Prostate cancer detection using modified transformer with optimal feature selection from MRI images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Prostate cancer is a leading malignancy in men, necessitating accurate early detection.
• Magnetic Resonance Imaging (MRI) is vital for prostate cancer diagnosis, but faces challenges like image noise and subtle lesion variations.
• Current detection methods struggle with precision, impacting patient outcomes.

Purpose of the Study:

• To introduce ProstateNet, a novel automated framework for enhanced prostate cancer detection using AI.
• To improve the accuracy and reliability of identifying cancerous regions in prostate MRI scans.
• To overcome limitations in current prostate cancer detection techniques.

Main Methods:

• MRI image preprocessing using a Bilateral Filter to reduce noise and preserve details.
• Feature extraction via a Residual Recurrent Model (RRM) for spatial dependencies.
• Feature refinement using the Iterative Secrecy Bird Optimization (ISBO) algorithm.
• Disease detection employing a Modified Transformer model with CapsuleNet for accurate classification.

Main Results:

• The proposed ProstateNet framework demonstrates enhanced precision in prostate cancer detection.
• The integration of Bilateral Filter, RRM, ISBO, and Modified Transformer improves robustness.
• The system accurately classifies cancerous and non-cancerous regions in MRI scans.

Conclusions:

• ProstateNet offers a significant advancement in automated prostate cancer detection.
• The framework shows promise for improving diagnostic accuracy and patient outcomes in prostate cancer management.
• This AI-driven approach addresses key challenges in interpreting prostate MRI data.