of
best Retrospective Clinical Data”, MCD 2007, LNAI 4944, pp. 238–251, 200
*mandatory [10] Tseng Vincent S., Lee Chao-Hui,and Chen Jessie Chia-Yu. “integrated
Bio-Signal Data Mining with Applications on Asthma Monitoring”,
Therefore, on this proposed system, to build comprehensive National Cheng Kung University, Taiwan
rules to gain good knowledge of a personal trigger factor of [11] Kudyba, S: Managing data mining : advice from experts. (2004)
asthma attack,then we will validate the rules extracted with [12] Zhou Qishen, Z. Yin, Q. Ying, W. Shahnhui, Intelligent Data Mining
medical expert. The rule that’s been used and validated from and Decision System for Commercial Decision Making.
TELEKOMNIKA Indonesian Journal of Electrical Engineering Vol.12,
classifier process, will be used to predict asthma attacks by the No 1, January 2014,pp.792-801
matching patient’s personal information, asthma allergy
[13] G.Gan,C.Ma, J Wu. “ Data Clustering Theory, Algorithm, and
records, stress level and internal and external trigger factor as Applications”. American Statistical Association Alexandria, Virginia,
input data. When a prefix sequential pattern is matched in these 2007
data, system could remind users to take medicines or to seek [14] Arun K Punjari, ―Data Mining Techniques‖, Universities (India) Press
medical treatment based on the prediction of asthma attack Private Limited, 2006
severity. [15] P. Bradley, U. Fayyad and C. Reina, "Scaling Clustering Algorithms to
Large Databases", In Proc. of the 4th ACM SIGKDD, pp. 9-15, 1998.
[16] Brijesh Kumar Baradwaj, Saurabh Pal, Data mining: machine learning,
IV. CONCLUSION AND FUTURE WORK statistics, and databases, 1996.
This research is concerned with the study and analysis of [17] Meghana Nagori, Pawar Suvana and Vivek Kshirsagar, “ Managing
data mining with data clustering and classification algorithms, asthma in children and analyzing best possible treatment with data
analyzing the existing methods for predicting trigger factors mining approach of classification”, in Proc of WCECS 2009, San
Francisco, USA.
of personalized asthma attack and to design an efficient and
effective method for predicting severity of asthma attack based
on personalized patient trigger factors. We proposed K-Means
123
Proceeding of International Conference on Electrical Engineering, Computer Science and Informatics (EECSI 2015), Palembang, Indonesia, 19 -20 August 2015
Decision Support System for Heart Disease

Diagnosing Using K-NN Algorithm
Tito Yuwono Noor Akhmad Setiawan, Hanung Adi Ipin Prasojo, Sri Kusuma Dewi, Ridho
Department of Electrical Engineering Nugroho, Anugrah Galang Persada Rahmadi
Islamic University of Indonesia Department of Electrical and Information Department of Informatic
Yogyakarta Technology Islamic University of Indonesia
Address: Kaliurang Street KM 14 Gadjah Mada University Yogyakarta
Yogyakarta, Indonesia Address: Jalaan Grafika No 2 Yogyakarta Address: Kaliurang Street KM 14
Email:tito@uii.ac.id Email:noorwewe@ugm.ac.id, Yogyakarta, Indonesia
adinugroho@ugm.ac.id Email:rm_prasojo@yahoo.com, cicie,
ridhorahmadi@uii.ac.id
Abstract— Heart disease is a notoriously dangerous disease which feature extraction [8], and a software development for automatic
possibly causing the death. An electrocardiogram (ECG) is used for detection [9].
a diagnosis of the disease. It is often, however, a fault diagnosis by a This present work provides an implementation of computer
doctor misleads to inappropriate treatment, which increases a risk assistance to help the diagnosis. We implement a method called
of death. This present work implements k-nearest neighbor (K-NN) k-nearest neighbor (K-NN) on ECG data which intended to give
on ECG data to get a better interpretation which expected to help a
a better and accurate interpretation.
decision making in the diagnosis. For experiment, we use an ECG
data from MIT BIH and zoom in on classification of three classes; Furthermore, some practical purposes that motivate our study
normal, myocardial infarction and others. We use a single decision are: (1) to help Physicians in interpreting the ECG signal, (2)
threshold to evaluate the validity of the experiment. The result To transform the ECG signal into a form that is ready for a
shows an accuracy up to 87% with a value of K = 4. further process of computation.
Keywords—K-NN; ECG; Diagnosis; Heart Disease The remainder of the paper is organized as follows: the
methodology is described in Section II, the experimental results
is discussed in Section III, and the conclusion of this study is
summarized in Section IV.
I. Introduction
Heart disease (cardiovascular) is a notoriously dangerous II. methodology
disease. According to World Health Organization (WHO), the A. k-Nearest Neighbor
disease is a major cause of death [1]. Moreover, The Department K-nearest neighbor (k-NN) is a non-parametric technique
of Health of the Republic of Indonesia also states that the for classification and regression. It is an instance-based learning
disease is one of the main causes of the death in Indonesia [2]. method to group data points based on the majority class in a
Some alternatives have been suggested to decrease the risk neighborhood with size k. The basic idea is, to find the k nearest
of death caused by this disease; an early screening of heart neighbors for each data point, and to classify it based on the
activity, and an accurate diagnosis. An electrocardiogram (ECG) majority. In order to do that, the distances between this
is used to check the disease. It is often, however, a diagnosis that particular data point to the other k data points are computed.
is conducted by a doctor based on the ECG data misleading to Technically k-NN works as the following steps. Suppose
an inappropriate decision. A computer-aided diagnosis is an we are given a training data set (Xi,Yi),....,(Xn,Yn) (with m-
alternative to solve this problem. dimensional attribute vector X and its corresponding class Y.
Several previous studies have been conducted. For Each X is classifiable into P classes, namely C1,....,Cp Then we
example, to use the Wavelet and template matching method to are given a new instance Xn+1 to classify.
measure the QT interval of the ECG signal [3], or to use pattern The k-NN first computes the distances between Xn+1 and
recognition [4][5][6], or to use EMD (Empirical Mode X1,...,Xn by means of Euclidean distance measurement.
Decomposition) combined with R peak detection
method and CWT (Continuous Wavelet Transform)[7], or to use

