Deep Learning-Based Semantic Segmentation Models for Prostate Gland Segmentation

Prostate cancer (PCa) is one of the prevalent forms disease found in males due to unusual development cells. Early diagnosis this PCa can be useful terms treatment and medication. Segmentation classification through manual observation are methods, but it highly challenging complex boundaries features. Machine learning-based semantic segmentation architecture models consume more energy processing time will lead reduced scalability reliability. In order tackle these limitations, deep used as has advantages discriminating features lesions efficiently accurately. The main aim work segment accurately efficiently. Hence, model-based architectures such U-Net, Linknet, PSPNet proposed research. These equipped with a backbone Inception-ResNet-v2 CNN for prostate gland segmentation. Nearest neighbour interpolation normalization methods employed preprocessing technique enhancing MRI images. normalized image was taken various settings LinkNet PSP-Net performing optimizing models, Adam, Adamax Nadam optimizers used. experiment performed using NCI-ISBI 2013 dataset. Performance analysis evaluated Intersection Union (IoU) scores, where optimized obtained best IoU score 0.763337802.