Extensive Analysis of Deep Learning Algorithms for Automatic Detection of Liver Anomalies Using MRI and CT Scanned Images

The liver's segmentation during medical imaging is crucial for the diagnosis and prognosis of liver diseases., including fibroid cysts, malignant growths, and other pathological problems. The World Health Organization says that alcohol addiction, non-alcoholic fatty liver disease, chronic hepatitis infections, and genetic predispositions are among the risk factors that make liver cancer a primary cause of cancer-related death. Radiologists' traditional manual assessment of CT and MRI images takes a lot of time and is prone to mistakes. Medical picture segmentation has undergone a revolution because of developments in deep learning, namely convolutional neural networks (CNNs)., which automate feature extraction and increase accuracy. This comprehensive analysis investigates many deep learning techniques, such as CNNs, U-Net, autoencoders, GANs, and transformers, that are used for liver segmentation. The above-mentioned models have revealed a significant improvement in efficiency and accuracy in segmentation. These models” evaluation and assessment have been eased with datasets such as LiTS, CHAOS, MSD, SLIVER07, and many others. Some of the performance indices that are used to measure the effectiveness of these mathematical models. Despite this, there are still problems that persist such as the processing requirements, morphological variation, and dependency on labeled data. To improve model clinical relevance in the future, more work should be done to make them less sensitive to noise and to increase model interpretability. This analysis raises questions for further development to overcome the current limitations and reveal the potential of deep learning to transform the identification of liver abnormalities.