• Deep intelligent transformer for pancreas segmentation and abnormality classification

    Banavathu Sridevi & B. John Jaidhan 

    Abstract

    The task of identifying tissues and classifying disease types from Pancreas MRI images can pose a significant challenge. Traditional models have struggled to accurately analyse abnormalities in tissue density, leading to classification inaccuracies in numerous cases. To address this issue, a novel Vanilla Termite Transformer Neural System (VTTNS) was introduced in this research study. The system was initially loaded with the MRI image data of the Pancreas. Preprocessing is done to eliminate noisy characteristics. Furthermore, the termite optimisation best solution function is utilized to extract disease features. Based on the selected features, the pancreatic diseases are predicted, and the affected region is segmented. The classification layer processes the relevant information to classify the Pancreas MRI image. The Pancreas MRI images are used to evaluate the suggested VTTNS. Lastly, the performance is evaluated using measures such as precision, F-measure, recall, accuracy, error rate, Jaccard score, and Dice score. With comparatively small errors, the model achieved a high accuracy of 99.80%.