An integrated method for identifying liver tumors utilizing convolutional neural networks and residual networks

Abstract

The liver is located in the upper right quadrant of the belly. The liver performs essential tasks such as filtering blood, detoxifying chemicals, and metabolizing drugs and alcohol. Tumors develop as a result of an increase in cell proliferation. Metastatic liver cancers are the most common type. Hepatocellular carcinoma accounts for one million out of the total two million liver disorders that result in fatalities annually. The incidence and mortality rates of liver cancer are projected to increase by 55% by the year 2040. This would classify it as the third most prevalent cancer globally and one of the top five most severe cancers in 90 nations. A study conducted by the World Health Organization (WHO) was published in the Journal of Hepatology in October 2022. Medical imaging techniques can be used to identify the presence of these abnormalities, and a liver biopsy is performed to definitively establish the diagnosis. This paper presents a novel crossover technique that utilizes Convolutional Neural Networks (CNN) and Residual Networks (Resnet) to precisely identify liver cancers. This study presents a new approach for predicting lung cancer using a Convolutional Neural Network (CNN) that incorporates modification injection and deep learning-validated features. Picture analysis utilizes modern techniques such as deep learning and image processing. CNN-learned hierarchical features aid in the diagnosis of lung tumors by detecting and analyzing intricate patterns and textures. One of the model’s important features is the utilization of extensive image datasets to facilitate transfer learning on pre-trained models. The technique improves and eliminates noise from photographs of the skin. The skin ailment is classified using the Softmax Classifier after extracting visual features using a convolutional neural network (CNN). The device has the ability to rapidly categorize skin conditions with an accuracy rate over 95%.