generalized tensor rank one discriminant analysis on
160
ECG Signal ECG File

𝑑𝑖 = √∑𝑚
𝑗=1( 𝑋𝑛+1 − 𝑋𝑖𝑗 )
2 (1)
A. Block Diagram
In this section we provide a block diagram as a
visualization of the steps of our implementation. Figures 1 and 2
give sketches of this block diagram. First we collect the ECG Data Extraction
signal and ECG files (containing prior knowledge base) as the
input. The distance metric is then computed between the new
ECG signals and the prior knowledge base, based on an attribute
vector, namely PR interval, PR segment, QRS interval, ST
segment and QT interval. This step assigns a new instance of PR Interval PR Segment QRS Interval ST Segment QT Interval
ECG signal into a class.

The prior knowledge base that we plug in contains three
classes, representing three different types of patient states:
disease Healthy Controls (normal), Myocardial Infarction, and
the other. The distribution of the disease as shown in Figure 3
and Appendix A. Inference Machine
K-NN
ECG Signal Knowledge Base
Prediction of
Expert System Diagnosis
Prediction of Diagnosis Fig. 2. The model of the decision support system

Inference Machine
Name of Disease
ECG File K-NN
Knowledge
Base
Normalothers
Fig. 1. Block Diagram of the intelligence system for ECG interpretation. Infarc
Others
Normalothers
Infarc
Others
Fig. 3. Heart disease distribution

((PRinterval x QRSinterval), (STsegment x QTinterval)) and ((PRinterval
x STsegment), (QRSinterval x QTinterval)).
161
III. Experimental result

