SBIR Phase II: Low-cost needle guidance system for bedside lumbar puncture

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase II project is to improve the quality and efficiency of bedside spinal access procedures by developing a needle guidance system that interfaces with existing ultrasound machines. Lumbar punctures (LPs) are performed to diagnose and treat neurological conditions such as meningitis. The traditional LP technique requires practitioners to manually feel for spinal landmarks to form a mental image of the anatomy before blindly inserting the needle into the spine, aiming for a small target containing spinal fluid. This methodology can be challenging, costly, and time-consuming and is highly dependent on practitioner experience and patient body type. Up to 42% of procedures fail to access the target. Failed cases require radiological intervention, increasing the cost of care, lengthening patient stay, and exposing patients to radiation. Hospitals in the United States lose an estimated $2 billion annually due to inefficiencies and failures in LPs. The proposed needle guidance system will enable bedside spinal access procedures to be performed under real-time ultrasound imaging, significantly improving first-attempt success rates. The guidance technology can further be applied in other clinical segments, improving the quality of care across a variety of needle-based procedures. This Small Business Innovation Research (SBIR) Phase II project aims to accomplish two primary objectives: 1) Complete development of a needle guidance system designed to interface with existing ultrasound machines; and 2) Develop an AI-powered anatomy detection software feature for the guidance system. These combined objectives will generate a functional, intuitive, and accessible solution that minimizes barriers to adoption while maximizing clinical and operational value. The proposed research involves conducting an array of safety and reliability studies to investigate the performance of the guidance system under realistic conditions. In developing a robust anatomy detection software feature, a spinal ultrasound data set will be created from a variety of non-patient volunteers. The data will be analyzed, processed, and used to train a machine learning model for anatomy detection. It is anticipated that the results of this project will demonstrate a sufficiently safe and efficacious system that can consistently guide a needle to an intended target with an error of less than 3 millimeters. The anatomy detection feature is expected to perform with at least 93% sensitivity and 85% specificity in identifying five key spinal landmarks; this level of performance would significantly reduce the knowledge barrier to performing ultrasound-guided interventions. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.