Diagnosing Prostate Cancer: An Implementation of Deep Machine Learning Fusion Network in MRI Using a Transfer Learning Approach

Abstract

Of all the terminal cancers that plague men, prostate cancer remains one of the most prevalent and ubiquitous. Data shows prostate cancer is the second leading cause of cancer death worldwide among men. About 11% of men have prostate cancer at some time during their lives. As it happens, we have dedicated our entire research to developing an approach that can improve the existing precision of prostate cancer diagnosis. In our research, we have dedicated a Transfer Learning approach for the Deep Learning model to compare the accuracy in results using Machine Learning classifiers. In addition, we evaluated individual performance in classifications with different evaluation measures using a Deep Learning pre-trained network, VGG16. During our evaluation, we assessed several performance metrics such as Precision, Recall, F1 Score, and Loss Vs. Accuracy for performance analysis. Upon implementing the Transfer Learning approach, we recorded the optimum performance using the VGG16 architecture compared to other popular Deep learning models such as MobileNet and ResNet. It is important to note that we have used the convolutional block and dense layers of VGG16 architecture to extract features from our image dataset. Afterward, we forwarded those features to Machine Learning classifiers to tabulate the final classification result. Upon successful tabulation, we have secured significant accuracy in prognostication using the Deep Machine Learning method in our research.