融合视觉与触觉传感器，提升手术中组织边界识别精度
Where is the Boundary: Multimodal Sensor Fusion Test Bench for Tissue Boundary Delineation
- 用视觉引导定位边界，再结合麦克风和力传感器精修
- 多模态融合使组织分类准确率显著提升
- 适合研究手术机器人感知系统或医疗传感的团队
机器人辅助神经外科手术因工具更灵活、精准和可控而日益受到关注，可改善患者预后。然而，此类系统常削弱外科医生的自然感官反馈，这对识别组织（尤其在肿瘤手术中区分健康与病变组织）至关重要。尽管成像和力感测量缓解了反馈缺失问题，但针对组织边界精确划分的多模态传感研究仍有限。本文提出一个用户友好、模块化的测试平台，用于评估与集成互补的多模态传感器以实现组织识别。系统首先通过视觉引导估计边界位置，再利用接触式麦克风和力传感器采集的数据进行优化。支持实时数据采集与可视化，通过交互式图形界面实现。实验结果表明，多模态融合显著提升了材料分类准确率。该平台为探索手术应用中的传感器融合提供了可扩展的软硬件解决方案，并展示了多模态方法在实时组织边界界定中的潜力。
原文摘要 · Abstract (English)
Robot-assisted neurological surgery is receiving growing interest due to the improved dexterity, precision, and control of surgical tools, which results in better patient outcomes. However, such systems often limit surgeons' natural sensory feedback, which is crucial in identifying tissues -- particularly in oncological procedures where distinguishing between healthy and tumorous tissue is vital. While imaging and force sensing have addressed the lack of sensory feedback, limited research has explored multimodal sensing options for accurate tissue boundary delineation. We present a user-friendly, modular test bench designed to evaluate and integrate complementary multimodal sensors for tissue identification. Our proposed system first uses vision-based guidance to estimate boundary locations with visual cues, which are then refined using data acquired by contact microphones and a force sensor. Real-time data acquisition and visualization are supported via an interactive graphical interface. Experimental results demonstrate that multimodal fusion significantly improves material classification accuracy. The platform provides a scalable hardware-software solution for exploring sensor fusion in surgical applications and demonstrates the potential of multimodal approaches in real-time tissue boundary delineation.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。