Mining balance disorders' data for the development of diagnostic decision support systems.

In this work we present the methodology for the development of the EMBalance diagnostic Decision Support System (DSS) for balance disorders. Medical data from patients with balance disorders have been analysed using data mining techniques for the development of the diagnostic DSS. The proposed methodology uses various data, ranging from demographic characteristics to clinical examination, auditory and vestibular tests, in order to provide an accurate diagnosis. The system aims to provide decision support for general practitioners (GPs) and experts in the diagnosis of balance disorders as well as to provide recommendations for the appropriate information and data to be requested at each step of the diagnostic process. Detailed results are provided for the diagnosis of 12 balance disorders, both for GPs and experts. Overall, the reported accuracy ranges from 59.3 to 89.8% for GPs and from 74.3 to 92.1% for experts.

Texture-based classification of hysteroscopy images of the endometrium.

The objective of this study was to classify hysteroscopy images of the endometrium based on texture analysis for the early detection of gynaecological cancer. A total of 418 regions of interest (ROIs) were extracted (209 normal and 209 abnormal) from 40 subjects. Images were gamma corrected and were converted to gray scale. The following texture features were extracted: (i) statistical features, (ii) spatial gray level dependence matrices (SGLDM), and (iii) gray level difference statistics (GLDS). The PNN and SVM neural network classifiers were also investigated for classifying normal and abnormal ROIs. Results show that there is significant difference (using Wilcoxon rank sum test at a=0.05) between the texture features of normal and abnormal ROIs for both the gamma corrected and uncorrected images. Abnormal ROIs had lower gray scale median and homogeneity values, and higher entropy and contrast values when compared to the normal ROIs. The highest percentage of correct classifications score was 77% and was achieved for the SVM models trained with the SF and GLDS features. Concluding, texture features provide useful information differentiating between normal and abnormal ROIs of the endometrium.

Texture analysis of the endometrium during hysteroscopy: preliminary results.

The objective of this study is to investigate the usefulness of texture analysis in the endometrium during hysteroscopy in endoscopic imaging of the uterine cavity. Endoscopy images from the endometrium from three subjects, at optimum illumination and focus, were frozen and digitized at 720x576 pixels using 24 bits color. Regions of interest (ROI) of normal (N=61) and abnormal (N=69) regions were manually selected by the physician. ROI images were converted into gray scale and statistical features (SF) and spatial gray level dependence matrix features (SGLDM) were computed. The nonparametric Wilcoxon rank sum test at a=0.05 was carried out for comparing the differences between normal and abnormal tissue. There was significant difference between normal and abnormal endometrium for the SF features variance, energy and entropy and for the SGLDM feature of angular second moment. There was no significant difference for the SF features mean, median, and SGLDM features of contrast, correlation and homogeneity.