Colorectal cancer detection with enhanced precision using a hybrid supervised and unsupervised learning approach

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: Jan. 25, 2025

The current work introduces the hybrid ensemble framework for detection and segmentation of colorectal cancer. This will incorporate both supervised classification unsupervised clustering methods to present more understandable accurate diagnostic results. method entails several steps with CNN models: ADa-22 AD-22, transformer networks, an SVM classifier, all inbuilt. CVC ClinicDB dataset supports this process, containing 1650 colonoscopy images classified as polyps or non-polyps. best performance in ensembles was done by AD-22 + Transformer model, AUC 0.99, a training accuracy 99.50%, testing 99.00%. group also saw high 97.50% Polyps 99.30% Non-Polyps, together recall 97.80% 98.90% hence performing very well identifying cancerous healthy regions. proposed here uses K-means combination visualisation bounding boxes, thereby improving yielding silhouette score 0.73 cluster configuration. It discusses how combine feature interpretation challenges into medical imaging localization precise malignant A good balance between generalization shall be hyperparameter optimization-heavy learning rates; dropout rates overfitting suppressed effectively. schema treats deficiencies previous approaches, such incorporating CNN-based effective extraction, networks developing attention mechanisms, finally fine decision boundary support vector machine. Further, we refine process via purpose enhancing procedure. Such holistic framework, hence, further boosts results generating outcomes rigorous benchmarking detecting cancer higher reality towards clinical application feasibility.

Language: Английский

---

Automated Detection of Colorectal Polyp Utilizing Deep Learning Methods With Explainable AI

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 78074 - 78100

Published: Jan. 1, 2024

Detecting colorectal polyps promptly and accurately is crucial in preventing the progression of cancer. These cause severe conditions colon or rectum, presenting a significant diagnostic challenge. Traditional manual detection through medical imaging not only bulky prone to errors but also incurs substantial costs, requiring expert endoscopist. Inefficient treatment can lead critical health complications. Addressing these issues, we extensively employed various configurations state-of-the-art YOLOv8 (n-nano, s-small, m-medium, l-large, x-extra-large) models for effective polyp localization. Complementing this, proposed novel TR-SE-Net model segmentation, integrating Squeeze-and-Excite Networks (SE-Net) elevate performance real-time processing capabilities. The Kvasir-SEG dataset utilized training testing models, supplemented by external validation CVC-ClinicDB, PolypGen, ETIS-LaribPolypDB, EDD 2020, BKAI-IGH confirm their efficacy unseen, data. This study delves into interpretability using explainable AI (XAI), such as eigen visualization localization heatmap analysis segmentation. exploration provides deeper insights decision-making processes thereby enhancing reliability. Notably, YOLOv8m showcased remarkable prediction speed (approximately 16.61 ms) excelled precision (0.946), recall (0.771), F1-score (0.85), mAP ₅₀ (0.886), xmlns:xlink="http://www.w3.org/1999/xlink">50-95 (0.695), catering diverse clinical scenarios. demonstrated improvements segmentation performances, including DSC (0.8754), F2-score (0.8786), (0.9027), (0.8879), accuracy (0.9647), competitive mIoU (0.7961), FPS (54), parameters (27.27 million), flops (10.59 GMac). Furthermore, A graphical Computer Aided Diagnosis (CAD) system developed utilizing both substantially reduce miss rate because will assist vice versa if fails. Conclusively, advanced computer-aided methods enhances colonoscopy procedures mitigating risks

Language: Английский

---

Rethinking encoder-decoder architecture using vision transformer for colorectal polyp and surgical instruments segmentation

Engineering Applications of Artificial Intelligence, Journal Year: 2024, Volume and Issue: 136, P. 108962 - 108962

Published: July 26, 2024

Language: Английский

---

Applications of Machine Learning in Cancer Imaging: A Review of Diagnostic Methods for Six Major Cancer Types

Electronics, Journal Year: 2024, Volume and Issue: 13(23), P. 4697 - 4697

Published: Nov. 27, 2024

Machine learning (ML) methods have revolutionized cancer analysis by enhancing the accuracy of diagnosis, prognosis, and treatment strategies. This paper presents an extensive study on applications machine in analysis, with a focus three primary areas: comparative medical imaging techniques (including X-rays, mammography, ultrasound, CT, MRI, PET), various AI ML (such as deep learning, transfer ensemble learning), challenges limitations associated utilizing analysis. The highlights potential to improve early detection patient outcomes while also addressing technical practical that must be overcome for its effective clinical integration. Finally, discusses future directions opportunities advancing research.

Language: Английский

---

Colorectal Polyp Detection and Comparative Evaluation Based on Deep Learning Approaches

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 135074 - 135089

Published: Jan. 1, 2023

Colorectal cancer has been one of the leading causes mortality over past decade, and colorectal polyps are cause this disease. Conventional polyp detection techniques insufficient for proper detection; thus, an efficient method is indispensable. In study, we collected images from a hospital in Taiwan, annotated ground truth locations, integrated them with public dataset to create colonoscopy dataset. Data augmentation further used increase training dataset’s diversity improve models’ performance. By developing comparison system based on recent state-of-the-art methods (i.e., FasterRCNN, SSD, YOLOv3, YOLOv4), compared measurement metrics statistically analyzed performance models identify significant statistical difference Moreover, developed error handling mechanism each model discard false null predictions. Finally, our selects proposes best performing deep learning detect classify polyps. We expect that proposed will accurately locate different types Eventually, approach ensure valuable medical aid model.

Language: Английский

---

HemoSet: The First Blood Segmentation Dataset for Automation of Hemostasis Management

Published: June 3, 2024

Language: Английский

---