Uji Diagnostik Pemeriksaan Tulang Osteolitik Berbasis Intensitas Citra Digital

Susilo Susilo, Maesadji Tjokro Nagoro, Kusminarto Kusminarto, Wahyu Setia Budi


Abstract


Diagnostic test of osteolytic bone examination based on digital image intensity

Background: The visual examination of bone radiographs using digital computed radiography (CR) is an examination for the diagnosis of bone-metastatic cancer. The subjectivity of interpretation of bone radiographs may lead to doctor’s doubt in making decision to treatment patients with bone-metastatic cancer. Software Matlab-based computer application program makes a standard method to organize the results of bone radiographs. The objective of this study is to develop a software based on Matlab to analyze the diagnostic values, and to determine the optimal of cut off point to diagnose of osteolytic bone.

Method: The researches data are collected from Department of Radiology of three hospitals i.e. Dr. Kariadi Hospital Semarang, Dr. Sardjito Hospital and Bethesda Hospital Yogyakarta. This research was carried out during four month from April to August 2009. Radiographs of osteolytic bone interpreted by radiologist were compared with PA examination result of the osteolytic bone which were viewed as the gold standard. The steps in this study i.e. patients are classified as a normal or osteolytic bone patients based on the cut off point that had been determined, calculate the value of the diagnostic test using 2x2 tables, determined the area under the curve (AUC) by the procedure of receiver operating characteristic (ROC), and determined the optimal of cut off point.

Result: The results of study show that the diagnostic test for osteolytic bone by using Matlab-based software has sensitivity of 0.88, specificity of 0.891, positive expected value of 0.897, negative expected value of 0.950 and the cut off point at 0.93, while, the value of area under the curve (AUC) is 94% (95% CI: 89.7%-98.3%), and the accuracy is 0.881 for the case of osteolytic bone.

Conclusion: Matlab-based software being used for diagnosing osteolytic bone has relatively high sensitivity and specificity.

Keywords: Digital image, bone metastases, osteolytic, optimum cut off point

 

ABSTRAK

Latar belakang: Pemeriksaan radiograf tulang secara visual menggunakan sistem radiografi digital CR (computed radiography) merupakan pemeriksaan untuk diagnosis kanker metastasis tulang. Subyektivitas interpretasi radiograf tulang dapat menyebabkan keraguan dokter dalam mengambil keputusan untuk pengobatan pasien dengan kanker tulang metastatik. Software berbasis program aplikasi computer Matlab membuat suatu metode standard untuk mengorganisasikan hasil radiograf tulang. Tujuan penelitian
adalah mengembangkan software berbasis Matlab untuk menganalisis nilai-nilai diagnostik, cut off point optimal dan akurasi pemeriksaan pada diagnosis tulang osteolitik.
Metode: Data penelitian diambil di bagian radiologi dari tiga rumah sakit, yaitu RSUP Dr. Kariadi Semarang, RSUP Dr. Sardjito dan Rumah Sakit Bethesda Yogyakarta. Penelitian dilakukan selama empat bulan dari April sampai Agustus 2009. Radiograf tulang  osteolitik yang diinterpretasikan oleh radiolog ini dibandingkan dengan hasil pemeriksaan PA tulang osteolitik yang dianggap sebagai gold standard. Langkah yang dilakukan dalam penelitian ini adalah pasien diklasifikasikan sebagai pasien tulang normal dan pasien osteolitik berdasar cut off point yang telah ditetapkan, menghitung nilai uji diagnostik menggunakan tabel 2x2, menghitung luasan di bawah kurva (AUC) dengan cara receiver operating characteristic (ROC), serta menetapkan cut off point optimal.

Hasil: Hasil penelitian menunjukkan bahwa uji diagnostik tulang osteolitik menggunakan software berbasis Matlab memiliki sensitivitas 0,875, spesifisitas 0,891, nilai duga
positif 0,897, nilai duga negatif 0,950 dan cut off point 0,93. Nilai luasan di bawah kurva (AUC) ROC adalah sebesar 94%


Keywords


Digital image, bone metastases, osteolytic, optimum cut off point

Full Text:

PDF


Visitor Stat :
Media Medika Indonesiana Statistics