Deep Learning Approach for Predicting Prostate Cancer from MRI Images
Keywords:Deep learning, convolutional neural network, classification, prostate cancer, magnetic resonance imaging
According to medical data, prostate cancer has been one of the most lethal malignancies in recent years. Early detection of prostate cancer significantly influences the tumor's treatability. Image analysis software that operates using a machine learning or deep learning algorithm is one of the techniques utilized to aid in the early and rapid identification of prostate cancer. This paper evaluates the performance of three deep learning Convolutional neural network (CNN) algorithms in detecting prostate cancer. Using Python, three deep learning models, ResNet50, InceptionV3, and VGG16, are subsequently created on the Kaggle platform. These three models have been applied to various medical image diagnostic problems and have won several contests. This study used 620 image samples from the Cancer Imaging Archive (TCIA) data source. Accuracy, f1 score, recall, and precision are used to evaluate the performance of the three models. The extracted test results indicate that the VGG16 achieves the highest level of accuracy at 95.56 percent, followed by the ResNet50 at 86.67 percent and the InceptionV3 at 85.56 percent.