This document summarizes a research study that used artificial neural networks (ANN) to classify ultrasound images of normal and abnormal kidneys. The study involved the following key steps:
1. Ultrasound images of normal and abnormal kidneys were collected and preprocessed, including cropping, rotation, edge detection and background removal.
2. Texture features were extracted from the preprocessed images using four different techniques: intensity histogram, gray-level co-occurrence matrix, gray-level run length matrix, and invariant moments. This resulted in a total of 4324 features.
3. Feature selection was performed to identify the most important features, resulting in a reduced feature set.
4. An ANN with a back