Abstract: This study presents an optimized deep learning model designed to enhance the accuracy of liver disease detection using ultrasound imaging. Ultrasound images often suffer from noise, low contrast, and operator variability, creating challenges in clinical interpretation. To address these issues, an integrated approach combining targeted preprocessing, a lightweight CNN architecture, and balanced augmentation strategies was developed. The model demonstrates improved diagnostic consistency, effectively distinguishing normal and abnormal liver tissue patterns. This case study highlights the model’s performance, practical significance, and potential to serve as a supportive diagnostic tool in healthcare environments.

Keywords: Liver Disease Detection, Deep learning, Ultrasound imaging, Noise reduction, Medical Image Analysis

Download: | DOI: 10.17148/IMRJR.2026.030108