Diagnosis of Liver Tumor from CT Scan Images using Deep Segmentation Network with CMBOA based CNN

Accurate segmentation and categorization of liver tumours are crucial for the diagnosis and management of carcinoma or metastases. The liver tumour presents a challenging problem for precise and automated tumour segmentation and classification because of its blurry boundaries and large variety of potential forms, sizes, and placements. New AI models have emerged as computer technology has progressed. The NLP community has had such success with the transformer paradigm that the CV community has adopted it as well. Although established methods exist for categorising the liver, especially in clinical settings, they might be refined to be more accurate. Two deep learning-based models are used to do the segmentation and classification of the liver tumour in this study. As a first step, the input photos are median filtered and their histograms are equalised to prepare them for further processing. Then, Deep Segmentation Network (DSegNet) extracts the liver from the input pictures. The tumour is then categorised using an Optimised Convolutional Neural Network (OCNN) model, with the CNN's weight chosen using a Cat and Mouse Based Optimisation Algorithm (CMBOA). Two openly accessible datasets are used for the experimental study, with a focus on certain key metrics. The results demonstration that associated to pre-existing deep learning representations, the suggested model is about 98% more accurate in its classifications.