Related Experiment Video
Updated: Jun 17, 2026
10:26
A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
Published on: May 19, 2023
Detection and Classification of Pancreatic Cancer Nodules on CT Images using U-Net and Ensemble Models
V M Manesh1,2, M Subramoniam1, S Poornapushpakala1
1School of Electrical and Electronics Engineering, Sathyabama Institute of Science and Technology, Chennai, Tamil Nadu, India.
Current Medical Imaging
|June 16, 2026
Summary
This study introduces an AI framework for early pancreatic cancer detection using CT scans. The model accurately identifies small tumors, improving diagnostic capabilities for better patient outcomes.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Pancreatic Adenocarcinoma (PDAC) is a leading cause of cancer mortality, with early detection hindered by current imaging techniques.
- Deep learning (DL) offers potential for improved segmentation and early detection of pancreatic tumors.
Purpose of the Study:
- To develop and validate an AI-based framework for early detection of PDAC using CT images.
- To enhance the accuracy and reliability of pancreatic cancer diagnosis, particularly for small tumors.
Main Methods:
- Utilized a Boosted Adaptive Diffusion Filter (BADF) for noise reduction in CT images.
- Employed a modified U-Net model with Adam optimizer for precise tumor segmentation.
- Implemented an ensemble learning approach for robust classification of PDAC.
Main Results:
- Achieved high performance metrics: 98.7% accuracy, 98.7% precision, 97.92% specificity, and 99.63% AUC.
- Demonstrated superior performance in detecting small pancreatic tumors (<2 cm) compared to existing DL methods.
Conclusions:
- The integrated framework (BADF, U-Net, ensemble learning) significantly improves robustness and detection accuracy for PDAC.
- This AI approach shows substantial potential for early PDAC detection, aiding clinical decisions and improving patient care.