Table 1 lists the classes of diseases in the prior knowledge base. The accuracy of the system can be determined by
calculating the value of TP, TN, FP and FN of Table 4.
TABLE 1. THE DISTRIBUTION OF DISEASE CLASSES TP= 32 + 70 + 15 = 117
No Disease class Number of instancess TN= (70 + 15) + (32 + 15) + (32 + 70) = 234
1 Health Control (Normal) 41 FP= (8 + 2) + (8 + 6) + (1 + 1)= 26
2 Myocardial infarction 79 FN= (8 + 1) + (8 + 1) + (2 + 6) = 26
3 Others 23
TP  TN
The accuracy T  x100%
TP  TN  FP  FN
In order to test the performance of our implementation, we 117  234
T x100%  87%
conduct a validity test. This test is to compare the result of 117  234  26  26
prediction to the real outcome. The result of the validity test is Based on the above calculation can be seen that the level of
described in Table 4 in Appendix A. accuracy of the expert system that has been created is 87%.
Furthermore we use single decision threshold to determine Several algorithms have been used with each its accurattion are
the accuracy. The computation is based on the confusion matrix presented in Table 3.
as follows.
TABLE 3. THE COMPARATION OF ALGORITHM [10]
Algorithm Control Ventricula Myocardial Total of
Patiens r Infarction Accuration
N=382 Hypertrop N=547 (%)
hy N=1220
N=291
Padova 89.8 61.3 47.1 62.0
Nagoya- 89.3 42.6 63.7 65.6
Fukuda
IBM Medis 91.3 49.4 62.5 67.6
HP (Agilent) 93.5 51.0 64.5 69.3
1. TP (True Positive) is for correct prediction of a patient’s Glasgow 94.0 51.0 67.7 69.7
disease. GE 86.3 61.1 69.7 69.8
(Marquette)
2. TN (True Negative) is for a correct prediction that a patient Means 97.1 42.5 67.2 69.8
does not suffer from the disease. Hannover 86.6 72.1 79.0 75.8
3. FP (False Positive) is for a prediction that states a patient Louvaine 91.5 67.0 82.1 77.3
suffers from a disease, which actually not. (Louven)
8
4. FN (False Negative) is for a prediction that states a patient
Cardiologists 97.1 60.4 80.3 79.2
does not suffers from a disease, which actually yes. Combined
Scores
Table 2 shows comparison between reality with the diagnosis
system
Table 3 is a publication of QRS diagnostic using variation of
TABLE 2. THE COMPARISON BEETWEN THE REAL DIASEASE WITH algorithms. This table shows Louvaine has the best total
THE DIAGNOSIS accuracy with 77.3%. The accuracy for diagnosing Control
Diagnosis using K-NN Patiens (normal) is 91.5%, Ventricular Hypertrophy is 67.0%
Myocardial and Myocardial Infarction is 82.1%.
Diagnosis Normal Others
Infarction
Normal 32 8 1
Disease
Real
Myocardial
8 70 1
Infarction
Others 2 6 15 IV. Conclusion
The accuracy achieved in this study is 87% with k=4. Although
this percentage is incomparable with the results provided in
Table 5 (as the sample size is different), but our result shows
that k-NN is able to return a reasonably good result. As the
future work, we suggest to increase both number of diseases and
members of each class.
162
Acknowledgment 50 Myocardial Infarction Myocardial Infarction

Thanks to the minister of research-technology and higher 51 Normal Normal
52 Normal Normal
education that has funded this research. 53 Others Myocardial Infarction
54 Normal Normal
55 Normal Normal
56 Others Myocardial Infarction
57 Normal Normal
Appendix A 58 Normal Myocardial Infarction
59 Others Others
60 Others Others
TABLE 5. THE DIAGNOSIS VALIDATION FOR K=4 61 Myocardial Infarction Normal
62 Others Myocardial Infarction
No Real Deasease Diagnosis 63 Normal Normal
1 Myocardial Infarction Myocardial Infarction 64 Myocardial Infarction Myocardial Infarction
4 Myocardial Infarction Myocardial Infarction 67 Myocardial Infarction Normal
6 Myocardial Infarction Myocardial Infarction 69 Others Others
8 Myocardial Infarction Myocardial Infarction 71 Normal Myocardial Infarction
11 Myocardial Infarction Myocardial Infarction 74 Others Normal
13 Myocardial Infarction Myocardial Infarction 76 Others Myocardial Infarction
14 Myocardial Infarction Normal 77 Normal Normal
16 Myocardial Infarction Normal 79 Others Others
17 Myocardial Infarction Myocardial Infarction 80 Normal Normal
20 Myocardial Infarction Others 83 Others Others
31 Myocardial Infarction Myocardial Infarction 94 Myocardial Infarction Normal
32 Myocardial Infarction Myocardial Infarction 95 Others Myocardial Infarction
38 Myocardial Infarction Normal 101 Others Others
163
113 Normal Normal

139 M M M M M M M M M M M M
114 Normal Normal
115 Normal Myocardial Infarction 140 L L M M L L N L N N L L
116 Normal Normal 141 M M M M M M M M M M M M

117 Normal Myocardial Infarction
118 Normal Normal
119 Myocardial Infarction Myocardial Infarction 143 M M M M M M L L N L L N
120 Normal Normal

