This document presents a proposed method for classifying CT liver images to diagnose diseases using a probabilistic neural network (PNN) with a radial basis function kernel and kernel weighted fuzzy clustering. Existing methods have drawbacks like poor accuracy, inability to handle multiple images quickly, and less discriminatory power. The proposed method extracts Haralick texture features from normalized subband coefficient transformed images and inputs them into a PNN classifier trained on reference samples to classify images as normal, benign, or malignant. Kernel weighted fuzzy clustering is also used for segmentation. The method aims to accurately diagnose liver diseases from CT images in a short time through automated classification and segmentation.