Deep Learning-Based Diagnosis System for Liver Cancer Through Multimodal Determination for Early Treatment

Liver cancer is still one of the top causes of worldwide mortality from cancer, which highlights the demand for earlier diagnostic tools aimed at improved treatment outcomes. Although these traditional diagnostic practices are helpful, they may take time and also be subject to human error. The present study aims to fill the gap in existing diagnostic tools for liver cancer by evaluating prostate deep learning models (VGG 16, VGG 19 and ResNet 50) on CT scan images. This problem statement is related to improving the accuracy and reliability of liver cancer detection using powerful computational techniques. We processed a dataset of 1020 CT scan images by resizing, then normalizing the data, and performing an augmentation and some noise reduction. The performance of each model was evaluated in terms of precision, recall and F1 score using Dice Similarity Coefficient to identify the liver tumours. VGG 19 was the top performer in terms of accuracy and another score, followed by VGG 16 and ResNet 50. The use of such models in clinical settings can significantly improve diagnostic accuracy and, ultimately, diagnosis time, which will lead to better patient outcomes. The study ultimately underscores the promise of deep learning models (specifically VGG 19) as an essential resource in medical imaging by developing new and prospective methods for computerized liver cancer diagnosis that will help improve healthcare.