Hybrid Optimization Enabled Deep-Learning for Prostate Cancer Detection.

Prostate cancer (PCa) is now one of the most common types of cancer in gents and one of the most important causes of death in the United States. The best non-invasive imaging technique for diagnosing PCa is Magnetic Resonance Image. Each step in the conventional methods is optimized without allowing for error tolerance, resulting in significant computational costs. Hence, a Developed LHFGSO-DMN for PCa detection is introduced to reduce the proportion of PCa deaths. Modifying the objective function with dice coefficient and pixel-wise cross-entropy creates the optimized multi-objective semantic segmentation model known as multi-objective SegNet. This model is trained using the proposed Light Henry Firefly Gas Solubility Optimization (LHFGSO). The LHFGSO is the integration of a Henry Firefly Gas Solubility Optimization (HFGSO) and Light Spectrum Optimizer, and HFGSO is the formation of Henry Gas Solubility Optimization and Firefly Algorithm. Additionally, the LHFGSO method is used to train the Deep Maxout Network (DMN), which is used for cancer detection. The introduced multi-objective SegNet model with DMN for PCa growth recognition strategy reached superior execution measures with an accuracy of 94.63%, sensitivity of 93.46% and specificity of 95.72%. [ABSTRACT FROM AUTHOR]