A deep learning framework for efficient prostate tumor segmentation using deep learning model powered by wombat optimization

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Medical image segmentation plays a vital role in the early and precise detection of prostate cancer, which continues to be a major cause of morbidity among men. The research proposes a strong framework for prostate cancer segmentation that achieves high precision by combining deep learning and optimization methods. In order to improve data diversity and quality, the system first gathers and prepares prostate MRI datasets. Next, it performs sophisticated preprocessing operations as rotation, reshaping, Contrast Limited Adaptive Histogram Equalization (CLAHE), and image flipping. For efficient learning and assessment, the pre-processed data is divided into training (80%) and testing (20%) sets. A new deep learning model designated as SwinDense-GAM-UNet + + is used for segmentation. It combines the advantages of Swin Transformer module for capturing long-range dependencies and enhance contextual understanding. Gradient Attention Modules (GAM) for realistic segmentation learning, dense connections for deeper gradient flow, and UNet + + for hierarchical feature extraction. Furthermore, to enhance training efficiency and optimize model hyperparameters, the Wombat Optimization Algorithm (WOA) is utilized. Python software is used to implement the proposed system for image processing, model training, and evaluation. The model’s performance is evaluated using common measures like as recall, accuracy, and precision. The proposed structure offers a potential alternative for clinical diagnostic support, as evidenced by the results, which show that it performs noticeably better than current models in prostate cancer segmentation tasks.