121 Myocardial Infarction Myocardial Infarction Perbedaan
0 37 26 40 43 55 46 53 50 48 51
Diagnosis
122 Normal Normal
123 Normal Normal
124 Myocardial Infarction Myocardial Infarction
125 Normal Others References
126 Normal Normal [1] WHO, Cardiovascular Disease,2012.
127 Myocardial Infarction Myocardial Infarction [2] Menkes RI,”Situasi Kesehatan Jantung,” Pusat Data dan Informasi
128 Myocardial Infarction Myocardial Infarction Kemenkes RI, 2014.
129 Myocardial Infarction Myocardial Infarction [3] Kinckmerova, Metodhes for Detection and Classification in ECG analysis.
130 Myocardial Infarction Myocardial Infarction Doctoral Dissertation, Brno University of Technology, Brno, 2009.
131 Myocardial Infarction Myocardial Infarction [4] K.O. Gupta & P.N. Chatur, “ECG Signal Analysis and Classification using
132 Others Normal Data Mining and Artificial Neural Networks, “International Journal of
133 Normal Normal Emerging Technology and Advanced Engineering, Vol 2, Issue 1, 2012.
134 Normal Normal
[5] J.Kaur & J.P.S Raina, “An Intelligent Diagnosis System for
135 Myocardial Infarction Myocardial Infarction Electrocardiogram (ECG) Images Using Artificial Neural Network
136 Myocardial Infarction Myocardial Infarction (ANN),” International Journal of Electrical, Electronics and Computer
137 Myocardial Infarction Myocardial Infarction Engineering 1(1):47-51(2012)
138 Normal Normal [6] J.Kaur & J.P.S Raina, “An Intelligent Diagnosis System for
139 Myocardial Infarction Myocardial Infarction Electrocardiogram (ECG) Images Using Artificial Neural Network
140 Others Myocardial Infarction (ANN),” International Journal of Electrical, Electronics and Computer
141 Myocardial Infarction Myocardial Infarction Engineering 1(1):47-51(2012)
142 Myocardial Infarction Myocardial Infarction [7] Priyadarshini, ECG Signal Analysis: Enhancment and R-Peak detection.
143 Myocardial Infarction Myocardial Infarction Natioal Institut of Rourkela, India, 2010.
[8] K.Huang & L. Zhang, “Cardiology knowledge free ECG feature extraction
using generalized tensor rank one discriminant analysis,”EURASIP Journal
Appendix B and Advances Signal Processing, 2014.
[9] M.H.Imtiaz & M.A. Kiber, “Design and Implementation of a Real-Time
TABLE 6. THE VARIATION OF K VALUES Automated ECG Diagnosis (AED) System,” Gobal Journal of Researches
in Engineering and Electronics Engineering, Vol 13 Issue 11,2013.
No
Real
diasease
K Variation [10] J.L. Willems, et al., “The Diagnostic Performance of Computer Programs
K=2 K=3 K=4 K=5 K=6 K=7 K=8 K=9 K=10 K=11 K=12 for the Interpretation of Electrocardiograms”, New England Journal of
Medicine, 325:1767-1773,1991.
2 M M M M N N N M N N N N
3 M M M M N M N N N N N N
5 M M N M M M M M M M M M
6 M M N M M M M M M M M M
.
.
135 M M M M M M N N M M M M
137 M M M M N N N N N N N N
138 N N M N N N N N N N N N
164
IOP Conference Series: Materials Science and Engineering
PAPER • OPEN ACCESS
Noise Reduction in Breath Sound Files Using Wavelet Transform Based

Filter
To cite this article: M F Syahputra et al 2017 IOP Conf. Ser.: Mater. Sci. Eng. 190 012040
View the article online for updates and enhancements.
This content was downloaded from IP address 138.68.70.63 on 19/03/2018 at 02:56

International Conference
IAES International on Recent
Conference Trends inEngineering,
on Electrical Physics 2016Computer
(ICRTP2016)
Science and Informatics IOP Publishing
Journal of Physics:
IOP Conf. Conference
Series: Materials Seriesand
Science 755 (2016) 011001
Engineering doi:10.1088/1742-6596/755/1/011001
190 (2017) 012040 doi:10.1088/1757-899X/190/1/012040
1234567890
Noise Reduction in Breath Sound Files Using Wavelet

Transform Based Filter
M F Syahputra1, S I G Situmeang2, R F Rahmat1, R Budiarto3

