Efficient CBCT Segmentation via nnU-Net with Structure-Aware Post-processing and Interactive Refinement

Changkai Ji, Yusheng Liu, Yuxian Jiang, Lisheng Wang

Keywords: nnU-Net, Structure Aware Post-processing, Computational efficiency

TL;DR: This paper presents an nnU-Net-based method with Structure Aware Post-processing and optimized inference strategies for efficient and accurate multi-class anatomical structure segmentation in CBCT.

Abstract: Accurate segmentation of anatomical structures from cone-beam computed tomography (CBCT) is essential for clinical applications in dentistry, maxillofacial surgery, and orthodontics. The ToothFairy3 Challenge has a comprehensive 77-class segmentation task, emphasizing both accuracy and computational efficiency. In this work, we present a method based on the nnU-Net framework, enhanced with a Structure Aware Post-processing (SAP) strategy. nnU-Net serves as a backbone for multi-class segmentation, while SAP refines predictions by introducing individualized thresholds for each anatomical structure, thereby mitigating noise and preserving clinically important fine structures. To further improve efficiency, we disabled mirroring augmentation during training and employed inference acceleration strategies, including the removal of test-time augmentation and optimized interpolation on floating-point tensors. Experimental results validate the effectiveness of our approach in balancing segmentation accuracy with computational efficiency. To further ensure robustness in challenging clinical scenarios, we also utilize an interactive refinement module based on nnInteractive. This strategy allows clinicians to correct local segmentation errors with minimal user guidance, providing a safety net for complex anatomical variations.

Changes Summary:

Following the suggestion, we have integrated the core contributions of our second submission ("Balancing Accuracy and Efficiency in Inferior Alveolar Canal Segmentation from CBCT Scans") into this camera-ready manuscript.

Specifically, the following major revisions have been made:

Title Update: We refined the title to "Efficient CBCT Segmentation via nnU-Net with Structure-Aware Post-processing and Interactive Refinement" to reflect the expanded scope.

Methodology: A new subsection (Section 2.5: Interactive Refinement Module) has been added, detailing the nnInteractive framework, the early-prompt fusion strategy, and the AutoZoom mechanism.

Experiments: We expanded the Results section to include the performance evaluation of the interactive module.

Discussion: We updated the Abstract, Introduction, and Conclusion to highlight how the interactive module serves as a robust "human-in-the-loop" complement to the automated pipeline.

Main Tex File: samplepaper.tex

Confirm Latex Only: false

Authors Changed: false

Submission Number: 15