Computer-Aided Diagnosis in Hysteroscopic Imaging

Abstract

The paper presents the development of a computer-aided diagnostic (CAD) system for the early detection of endometrial cancer. The proposed CAD system supports reproducibility through texture feature standardization, standardized multifeature selection, and provides physicians with comparative distributions of the extracted texture features. The CAD system was validated using 516 regions of interest (ROIs) extracted from 52 subjects. The ROIs were equally distributed among normal and abnormal cases. To support reproducibility, the RGB images were first gamma corrected and then converted into HSV and YCrCb. From each channel of the gamma-corrected YCrCb, HSV, and RGB color systems, we extracted the following texture features: 1) statistical features (SFs), 2) spatial gray-level dependence matrices (SGLDM), and 3) gray-level difference statistics (GLDS). The texture features were then used as inputs with support vector machines (SVMs) and the probabilistic neural network (PNN) classifiers. After accounting for multiple comparisons, texture features extracted from abnormal ROIs were found to be significantly different than texture features extracted from normal ROIs. Compared to texture features extracted from normal ROIs, abnormal ROIs were characterized by lower image intensity, while variance, entropy, and contrast gave higher values. In terms of ROI classification, the best results were achieved by using SF and GLDS features with an SVM classifier. For this combination, the proposed CAD system achieved an 81% correct classification rate.

The effectiveness of automatic laparoscopic diagnostics of liver pathology using different methods of digital images classification

“…Texture features extracted from pathological endometrium were different from normal one, and were characterized by lower image intensity, while variance, entropy and contrast gave higher values [4]. Thus, for the hysteroscopic images the heightened mediana of grey scale is clearly identified and along with more homogenous and less contrast might be treated as informative differential index for normal state identification [4].…”

“…Контрольне тестування засвідчило, що найвищим показник повноти діагностики цирозу печінки був при використанні AdaBoost класифікатора, котрий було навчено за допомогою MCLBT-дескрипторів, що отримали з кадрів у HSV форматі -0,655, a також під час діагностики метастатичного ураження печінки -при використанні MCLBT-де-скрипторів, що одержали з кадрів у RGB форматі -0,925. Отже, КАД на основі AdaBoost класифікатора дає можливість Computer automatic diagnostic (CAD)/classification of videoimages is actual for minimal invasive abdominal surgery and endoscopy [1][2][3][4][5]. CAD systems are intensively developed for tracking laparoscopic instrumentation [6], identification of zones of pathology [4,7,8] and the diminution the risk of damage of healthy tissues [2].…”

“…Отже, КАД на основі AdaBoost класифікатора дає можливість Computer automatic diagnostic (CAD)/classification of videoimages is actual for minimal invasive abdominal surgery and endoscopy [1][2][3][4][5]. CAD systems are intensively developed for tracking laparoscopic instrumentation [6], identification of zones of pathology [4,7,8] and the diminution the risk of damage of healthy tissues [2]. Thus, A. Lahane et al [2] used series of computer vision algorithms to track surgical instruments and coarse location of the cystic artery.…”

“…The quick change of the angle of object along with changes of their illumination and specular reflections significantly impact on the result of automatic recognition [5]. Hence, the usage of gamma-correction in the course of preprocessing of primarily gathering information, which permits to identify relations between quantitative pixel characteristics and their actual brightness is justified [4].…”

“…Thus, for the hysteroscopic images the heightened mediana of grey scale is clearly identified and along with more homogenous and less contrast might be treated as informative differential index for normal state identification [4]. That is why we decided to use both color and texture features for training Haarfeatures based classifier.…”