Pancreatic tumor segmentation improved by lightweight TRIUNE-Net model
TRIUNE-Net: Harmonizing Scale, Shape, and Efficiency in Pancreatic Tumor Segmentation
Computer Vision and Pattern Recognition
Summary
Pancreatic tumors are tricky to spot and outline in 3D CT scans because both the pancreas and tumors vary greatly in size and shape. The authors created TRIUNE-Net, a new computer model that adapts to these wide ranges by looking at different scales and focusing on irregular tumor shapes. It also keeps all spatial details during image processing stages to avoid missing tiny tumors. Tests show TRIUNE-Net is better and more efficient than previous methods, helping reduce false positives and missed tumors without needing extra data to learn from.
What this means in practice
- •For medical imaging teams: Improve tumor detection accuracy and efficiency in pancreatic CT scan analysis within clinical workflows using TRIUNE-Net.
- •For radiology software developers: Integrate TRIUNE-Net for enhanced segmentation modules in radiology tools targeting better pancreatic tumor delineation.$Commercial implications: Enables creation of advanced imaging software with improved tumor detection, appealing to hospitals and diagnostic centers.
Authors
Amir Hossein Saleknia, Alireza Kheyrkhah, Sanaz Karimijafarbigloo, Reza Azad, Sina Houshmand, Ulas Bagci, Dorit Merhof, Alaa Sulaiman
Abstract
Pancreatic tumor segmentation in 3D CT volumes is challenged by extreme scale variability across both the pancreas and tumor, and highly irregular tumor morphology. While recent advances have pushed segmentation performance, existing methods do not explicitly address these challenges and come at the cost of excessive computational complexity, limiting their practicality in resource-constrained clinical environments. We propose TRIUNE-Net, a lightweight unified architecture that harmonizes scale, shape, and efficiency through three synergistic innovations. A multi-scale context aggregation module with stage-adaptive dilated convolutions enables the model to reason across the broad range of anatomical scales present in both organs. A serial linear-deformable attention mechanism combines large effective receptive fields with shapeadaptive deformable convolutions to capture irregular, non-convex tumor morphologies. Finally, an information-preserving downsampling module replaces conventional max pooling entirely, retaining all spatial information while adding negligible parameters, preventing small tumors from being discarded before they can be recognized. On both the MSD Pancreas and NVD Pancreas datasets, TRIUNE-Net achieves state-of-theart results with only 5.86 M parameters and no external pre-training, outperforming all baselines across all key tumor metrics. Specifically, it surpasses the next-best model by 0.45% in tumor Dice, 6.0 points in F1 score, 6.6 points in sensitivity, and 3.4 points in precision, simultaneously reflecting its ability to suppress both missed tumors and false alarms in clinically realistic conditions. Our code is available at: https://github.com/abdora-ai/TRIUNE-Net