Color multiscale texture classification of hysteroscopy images of the endometrium.

The objective of this study was to investigate the diagnostic performance of a Computer Aided Diagnostic (CAD) system based on color multiscale texture analysis for the classification of hysteroscopy images of the endometrium, in support of the early detection of gynaecological cancer. A total of 416 Regions of Interest (ROIs) of the endometrium were extracted (208 normal and 208 abnormal) from 45 subjects. RGB images were gamma corrected and were converted to the YCrCb color system. The following texture features were extracted from the Y, Cr and Cb channels: (i) Statistical Features (SF), (ii) Spatial Gray Level Dependence Matrices (SGLDM), and (iii) Gray Level Difference Statistics (GLDS). The Probabilistic Neural Network (PNN), statistical learning and the Support Vector Machine (SVM) neural network classifiers were also applied for the investigation of classifying normal and abnormal ROIs in different scales. Results showed that the highest percentage of correct classification (%CC) score was 79% and was achieved for the SVM models trained with the SF and GLDS features for the 1x1 scale. This %CC was higher by only 2% when compared with the CAD system developed, based on the SF and GLDS feature sets computed from the Y channel only. Further increase in scale from 2x2 to 9x9, dropped the %CC in the region of 60% for the SF, SGLDM, and GLDS, feature sets, and their combinations. Concluding, a CAD system based on texture analysis and SVM models can be used to classify normal and abnormal endometrium tissue in difficult cases of gynaecological cancer. The proposed system has to be investigated with more cases before it is applied in clinical practise.