Technology Category
Computer Vision: SegmentationMachine Learning: Calibration & Uncertainty QuantificationKnowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
Performance Analysis of Deep Learning Models for Femur Segmentation in MRI Scan
🤖 AI Summary
This study addresses poor generalizability in MRI-based femur segmentation—caused by small anatomical structures and dataset bias—by systematically evaluating U-Net, Attention U-Net, U-KAN (a Kolmogorov–Arnold network-based architecture), and the prompt-driven vision transformer SAM 2 across 11,164 clinical MRI scans. It represents the first comparative assessment of CNNs versus prompt-guided transformer architectures for this task, with segmentation accuracy quantified via Dice scores (0.932–0.954). Results indicate that Attention U-Net achieves the best overall performance, while U-KAN demonstrates significantly improved accuracy in fine-grained regions—particularly the intercondylar ridge—highlighting superior modeling capability for small structures. The findings provide empirically grounded guidance for model selection and architectural design in medical image segmentation of subtle anatomical features.
Technology Category
Application Category
📝 Abstract
Convolutional neural networks like U-Net excel in medical image segmentation, while attention mechanisms and KAN enhance feature extraction. Meta's SAM 2 uses Vision Transformers for prompt-based segmentation without fine-tuning. However, biases in these models impact generalization with limited data. In this study, we systematically evaluate and compare the performance of three CNN-based models, i.e., U-Net, Attention U-Net, and U-KAN, and one transformer-based model, i.e., SAM 2 for segmenting femur bone structures in MRI scan. The dataset comprises 11,164 MRI scans with detailed annotations of femoral regions. Performance is assessed using the Dice Similarity Coefficient, which ranges from 0.932 to 0.954. Attention U-Net achieves the highest overall scores, while U-KAN demonstrated superior performance in anatomical regions with a smaller region of interest, leveraging its enhanced learning capacity to improve segmentation accuracy.
Problem
Research questions and friction points this paper is trying to address.
Evaluating deep learning models for femur segmentation in MRI scans
Comparing CNN-based and transformer-based models for medical image segmentation
Assessing model performance using Dice Similarity Coefficient metrics
Innovation
Methods, ideas, or system contributions that make the work stand out.
Uses U-Net, Attention U-Net, and U-KAN models
Compares CNN and transformer-based segmentation
Evaluates performance with Dice Similarity Coefficient
🔎 Similar Papers
No similar papers found.