1
Department of Information Technology, Faculty of Computer Science and
Information Technology University of Sumatera Utara, Medan, Indonesia
2
College of Engineering, National Chung Cheng University, Taiwan
3
Department of Information System College of Computer Science and Information
Technology, Albaha University, Saudi Arabia
E-mail: nca.fadly@usu.ac.id, samuel103m@cs.ccu.edu.tw, romi.fadillah@usu.ac.id,

rahmat@bu.edu.sa
Abstract. The development of science and technology in the field of healthcare increasingly
provides convenience in diagnosing respiratory system problem. Recording the breath sounds
is one example of these developments. Breath sounds are recorded using a digital stethoscope,
and then stored in a file with sound format. This breath sounds will be analyzed by health
practitioners to diagnose the symptoms of disease or illness. However, the breath sounds is not
free from interference signals. Therefore, noise filter or signal interference reduction system is
required so that breath sounds component which contains information signal can be clarified.
In this study, we designed a filter called a wavelet transform based filter. The filter that is
designed in this study is using Daubechies wavelet with four wavelet transform coefficients.
Based on the testing of the ten types of breath sounds data, the data is obtained in the largest
SNRdB bronchial for 74.3685 decibels.
1. Introduction
To make diagnosis of a human respiratory system, health practitioner or doctor uses a device called a
stethoscope. With it, medical practitioners can hear the sound of the lungs to diagnose a person's
physical condition and health. Breath sounds heard through the stethoscope, is still mixed with other
noises like the sound of the heart, skin friction with the stethoscope, and chest movements. It is
difficult for health practitioners to diagnose a person's physical condition and health. In signal
processing, other voices or extraneous noise is commonly called signal interference or noise.
The development of science and technology in the field of healthcare increasingly provide
convenience in diagnosing respiratory system problem. Breath sounds recording is one example of
these developments. Breath Sounds are recorded using a digital stethoscope, and then stored in a file
with sound format. As with conventional methods, respiratory sounds are not free from interference
signals. Therefore, filter noise or signal interference reduction is required so that breath sounds
components that contain information signal can be clarified. Research on reducing signal interference
had been done since the last few years, but has not been able to give a satisfying result.
Previous studies related to respiratory noise filter, including the separation of heart and lung sounds
from the breath sounds with modified spectro-temporal representation [1], and reduction of heart
sounds from lung sounds recordings using adaptive filter [2] have been carried out.
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
IAES International Conference on Electrical Engineering, Computer Science and Informatics IOP Publishing
IOP Conf. Series: Materials Science and Engineering 190 (2017) 012040 doi:10.1088/1757-899X/190/1/012040
1234567890
Furthermore, researches on the wavelet transform including separation of discontinuous

