Early diagnosis of prostate cancer using image processing techniques

This paper gives an overview of the method of detecting prostate cancer in various stages by associating Region of interest segmentation method with artificial neural networks system. Prostate cancer is commonly prevalent carcinoma detected in most of the male population. A diagnosis of prostate cancer was complicated due to unclear symptoms and involves many procedures. One of these procedures involves the study of prostate tissue biopsy to find cancer affected region. However, no boundary specified region was considered for further studies. Recent developmental techniques in the medical imaging field, especially in neural networks, have paved the way for prostate carcinoma detection in different stages. The proposed method can be used for early stage detection of prostate carcinoma and its current stage. The MRI image of the prostate gland is preprocessed to reduce noise effects and Region of interest is obtained with the artificial neural network system and segmentation is done. The core idea of this paper is to assume that every region of prostate tissue could be related to malignant or unnatural tissues by making use of Radial Basis Function network method which uses the weighted mean algorithm along with Gaussian kernel for pattern recognition application. Experimental results of this method show improvement in terms of sensitivity, accuracy, specificity and reduce the number of false positive test results.