Kidney Abnormality Detection and Segmentation

As kidney disease is one of the deadliest cancers at present times, its detection and segmentation is one of the most important operations. The kidney disease can be identified by various methods which include CT scan, MRI and ultrasound. Among these techniques, US imaging in mostly preferred because of its low cost and non-invasive. However, the ultrasound images have low contrast and mainly contain speckle noise which creates a challenging task in kidney abnormalities detection. In this paper, an effective approach is developed that can eliminate the speckle noise from images and increase the accuracy of the system. For this, different images are taken from the available dataset or can be captured form the real world through camera. The images are pre-processed by using the Kuwahara filter so that any noise present in the image can be removed and enhanced image is obtained. In addition to this, the kuwahara filter preserves the edges of the image. Once the refined images are obtained, feature extraction process starts in which seven GLCM features i.e. contrast, correlation, energy, homogeneity, pixel value, min pixels value and max pixel value are extracted. The implementation of the CFA algorithm in the model reduces the overall complexity of the model by selecting only those features which are important and contain necessary information for segmenting and classification process. The selected features are then given to the PNN network for training and testing. The performance of the proposed model is evaluated in the MATLAB software. The simulation outcomes obtained proved that the proposed model is more precise, efficient and effective in identifying various kidney diseases.