Efficient Liver and Tumor identification from CT Images and Clinical Data Using a Hybrid Deep Learning and Machine Learning Model

The present research investigates the integration of machine learning and artificial intelligence models for precise liver and tumour segmentation from CT images, with the goal of increasing diagnostic accuracy in liver cancer. A full dataset size of 2000 patients’ records were used, with standardised dosage information, treatment, as well as CT scans. The comprehensive preprocessing of the data, including normalisation, noise reduction, contrast enhancement, and segmentation, provided an opportunity not only to improve the image quality but also to select the desired features. The tracking of the tumor responses to the treatment was completed with the help of such machine learning models as Support Vector Machines, deep learning models such as VGG 16, VGG 19, Inception V3. The SVM + Inception V3 hybrid model was chosen as the most appropriate one, showing 98.76% of accuracy in detecting the correct responses to liver cancer. The SVM ensured the proper temporal data handling, and the Inception V3 made use of multi-scale feature extraction capabilities to localise and classify cancerous regions within CT scans accurately. The VGG 19 achieved an accuracy of 94.5%, and VGG 16 achieved an accuracy of 91.2%. Both VGG 16 and VGG19 performed commendably because of their image analysis capabilities. SAR values further emphasise that SVM + Inception V3 achieves the max precision and recall rates, while VGG displays slight limitations in this regard. The implications for this study are that hybrid models can provide a practical implementation of machine learning/deep learning artificial intelligence applications for medical imaging.