Adaptive U-shaped Convolutional Network for Segmentation and Analysis of Mouse Cardiac Images

Overview

This study introduces a Dynamic U-shaped Convolutional Network (DUCNet) for the segmentation of mouse cardiac slice images affected by myocardial infarction, achieving a Dice coefficient of 80.68%.

Background

Accurate segmentation of mouse cardiac images is crucial for understanding myocardial infarction and related cardiovascular diseases. Traditional manual segmentation methods are time-consuming and subject to variability.

Data Highlights

Key Findings

• The DUCNet utilizes a dynamic convolution to enhance the perception of irregular U-shaped local structures.
• It incorporates a dual-stream fusion block to improve segmentation performance.
• An attention gate mechanism is employed to suppress irrelevant information and highlight key features.
• The model was validated on a dataset of 243 mouse cardiac slice images.
• The DUCNet outperformed existing models in both average Dice coefficient and infarct size segmentation.

Clinical Implications

The DUCNet provides a robust automated solution for the segmentation and quantification of myocardial infarction in mouse models, potentially improving the efficiency and accuracy of cardiovascular research. This advancement may facilitate better understanding of myocardial injury severity.

Conclusion

The introduction of DUCNet offers improved segmentation capabilities for irregular infarct shapes.

Related Resources & Content

1. npj Digital Medicine, 2026 -- CFG-MambaNet: A Novel Mamba Network Utilizing Contextual and Frequency Guidance for Enhanced Medical Image Segmentation
2. npj Digital Medicine, 2026 -- Hierarchical Mamba-CNN Transducer for Enhanced Liver Tumor Segmentation in CT Imaging
3. Hierarchical Patch-Based Convolutional Neural Networks for Automated Segmentation of Head CT Scans in Computer-Assisted Craniomaxillofacial Surgery, 2022
4. ACC, AHA Issue New Acute Coronary Syndromes Guideline - American College of Cardiology, 2025
5. npj Digital Medicine — Enhanced Mamba Filtering Networks for Precise Segmentation of Hepatocellular Carcinoma Lesions in Abdominal CT Scans
6. Society for cardiovascular magnetic resonance expert consensus statement on quantitative myocardial perfusion cardiovascular magnetic resonance imaging
7. Prognostic impact of persistent microvascular obstruction on cardiac magnetic resonance after STEMI: A systematic review and meta-analysis
8. ACC, AHA Issue New Acute Coronary Syndromes Guideline - American College of Cardiology
9. STEMI-DTU: Left Ventricular Unloading in Anterior STEMI Without Shock - American College of Cardiology
10. Ex-vivo validation of nine algorithms for quantifying infarcts with late gadolinium enhancement cardiovascular magnetic resonance - ScienceDirect
11. Guidelines for experimental models of myocardial ischemia and infarction - PMC
12. A refined TTC assay precisely detects cardiac injury and cellular viability in the infarcted mouse heart | Scientific Reports
13. Infarctsize-AI: an efficient infarct size image analysis tool for small rodent myocardial infarction studies | Sciety Labs (Experimental)