Enhancing Peripheral Nerve Blocks: Real-Time Ultrasound Guidance for Nerve Detection and Needle Placement

Ultrasound-guided nerve blocks are an essential element of regional anesthesia, providing effective localized pain relief while minimizing the risks associated with systemic medications. However, their adoption remains limited by procedural complexity, reliance on practitioner expertise, and the inherent challenges of ultrasound image interpretation. This study introduces an AI-assisted platform that integrates simultaneous real-time detection of the nerve and needle for the transversus abdominis plane block, addressing a critical gap in the literature. Utilizing a dataset of annotated ultrasound images, the authors developed and trained segmentation models based on the R2U-Net architecture. The nerve detection model achieved a Dice score of 0.8384, while the needle detection model reached a Dice score of 0.8309. These models were integrated into an Android-based application, offering real-time visualization and intuitive controls with low latency (<200 milliseconds) and consistent frame rates (24 FPS). The system has the potential to improve procedural precision, reduce variability across practitioners, and expand access to nerve blocks for non-specialists. Future work will focus on expanding clinical datasets, conducting validation studies, and exploring model enhancements to further refine its applicability.Clinical Relevance- This system serves as a valuable companion tool in ultrasound-guided nerve blocks, providing real-time visualization and feedback to support clinicians, particularly those with less experience. By enhancing procedural guidance, it facilitates teaching and improves confidence, enabling a broader range of practitioners to perform nerve blocks effectively and safely.