Development of an Ultrasound Image Extraction Method for Detection and Classification of Kidney Abnormalities Using a Convolutional Neural Network

The Kidney is an important organ that filters waste, toxins, and excess fluids from the blood and removes the waste in the urine. Kidney disease is a condition that occurs when it becomes damaged or impaired. In this study, a model was proposed using a convolutional neural network (CNN) and ML technology, which were trained on a dataset of 306 kidney ultrasound images. This development included Canny extraction methods, grey level co-occurrence matrix (GLCM), and principal component analysis (PCA) to obtain kidney abnormalities detection features. Additionally, the new method created was called Canny Grey Principal Component Analysis Pattern (CGPCAP). This CGPCAP method was developed to achieve better results in feature extraction and classification multi detection kidney used scale invariant feature transform (SIFT) to detect extreme, low-contrast, and edge areas in kidney images. Following this discussion, CGPCAP was tested on an image retrieved from the database, with a set of 16 features extracted. CGPCAP achieved high classification accuracy through the experimental use of a CNN classifier. Other objectives included performing feature extraction and classification between normal kidneys and kidney abnormalities. In this study, multi detection kidney was used to detect extreme, low-contrast, and edge areas in kidney images. CNN was used to classify kidney images based on the feature extraction results. Relating to this discussion, the results of the proposed method were compared with automatic feature extraction using Canny, GLCM, SIFT, and PCA. The extracted features were inputted to the CNN classifier, which achieved the highest accuracy of 97.5% compared to other abnormality detection methods. This suggests that the CGPCAP algorithm not only improves the model’s ability to make a more accurate prediction but is also more efficient at handling more complex data. Overall, the use of CGPCAP with the CNN test algorithm provides better results in the context of renal abnormality detection.