Glioma Segmentation on Multicenter 2D-Based Magnetic Resonance Imaging Using Low-Rank Adaptation Tuning of a Foundation Model: An External Test Evaluation.
Authors
Affiliations (17)
Affiliations (17)
- Department of Intelligent Information Engineering, Research Promotion Unit, School of Medical Sciences, Fujita Health University, Toyoake City, Aichi, Japan.
- Department of Neurosurgery, Tokyo Saiseikai Central Hospital, Minato-Ku, Tokyo, Japan.
- Department of Neurosurgery, Keio University School of Medicine, Shinjuku-Ku, Tokyo, Japan.
- Department of Intelligent Information Engineering, Research Promotion Unit, School of Medical Sciences, Fujita Health University, Toyoake City, Aichi, Japan. [email protected].
- Department of Radiological Technology, Faculty of Medical Technology, Niigata University of Health and Welfare, Niigata City, Niigata, Japan. [email protected].
- The Asahi Shimbun Company, Chuo-Ku, Tokyo, Japan.
- Center for Preventive Medicine and Department of Diagnostic Radiology, Keio University School of Medicine, Shinjuku-Ku, Tokyo, Japan.
- Visionary Imaging Services, Inc., Yokohama City, Kanagawa, Japan.
- Department of Radiology, Keio University School of Medicine, Shinjuku-Ku, Tokyo, Japan.
- Department of Neurosurgery, Fujita Health University School of Medicine, Toyoake City, Aichi, Japan.
- Department of Neurosurgery, Graduate School of Medical Science, Kanazawa University, Kanazawa City, Ishikawa, Japan.
- Department of Neurosurgery, Faculty of Medicine, Hokkaido University, Sapporo City, Hokkaido, Japan.
- Department of Neurosurgery, Graduate School of Medical and Dental Sciences, Kagoshima University, Kagoshima City, Kagoshima, Japan.
- Department of Neurosurgery, Kyushu University School of Medicine, Fukuoka City, Fukuoka, Japan.
- Department of Neurosurgery, Kumamoto University School of Medicine, Kumamoto City, Kumamoto, Japan.
- Department of Neurosurgery, Tohoku University Graduate School of Medicine, Sendai City, Miyagi, Japan.
- Department of Neurosurgery, International University of Health and Welfare Ichikawa General Hospital, Ichikawa City, Chiba, Japan.
Abstract
Research on foundation models is actively progressing. The segment anything model (SAM) and MedSAM are representative foundation models for image segmentation. Recently, low-rank adaptation (LoRA) has been developed, allowing parameter updates without retraining the entire model, thus solving the problem with large data and time required for task-specific fine-tuning. Although many studies have used public databases, few have focused on local data. Moreover, to our knowledge, no studies have fine-tuned MedSAM using LoRA. We aimed to evaluate SAM, MedSAM, and their LoRA-tuned variants (SAM-LoRA and MedSAM-LoRA) using brain magnetic resonance images of gliomas from five centers in Japan and to compare their performance. We used 2D-based fluid-attenuated inversion recovery axial images and conducted parameter optimization based on four-fold cross-validation (189 cases) and external test evaluation (75 cases) using cases collected retrospectively. Dice coefficients, intersection over union (IoU), and the 95% Hausdorff distance (HD95) were used to evaluate the performance of SAM, MedSAM, SAM-LoRA, and MedSAM-LoRA. Additionally, subgroup evaluations were performed according to scanner manufacturer, glioma location, and calcification status. In the external test at the case level using SAM-LoRA and MedSAM-LoRA, the Dice coefficients/IoU/HD95 were 0.9166/0.8506/1.3517 and 0.9051/0.8342/1.8061, respectively. Both SAM-LoRA and MedSAM-LoRA demonstrated significantly higher Dice coefficients and IoU, as well as significantly lower HD95 values, compared with SAM and MedSAM. Subgroup evaluations also showed highly accurate extraction across different scanner manufacturers, glioma locations, and calcification statuses. SAM-LoRA and MedSAM-LoRA achieved high accuracy on an externally evaluated dataset, suggesting potential utility.