Automatic MRI-Based segmentation of ablation zones in spine and liver: A comparative study of network performance and influencing factors.
Authors
Affiliations (8)
Affiliations (8)
- Institute for Medical Engineering, Otto-von-Guericke-University Magdeburg, Magdeburg, Germany. [email protected].
- Research Campus Stimulate, Magdeburg, Germany. [email protected].
- Institute for Medical Informatics and Artificial Intelligence, University Hospital Schleswig-Holstein Campus Kiel, Kiel, Germany. [email protected].
- Institute for Medical Informatics and Artificial Intelligence, University Hospital Schleswig-Holstein Campus Kiel, Kiel, Germany.
- Research Campus Stimulate, Magdeburg, Germany.
- Institute for Diagnostic and Interventional Radiology, Hanover Medical School, Hanover, Germany.
- Radiology and Nuclear Medicine, Medical Faculty, University Hospital Magdeburg, Magdeburg, Germany.
- Institute for Medical Engineering, Otto-von-Guericke-University Magdeburg, Magdeburg, Germany.
Abstract
Ablation therapies are a treatment option for cancer patients, particularly for conditions such as spinal metastases and liver tumors. Precisely delineating ablation zones is essential for accurately assessing treatment success. However, research in MRI-guided interventions remains limited for automated segmentation approaches and quantitative analysis. We developed a framework for automated segmentation of ablation zones following thermal interventions. The performance was tested on two representative types of clinical cases from different clinical sites: post-ablative liver lesions and spinal metastases. Four leading neural networks (nnUNet, TransUNet, SwinUNETR, and SwinUNETR-V2) were evaluated for their segmentation accuracy in segmenting necrotic tissue. Additionally, a statistical analysis was performed to investigate the influence of an optimized image ROI selection on the segmentation performance. The nnUNet achieved the highest segmentation performance, with a Dice Similarity Coefficient of 83.3 ± 13.2% for spinal metastases and 82.2 ± 12.4% for liver lesions. The statistical analysis revealed that cropping images to a standardized ROI size provided the optimal balance between user interaction and networks' performance. The automated segmentation accuracies are comparable to the inter-rater variability observed among radiologists for spinal imaging and establish the new state-of-the-art for liver MRIs, indicating the potential to facilitate the clinical routine for interventions.