The Precision diagnosis by integration of transfer learning for colorectal cancer polyp detection and classification
Keywords:
Colorectal Cancer (CRC),Polyp Detection, Medical Imaging,YOLOv8, Real-time Object Detection, Transfer Learning, Deep Learning, Computer-Aided Diagnosis (CAD),Cancer Screening, Artificial Intelligence in Healthcare, Precision, Medicine Endoscopy, Image Classification ,Tumour PredictionAbstract
Colorectal cancer (CRC) remains one of the most fatal cancers worldwide, emphasizing the critical need for early detection. Polyps, small abnormal growths in the colon or rectum, are often the earliest indicators of CRC. However, manual polyp identification during colonoscopies is time-consuming, prone to human error, and may lead to delayed treatment. This study proposes an advanced AI-driven approach utilizing the YOLOv8 deep learning model for real-time polyp detection and classification. By integrating transfer learning, our model leverages pre-trained knowledge, improving accuracy even with limited medical datasets. The system achieved an impressive detection accuracy of 92.8%, with a precision of 92.63%, recall of 93.62%, and an F1-score of 93.12%, outperforming traditional methods. Additionally, our model assesses the likelihood of polyps developing into malignant tumors, providing valuable insights for early diagnosis and treatment planning. The integration of AI in colorectal cancer screening has the potential to revolutionize medical imaging, reducing diagnostic errors, enhancing efficiency, and ultimately saving lives.
References
1. A. Garrido, R. Sont, and M. Guardiola, "Automated Polyp Detection Utilizing Microwave Endoscopy for Early Diagnosis and Prevention of Colorectal Cancer: Phantom Validation," IEEE Access, vol. PP, pp. 1–1, Oct. 2021, doi: 10.1109/ACCESS.2021.3124019 .
2. M. Liu, J. Jiang, and Z. Wang, "Colonic Polyp Detection in Endoscopic Videos with Single Shot Detection Based Deep Convolutional Neural Network," IEEE Access, Jun. 2019, doi: 10.1109/ACCESS.2019.2921027.
3. Y.-T. Chen and N. Ahmad, "Colorectal Polyp Detection and Comparative Evaluation Based on Deep Learning Approaches," IEEE Access, Dec. 2023, doi: 10.1109/ACCESS.2023.3337031.
4. M. S. Hiossian, M. N. Syeed, M. F. Uddin, and M. Hasan, "DeepPoly: Deep Learning-Based Polyps Segmentation and Classification for Autonomous Colonoscopy Examination," IEEE Access, vol. 11, pp. 95889–95902, Aug. 2023, doi: 10.1109/ACCESS.2023.3310541.
5. K. Elkarazle, V. Raman, P. Then, and C. Chua, "Enhanced MA-Net and Modified Mix-ViT Transformer for
Improved Colorectal Polyp Segmentation," IEEE Access, vol. 11, Jul. 2023, doi: 10.1109/ACCESS.2023.3291783.
6. S. Pathan, Y. Somayaji, T. Ali, and M. Varsha, "ContourNet: An Automated Segmentation Framework for Colonic Polyp Detection," IEEE Access, vol. 12, May 2024, doi: 10.1109/ACCESS.2024.3392947.
7. Q.-X. Huang, G.-S. Lin, and H. M. Sun, "Classification of Polyps in Endoscopic Images Using Self-Supervised Structural Learning," IEEE Access, May 2023, doi: 10.1109/ACCESS.2023.3277029.
8. T. Yang, N. Liang, J. Li, Y. Yang, Y. Li, and X. He, "Intelligent Imaging Technology for Colorectal Cancer Diagnosis Using Deep Learning," IEEE Access, vol. 7, pp. 178839–178847, 2019, doi:
10.1109/ACCESS.2019.2958124.
9. S. A. Kareem, A. Q. M. Sabri, and S. B. Mohamad, "Feature Extraction from Gut Microbiome Data for Deep Neural Network-Based Colorectal Cancer Classification," IEEE Access, vol. 9, pp. 23565–23578, 2021, doi: 10.1109/ACCESS.2021.3050838.
10. Y. Li, F. Zhang, and C. Xing, "Pathogenic Gene Screening for Colorectal Cancer and the Role of Deep Learning in Diagnosis," IEEE Access, vol. 8, pp. 114916–114929, Jun. 2020, doi: 10.1109/ACCESS.2020.3003999.
