본문으로 건너뛰기
← 뒤로

Explainable Lightweight Model Using Low-Rank and Convolutional Block Attention for Pancreatic Cancer Diagnosis.

The international journal of medical robotics + computer assisted surgery : MRCAS 2026 Vol.22(2) p. e70164

Tanwar V, Sharma B, Yadav DP, Liatsis P

📝 환자 설명용 한 줄

[BACKGROUND] Early and accurate pancreatic cancer (PC) detection remains a major clinical challenge.

이 논문을 인용하기

BibTeX ↓ RIS ↓
APA Tanwar V, Sharma B, et al. (2026). Explainable Lightweight Model Using Low-Rank and Convolutional Block Attention for Pancreatic Cancer Diagnosis.. The international journal of medical robotics + computer assisted surgery : MRCAS, 22(2), e70164. https://doi.org/10.1002/rcs.70164
MLA Tanwar V, et al.. "Explainable Lightweight Model Using Low-Rank and Convolutional Block Attention for Pancreatic Cancer Diagnosis.." The international journal of medical robotics + computer assisted surgery : MRCAS, vol. 22, no. 2, 2026, pp. e70164.
PMID 41999341
DOI 10.1002/rcs.70164

Abstract

[BACKGROUND] Early and accurate pancreatic cancer (PC) detection remains a major clinical challenge.

[METHODS] We introduce a novel hybrid deep learning framework for automated classification of CT images, which requires fewer computational resources while achieving high diagnostic performance. We integrated a lightweight MobileNetV3Small backbone with a convolutional block attention module and Low-rank Attention with Shared Efficient Representations (LASER) to enhance feature representation. Feature maps are projected via a 1 × 1 convolution into token sequences and processed through a transformer encoder to capture long-range dependencies. A parallel global average pooling extracts aggregated features, fused using a cross-type interaction (CTI) module.

[RESULTS] The model was evaluated on 18,942 CT images and achieved 99.34% accuracy, AUC-ROC of 0.9996, Cohen's Kappa of 0.9897, and MCC of 0.9859, outperforming ResNet50, EfficientNetB0, and ViT variants with only 1.26 million parameters.

[CONCLUSIONS] Explainability analyses using Grad-CAM, Grad-CAM++, and attention visualisation suggest that the model focuses on clinically relevant regions.

MeSH Terms

Pancreatic Neoplasms; Humans; Deep Learning; Tomography, X-Ray Computed; Algorithms; Image Processing, Computer-Assisted; ROC Curve