This document summarizes a research paper on computer-based automatic detection and classification of liver tumors using multilevel wavelet transformation and neural networks. The paper presents an algorithm that segments MRI images using k-means clustering to detect liver tumors at early stages. Feature extraction is performed on the images and a probabilistic neural network is used to classify tumors as benign, malignant, or normal. Experimental results showed clustering-based segmentation was more accurate than thresholding methods. The algorithm was able to automatically detect and analyze liver tumors in MRI/CT images to help clinicians.