An Efficient Biomedical Solicitation in Liver Cancer Classification by Deep Learning Approach

Abstract

The liver, a significant physical organ, is located in the upper right and left abdominal cavities just below the diaphragm. It produces several different chemicals that the body needs to function properly. Researchers can benefit from converting images into data to more easily exchange and generate precise results. Since the process of conversion relies on technology and an algorithm, it eliminates the possibility of human error. Liver cancer has the highest fatality rate of any cancer since its symptoms don't present until late in the disease's progression, making early detection difficult. Skimming, sifting, segmenting, feature abstraction, and presentation via ANN (Artificial Neural Network) are the primary topics of the first stage of the debate. Real-time data sets use Feed-Forward Neural Network (FFNN) for identifying liver cancer and classifying its severity. Filtering has two primary applications: noise suppression and edge rounding. Then, segmentation is employed to isolate the relevant area, allowing for more compact data storage. The Gray Level and Co-occurrence Matrix (GLCM) is used to extract features, and the resulting matrix can have many different forms. This criterion helps classify tumors as benign or malignant. Metrics such as accuracy, sensitivity, positive and negative predictive values, and precision are used to assess the rate. The experimental method for identifying liver tumors uses CT liver pictures to achieve an average accuracy of 99.45% for malignant images.