Machine Learning Algorithm for Abnormality Detection and Classification of Kidney Stones in Ultrasound Images

For feature extraction and classification, a novel Discrete Tchebichef Moment-based Machine Learning (DTM) technique is suggested. The echo patterns seen inside the kidney and its sub-regions are used by experts to make diagnoses. Hyperechoic, hypoechoic, isoechoic, and anechoic echos are classified as hyperechoic, hypoechoic, isoechoic, and anechoic, respectively, and need a high level of knowledge to understand. Various renal disorders with closely connected echo patterns necessitate the system's quantitative quantification of echoes. When considering the entire complete kidney area for feature extraction, this fixed ROI strategy may eliminate the size uncertainty. The classification was carried out with the help of a multiclass support vector machine classifier with fivefold cross validation. The suggested DTM is evaluated using quantitative criteria including accuracy, precision, recall, and F -Score. The suggested work's promising results show the potential for using CAD in ultrasound images to classify renal disorders.