adventitious sounds from vesicular sounds using wavelet transform stationary-non-stationary filters
[3], the separation splits the non-stationary input signals (heart sounds) from the stationary part (lung
sounds) using a stationary-non-stationary wavelet transform filters [4], and the reduction of heart
sounds from lung sounds have been recorded at low and medium flow rate using wavelet transform
based filter [5] have also being done.
In this research, we applied wavelet transform based filter to reduce interference signals and white
noise in the breath sounds recording, so we get a clearer signal without losing important information
contained in the signal. With breath sounds recording filtered, health practitioner or doctor will get
more detailed information for a better diagnosing method.The difficulty of diagnosing the disease
through the breath sounds recording is that it is mixed with other noises which makes it difficult to
diagnose a person's physical condition and health. A wavelet transform based filter is used with
wavelet daubechies and global threshold in the breath sounds recording to produce a better breath
sounds signal for diagnosis of diseases in the respiratory. In short, this work attempts to reduce noise
so that the breath sounds recording can be used to provide a better diagnosis.
2. Methodology
2.1 Respiratory sounds
Respiratory sounds are all sounds related to respiration including breath sounds, adventitious sounds,
cough sounds, snoring sounds, sneezing sounds, and sounds from the respiratory muscles. Voiced
sounds during breathing are not included in respiratory sounds. [6].
Respiratory sounds can be classified into two groups: breath sounds and adventitious sound
(abnormal) [7]. Breath sounds produced from healthy subject’s chest called normal breath sounds.
Normal breath sounds include both the inspiration and expiration. Both occur when the air moves in
and out during regular breathing cycle. Adventitious sounds are additional respiratory sounds in the
breath sounds. This sound occurs unexpectedly during regular breathing cycle.
There are several types of abnormal breath sounds. The four most common types are rales, rhonchi,
wheezing, and stridor [8]. Rales are small clicking, bubbling, or rattling sounds in the lungs. They are
heard when a person breathes in (inhales). And are believed to occur when air opens in closed air
spaces. Rales can be further described as moist, dry, fine, and course. Rhonchi are sounds that
resemble snoring. They occur when air is blocked or air flow becomes rough through the large
airways. Wheezing are high-pitched sounds produced by narrowed airways. They are most often heard
when a person breathes out (exhales). Wheezing and other abnormal sounds can sometimes be heard
without a stethoscope. Stridor is wheeze-like sound heard when a person breathes. Usually it is due to
a blockage of airflow in the windpipe (trachea) or in the back of the throat. Fort the characteristics of
Lung Sound and Noise, the peak sound of normal lung is usually found at frequencies below 100 Hz,
where lung sound energy decreases sharply between 100-200 Hz, but can still be detected at or above
800 Hz with a sensitive tool [2]. According to Earis & Cheetham [9], noises such as the sound of the
respiratory muscles, chest motion sounds, heart sounds, and other low-frequency noise, have a
frequency between 50 to 150 Hz [9].
2.2 Wavelet transform
Fourier method specifies only the spectral content of a signal in the frequency domain. The
disappearance of time information during Fourier transformation for preservation during the incident
is not considered. This condition can be ignored if the signal is stationary. However, for stationary
signals such as speech, time and frequency data is important to avoid significant loss of information in
the signal. Wavelet analysis can be used as an alternative method to solve the problem in Fourier
method [10]. By using the concept of multi resolution wavelet analysis (for example, the
representation of time and frequency scaling) to produce precise decomposition of the signal to obtain
an accurate representation, detailed characteristics such as small discontinuities, similarity, and even
higher order derivation hidden by conventional Fourier analysis can be revealed.
2
1234567890
Wavelet is a family of functions ψa,b(t) derived from a base wavelet ψ(t), called the "mother
wavelet", by dilation and translation [11], as shown in (1) as an example.
1 𝑡−𝑏
𝜓𝑎,𝑏 (𝑡) = 𝜓( ),𝑎 > 0, 𝑏 ∈  (1)
√ 𝑎 𝑎
Wavelet analysis is basically shifting and scaling a limited form of energy called the mother
wavelet ψ(t) of the desired signal. So, that the discrete wavelet transform can be written as in (2).
𝜓𝑗,𝑘 (𝑡) = 2𝑗⁄2 𝜓(2𝑗 𝑡 − 𝑘) (2)
2.3 Signal-to-noise ratio
Signal-to-noise ratio can be generally defined as the dimensionless ratio of signal power to the noise
power contained in a record (3) [12].
𝑃𝑠𝑖𝑔𝑛𝑎𝑙 𝐴𝑠𝑖𝑔𝑛𝑎𝑙 2
𝑆𝑁𝑅 = =( ) (3)
𝑃𝑛𝑜𝑖𝑠𝑒 𝐴𝑛𝑜𝑖𝑠𝑒
where
Psignal = mean of signal power
Pnoise = mean of noise power
Asignal= root mean square (RMS) of signal amplitude
Anoise = root mean square (RMS) of noise amplitude
2.4 General architecture of the proposed method
Explanation of the components contained in the general architecture as shown in figure 2 is as follows:
a. Respiratory voice recording file is used as input. Respiratory sound is a combination of sound
lungs and signal interference (noise).
b. Sounds can be played and printed out in the form of a signal.
c. Then respiratory sound is read by system and decomposed into an array of type byte and
stored in the data [].
d. Array Data [] is converted into the form of an array of type double.
e. Array Data [] repeatedly decomposed according to the specified level of decomposition,
produces two arrays, each array having half of the length of the data array []. The first array
called a low pass filter and a second array called a high pass filter.
f. In each array, apply wavelet transform to the coefficients.
g. Both arrays are reassembled in the data array [] with a low pass filter placed in the first half
and the high pass filter is placed at the end of the half.
h. Array Data [] is passed through a threshold, resulting in two arrays, array and the array of
respiratory sound signal noise.
i. Perform repeatedly reconstruction as many as the level of reconstruction that has been
assigned to each array.
j. Change the order in the data array [] of the previous half and half low pass filter high pass
filter, be alternating low pass filter high-pass filter to each array.
k. Reapply the wavelet transform coefficients of each array.
l. Array Data [] and is then converted from an array of type double into an array of type byte.
Audio formats and file name that has been set, is applied to the data [].
m. An array of data [signal] breathing sounds is rebuilt into a breathing sound file and the data
array [] noise rebuilt into a beam of noise.
3
1234567890
PREPROCESS
Breath Sounds
Read file wav and Return Converting data[] become

as byte array in data[] double array
THRESHOLD DECOMPOSITION
Threshold Do decomposition for as

much decomposition
setted
Set LPF In the half of the
output and HPF in the LPF HPF
end of the output
Apply coefficient wavelet
Breath sound transform
Noise (vn)
signal (vbs)
REKONSTRUKSI FINAL PROCESS
Do reconstruction as much Apply coefficient wavelet convert data[] to byte

reconstruction setted transform array format
audio
Change the value order from hal LPF and HPF to File
Adjustment
LPF-HPF by turns name
noise Breath sounds
Figure 1. General Architecture
3 Experiments, Results and Discussion

