Liver Tumor Detection and Classification Using GWO-ELM Model

Multimodal intelligence-based systems for medical analytics and decision-making are crucial in the healthcare industry. One of the most common types of cancer is liver cancer, and early detection is essential for successful treatment. The severity of irregular tumor forms varies depending on the malignancy stage and the tumor type. Identifying the liver and subsequent tumor segmentation are the two primary stages of tumor segmentation in the liver. In addition to detecting cancers from publically available data of liver scans, this research offers a novel deep learning-based segmentation with a grey wolf Optimization-Extreme Learning Model approach that exhibits excellent efficiency in results. To improve the efficacy of the liver tumor detection system, this work applies the GWO-ELM classifier and Haar wavelet transform. It uses one of the most widely used feature extractions. The GWO-ELM acts like a Support Vector Machine with a Neural Network structure and can solve multi and binary classification problems. In contrast, the Haar wavelet transform can extract the most pertinent features with low dimensionality. As a result, the GWO-ELM classifier and Haar wavelet transform characteristics are used to provide a useful method for classifying and extracting features from liver tumors. According to the results, the proposed GWO-ELM model performed very well, achieving an accuracy of 99.41 % for a multi-class dataset. This reveals that the GWO-ELM and Haar wavelet transform is a robust classifier for identifying liver tumors and might be used to handle various types of image data.