Weakly-supervised Mamba-Based Mastoidectomy Shape Prediction for Cochlear Implant Surgery Using 3D T-Distribution Loss

2025-05-23

AI Summary

To address the challenge of accurately predicting the mastoidectomy region in cochlear implantation surgery, this paper proposes a weakly supervised, three-dimensional anatomical shape prediction method based on preoperative CT scans. We introduce the 3D Mamba architecture—novel in otologic surgical planning—to construct an end-to-end voxel-level morphological prediction model. A custom 3D t-distribution loss function is designed to explicitly model geometric variability of the mastoid process. Furthermore, we propose a fully annotation-free weakly supervised paradigm, leveraging self-supervised model outputs as pseudo-labels to generate supervision signals. Evaluated on clinical datasets, our method significantly outperforms state-of-the-art approaches: predicted boundaries better align with anatomical constraints and surgical requirements, while robustness and generalizability are substantially improved. The framework delivers a reliable, deployable AI-assisted tool for preoperative surgical planning.

Technology Category

• Machine Learning: Learning with Manifolds
• Computer Vision: 3D Computer Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Large pretrained models with web data
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Cochlear implant surgery is a treatment for individuals with severe hearing loss. It involves inserting an array of electrodes inside the cochlea to electrically stimulate the auditory nerve and restore hearing sensation. A crucial step in this procedure is mastoidectomy, a surgical intervention that removes part of the mastoid region of the temporal bone, providing a critical pathway to the cochlea for electrode placement. Accurate prediction of the mastoidectomy region from preoperative imaging assists presurgical planning, reduces surgical risks, and improves surgical outcomes. In previous work, a self-supervised network was introduced to predict the mastoidectomy region using only preoperative CT scans. While promising, the method suffered from suboptimal robustness, limiting its practical application. To address this limitation, we propose a novel weakly-supervised Mamba-based framework to predict accurate mastoidectomy regions directly from preoperative CT scans. Our approach utilizes a 3D T-Distribution loss function inspired by the Student-t distribution, which effectively handles the complex geometric variability inherent in mastoidectomy shapes. Weak supervision is achieved using the segmentation results from the prior self-supervised network to eliminate the need for manual data cleaning or labeling throughout the training process. The proposed method is extensively evaluated against state-of-the-art approaches, demonstrating superior performance in predicting accurate and clinically relevant mastoidectomy regions. Our findings highlight the robustness and efficiency of the weakly-supervised learning framework with the proposed novel 3D T-Distribution loss.

Problem

• Predicting mastoidectomy regions from preoperative CT scans
• Improving robustness in weakly-supervised surgical shape prediction
• Handling geometric variability with 3D T-Distribution loss function

Innovation

• Mamba-based framework for mastoidectomy prediction
• 3D T-Distribution loss handles geometric variability
• Weak supervision eliminates manual data labeling