3.1 Data
The data used in this research is breath sounds records, available from Littmann company. The
existing data can be viewed as 10 different sounds with 10 different type of sounds. The data is
described as: (type, source) = (bronchial, left lower lobe), (coarse crackles, right lower lobe), (fine
crackles with deciduous bronchial, right middle lobe), (fine crackles, lung basis), (inspiratory stridor,
tracea), (normal tracheal, tracea interscapular), (normal vesicular, right and left lower lobe), (pleural
friction, right middle lobe), (rhonchus, right lower lobe), (wheezing, left lower lobe).
3.2 System analysis
Systems analysis aims to identify the system’s development. Analysis is needed as a basis for system
design. In this study, there is a five preprocess stage which is decomposition of the signal, the
threshold stage, reconstruction stage, and the final stage process.
To test the designed system, a scenario is prepared to measure the sound of respiratory filter and
the impact of the system with the objective and subjective criteria. Objective criteria is the signal-to-
noise ratio (SNR) and signal display. While subjective criteria is the human auditory signal and the
noise which is being observed.
4
1234567890
A breathing sound recording of data, ie, bronchial signal used in this work is shown in Figure 2.
Figure 2. Bronchial sound signal before reduced

Measuring processes of noise reduction using wavelet transform based filter to the data in Figure 2 are
as follow.
1. Change the data into a byte array type. After the array of type byte obtained, change again into
the array of type double.
2. Perform second level decomposition of the data.
3. Perform the threshold stage using the following equation:
𝑡 = 𝜎. √2𝑙𝑜𝑔𝑁 (4)
4. Perform level 2 reconstruction to the data.
5. Change the data back into the form of a byte array type.
Figure 3 and Figure 4 exhibits the filtered Bronchial signal and the noise of Bronchial signal of Figure
2, respectively.
Figure 3. Bronchial signal after through filter
Figure 4. Noise signal after through filter
3.3 Results
Experiment is done to investigate the signal-to-noise ratio (SNR) of each data. The results are shown
in Table 1.
5
1234567890
Table 1. Noise Reduction Testing Results of Breath Sounds

No. Type SNR SNRdB
1. Bronchial 2.7343069662486665E7 74.36847268923987
2. Coarse Crackles 3703.999776899587 35.68670951851344
3. Fine Crackles with Deciduous Bronchial Sound 1136888.1450540074 60.557177378843406
4. Fine Crackles 333675.91665527434 55.23324862210189
5. Inspiratory Stridor 223332.40934887607 53.48951751195278
6. Normal Tracheal Sound 551447.7650258812 57.41504380900588
7. Normal Vesicular Sound 15976.861637033 42.03491474128431
8. Pleural Friction 4389520.347856363 66.42417066561016
9. Rhonchus 294575.37479824974 54.69196438921298
10. Wheezing 686282.2578168823 58.365027713927475
4 Conclusion
The experimental results showed that the biggest SNRdB value is 74.36847268923987 decibel which is
obtained from the reduction of bronchial. If the threshold value is too large, the noise cannot be
reduced from the breath sounds signal. Conversely, if the threshold value is too small, breath sounds
signal will loose a lot of important information.
As for our future work, we plan to utilize machine learning methods in order to increase the accuracy
of the breath sounds analysis. In addition, we are not going to use the filters that have the level of
decomposition and reconstruction greater than 2.
References
[1] Falk T H and Chan W 2008 Modulation filtering for heart and lung sound separation from
breath sound recordings Proc. the 30th Annual Int. Conf. the IEEE Engineering in Medicine
and Biology Society, p 1859-1862
[2] Sukresno F, Rizal A and Iwut I 2009 Reduksi suara jantung dari rekaman suara paru-paru
menggunakan filter adaptif dengan algoritma recursive least square Proc. SENTIA 2009, p
A1-A7
[3] Hadjileontiadis L J and Panas S M 1997 Separation of discontinuous adventitious sounds from
vesicular sounds using a wavelet-based filter IEEE Transaction on Biomedical Engineering
44 1269-1281.
[4] Hadjileontiadis L J and Panas S M 1998 A wavelet-based reduction of heart sound noise from
lung sounds Int. Journal of Med. Informatics 52 183-190
[5] Hossain I and Moussavi Z 2003 An overview of heart-noise reduction of lung sound using
wavelet transform based filter Proc. the 25th Annual Int. Conf. the IEEE EMBS p 458-461
[6] Sovijärvi A R A, Dalmasso F, Vanderschool J, Malmberg L P, Righini G and Stoneman S A T
2000 Definition of terms for applications of respiratory sounds European Respiratory Review
2000 10 597-610
[7] Baydar K S, Ertuzun A and Kahya Y P 2003 Analysis and Classification of Respiratory Sounds
by Signal Coherence Method. Proc. the 25th Annual Int. Conf. the IEEE EMBS p 2950-2953
[8] Schriber A 2011 Breath sounds. http://www.nlm.nih.gov/medlineplus/ency/article/007535.htm
[9] Earis J E and Cheetham B M G 2000 Current methods used for computerized respiratory sound
analysis European Respiratory Review 10 586-590
[10] Walker S L & Foo S Y 2003 Optimal wavelets for speech signal representations Systemics,
Cybernetics and Informatics 1 44-46
[11] Cohen A and Kovačeviċ J 1996 Wavelets: The mathematical background. Proc. IEEE p 514-
522.
[12] Johnson D H 2006 Signal-to-Noise Ratio. http://www.scholarpedia.org/article/Signal-to-noise-
ratio

