Pufferfish Optimizer based Enhanced UNETR for Accurate Lung Cancer Segmentation and Classification

According to recent data released by the WHO, lung tumours are the top cause of death worldwide. One of the promising areas for effective lung cancer detection is tumour segmentation. To determine the location of tumours and the extent of malignancy, doctors use computerised tomography (CT) scans. Radiation oncologists face a daunting and timeconsuming challenge with medical imaging segmentation that requires handcrafting. This article proposes a comprehensive approach for early detection of lung cancer using CT scan imaging as a solution to this challenge. In order to improve patients’ chances of survival, this study presents an automated diagnostic algorithm. Part one of the suggested lung cancer diagnostic system is a segmentation constituent built on the UNEt TRansformers (UNETR) network, and part two is a classification component that uses the output segmentation component to label the output as benign or malignant, built on the self-supervised network. Pufferfish Optimisation Algorithm (POA) identifies the best values for UNETR’s trainable parameters. The suggested practise offers a potent resource for the early detection and treatment of lung cancer by utilising data from 3D-input CT scans. The three-stage suggested lung cancer classification model makes use of a modified U-Net for detection. To improve segmentation and classification outcomes, a large amount of trials have been conducted.