Texture analysis ? A new method of differentiating prostatic carcinoma from prostatic hypertrophy

Abstract

This study compares the results of a new computer assisted evaluation program of sonographic images of the prostate with the histological examination of the operative specimen. This computer program is based on the fact that the human eye is not capable of distinguishing picture distribution points which differ in higher than second order statistics. The sensitivity of the method was 89%, the specificity 86% and the diagnostic accuracy 88%. The results are discussed by comparing them to results of transrectal sonography and digital palpation as reported in the literature.

Texture Feature-Based Classification on Transrectal Ultrasound Image for Prostatic Cancer Detection

“…The sensitivity and specificity of this method are 80% and 88%, respectively, with good results. Kratzik et al [ 22 ] published a study on prostate testing. The study used texture feature analysis to obtain good results (specificity and sensitivity both exceed 80%) but did not specify how to evaluate.…”