Komunitas PDF

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Komunitas PDF

Diunggah oleh

Hak Cipta:

Format Tersedia

GAMBARAN DERAJAT KLINIS TERHADAP FAKTOR RESIKO PADA PASIEN

REAKSI HIPERSENSITIVITAS DI RUMAH SAKIT SANGLAH DENPASAR PERIODE

I Gede Aswin Arinata1, I Ketut Suardamana2

Kata Kunci: Reaksi Hipersensitivitas, faktor risiko, derajat klinis

Keyword : Hipersensitivity reactions, risk factors, clinical degree

PEMBAHASAN hipersensitivitas serupa dengan penelitian

Karakteristik Klinis dan Patologis Karsinoma Nasofaring di

Jurnal Kesehatan Andalas. 2016; 5(1)

PENDAHULUAN tahun dan menurun setelah umur 60 tahun. Angka

Jurnal Kesehatan Andalas. 2016; 5(1)

Jurnal Kesehatan Andalas. 2016; 5(1)

Jurnal Kesehatan Andalas. 2016; 5(1)

75,00% Rouvier sampai ke ruang retrofaringeal dan berlanjut

Jurnal Kesehatan Andalas. 2016; 5(1)

Jurnal Kesehatan Andalas. 2016; 5(1)

Jurnal Kesehatan Andalas. 2016; 5(1)

Design of Personalized Asthma Management System

III. PERSONALIZED ASTHMA MANAGEMENT SYSTEM WITH K-

In this proposed system, data are clustered into three groups

Decision Support System for Heart Disease

method and CWT (Continuous Wavelet Transform)[7], or to use

ECG Signal ECG File

ECG signal into a class.

ECG Signal Knowledge Base

Prediction of Diagnosis Fig. 2. The model of the decision support system

Fig. 3. Heart disease distribution

III. Experimental result

Acknowledgment 50 Myocardial Infarction Myocardial Infarction

113 Normal Normal

116 Normal Normal 141 M M M M M M M M M M M M

120 Normal Normal

PAPER • OPEN ACCESS

Noise Reduction in Breath Sound Files Using Wavelet Transform Based

View the article online for updates and enhancements.

This content was downloaded from IP address 138.68.70.63 on 19/03/2018 at 02:56

Noise Reduction in Breath Sound Files Using Wavelet

M F Syahputra1, S I G Situmeang2, R F Rahmat1, R Budiarto3

E-mail: nca.fadly@usu.ac.id, samuel103m@cs.ccu.edu.tw, romi.fadillah@usu.ac.id,

Furthermore, researches on the wavelet transform including separation of discontinuous

Read file wav and Return Converting data[] become

Threshold Do decomposition for as

REKONSTRUKSI FINAL PROCESS

Do reconstruction as much Apply coefficient wavelet convert data[] to byte

noise Breath sounds

Figure 1. General Architecture

3 Experiments, Results and Discussion

Figure 2. Bronchial sound signal before reduced

Figure 3. Bronchial signal after through filter

Figure 4. Noise signal after through filter

Table 1. Noise Reduction Testing Results of Breath Sounds

Anda mungkin juga menyukai