Liver diseases such as fatty liver, cysts, and tumors require early and accurate diagnosis to improve patient outcomes. Ultrasound imaging is widely used due to its non-invasive and cost-effective nature; however, its heterogeneous characteristics, including speckle noise, low contrast, and variability across devices, make diagnosis challenging. This paper proposes a deep learning-based approach for the classification of liver diseases using heterogeneous ultrasound images. The system employs pre-processing techniques such as noise reduction, normalization, and contrast enhancement to improve image quality. A YOLO-based architecture integrated with convolutional neural networks is used for feature extraction and simultaneous detection and classification of liver abnormalities. Experimental results show that the proposed model achieves improved accuracy and robustness compared to conventional methods. The system supports real-time analysis and can assist clinicians in reliable and efficient liver disease diagnosis.