The Classification of the Prostate Cancer based on Transfer Learning Techniques

Abstract

The most common cause of mortality worldwide and the most common male cancer is prostate cancer. According to the American Cancer Society. In the United States, there were 164,690 new instances of prostate cancer and at least 29,430 deaths from the disease in 2018, making up 9.5% of all new cancer cases. This will have a significant socioeconomic impact. Having the ability to determine the aggressiveness risk of confirmed prostate cancer could enhance the choice of proper treatment for individuals. This could lead to better outcomes, especially in terms of prostate cancer specific mortality. Deep learning-based significant prostate cancer classification has attracted a lot of attention because it may one day be used to support therapeutic decision-making. In this research we propose four models for classification the prostate cancer based on transfer learning algorithms (EfficentNet, DenseNet and Xception). We used two datasets for diagnosing prostate cancer. One of them is the standard dataset which consists of six grades of cancers and the other is a personal dataset from laboratory which is new dataset from treated patients at the hospital of Zagazig university. The results are obtained by using the standard dataset is approximately 93.6% accuracy using EfficientNetB7 pretrained model. The results of the created dataset are 97.08%, 97.98%, 98.87% by using EfiicientNet, DenseNet121 and Xception transfer learning models respectively. The results were compared with the state of arts, and it outperform all of them and also the model can be used in applications.

A systematic review of deep learning methods for the classification and segmentation of prostate cancer on magnetic resonance images

“…12 The integration of DL architectures such as CNNs and transfer learning algorithms like EfficientNet, DenseNet, and Xception has significantly improved classification accuracy rates, particularly in analyzing complex 3D MRI scans for detecting malignant lesions. 13 Contrastive learning techniques, including UKSSL (Underlying Knowledge-based Semi Supervised Learning), enable the extraction of valuable insights from limited labeled datasets, thereby enhancing the robustness and generalizability of classification models. 14 By harnessing the power of DL, these models not only enhance diagnosis accuracy but also facilitate personalized treatment planning and therapeutic decision-making in PCa management.…”

“…Various DL architectures, such as convolutional neural networks (CNNs), have been deployed to construct these classification models 12 . The integration of DL architectures such as CNNs and transfer learning algorithms like EfficientNet, DenseNet, and Xception has significantly improved classification accuracy rates, particularly in analyzing complex 3D MRI scans for detecting malignant lesions 13 . Contrastive learning techniques, including UKSSL (Underlying Knowledge‐based Semi Supervised Learning), enable the extraction of valuable insights from limited labeled datasets, thereby enhancing the robustness and generalizability of classification models 14 .…”