Uji Diagnostik Pemeriksaan Osteosklerotik Tulang dengan Sistem Radiografi Digital

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


Abstract


ABSTRACT
Diagnostic test of bone osteosclerotics examination using digital radiography system

Background: The interpretation of a conventional röntgen images have a high degree of subjectivity due to the limitation of humansight. The computationally detection help establish the accuracy of diagnosis of the radiologist. According to our knowledge, there has not been previous research on this issue in Indonesia. The research was aimed to examine a Matlab based software to determine the diagnostic value in the diagnosis of osteosclerotic bone.

Methods: This study was a diagnostic test which was conducted in Radiology Department of Dr. Kariadi Hospital Semarang, Dr.
Sardjito Hospital Yogyakarta and Bethesda Hospital Yogyakarta, April to August 2009. The radiographs of bone osteosclerotic resulted from computed radiography (CR) test results were taken consecutively, interpreted by a radiologist which was supported by the Anatomical Pathology Laboratory examination as a gold standard. Afterwards these steps were done respectively: classify patients as a normal or osteosclerotic patients based on the cut off point that was determined, calculate the value of the diagnostic by analysis of 2x2 tables, determine the area under the curve (AUC) by the procedure of receiver operating characteristic (ROC), and
determine the optimal of COP (cut off point) using ROC procedure.

Results: From the results of diagnostic tests of bone radiographs these following parameter values was obtained: AUC value of 97.6% (95% CI: 94.4%-100%), the optimal cut off point for bone oseosclerotic COP ≥1.05 with a sensitivity value of 93.0% and a specificity of 89.1%. Suitability kappa value of 0.818 K (95% CI: 0.757 to 0.879).

Conclusion: The radiographic examination of the results of CR using Matlab-based software can be used to diagnose bone
osteosclerotic with high sensitivity and specificity.

Keywords: Bone osteosclerotic, röntgen images, optimal of COP, Matlab software

ABSTRAK
Latar belakang: Pembacaan foto röntgen secara konvensional memiliki tingkat subyektivitas tinggi karena keterbatasan indra
penglihatan manusia. Pendeteksian secara terkomputasi membantu menegakkan diagnosis para radiolog. Sebagaimana diketahui, belum ada penelitian sebelumnya mengenai masalah ini di Indonesia. Penelitian ini bertujuan menguji perangkat lunak berbasis Matlab untuk menentukan nilai diagnostik pada diagnosis tulang osteosklerotik.

Metode: Penelitian ini adalah uji diagnostik yang dilakukan di Bagian Radiologi RSUP Dr. Kariadi Semarang, RSUP Dr. Sardjito
Yogyakarta dan Rumah Sakit Bethesda Yogyakarta, April-Agustus 2009. Radiograf osteosklerotik tulang hasil pemeriksaan CR (computed radiography) diambil untuk sampel secara konsekutif, kemudian diperiksa oleh radiolog yang didukung oleh pemeriksaan Laboratorium Patologi Anatomi sebagai baku emas. Tahapan yang dilakukan berturut-turut adalah: mengelompokkan pasien sebagai osteosklerotik dan normal berdasar titik potong yang ditentukan, menghitung nilai diagnostik dengan tabel analisis 2x2, menentukan AUC (area under the curve) dengan prosedur ROC (receiver operating characteristic), dan menentukan COP (cut off point) optimal dengan prosedur ROC.

Hasil: Hasil uji diagnostik radiograf tulang diperoleh nilai-nilai parameter sebagai berikut: nilai AUC adalah sebesar 97,6% (IK 95%: 94,4%-100%), titik potong optimal untuk osteosklerotik tulang COP ≥1,05 dengan nilai sensitivitas sebesar 93,0% dan spesifisitas sebesar 89,1%. Nilai kesesuaian kappa K sebesar 0,818 (IK 95%: 0,757-0,879).

Simpulan: Pemeriksaan radiografi hasil CR menggunakan
perangkat lunak berbasis Matlab dapat digunakan untuk
mendiagnosis osteosklerotik tulang dengan sensitivitas dan
spesifisitas tinggi


Keywords


Bone osteosclerotic, röntgen images, optimal of COP, Matlab software

Full Text:

PDF

Refbacks

  • There are currently no refbacks.


Visitor Stat :
Media Medika Indonesiana Statistics