Machine Learning Algorithms to Improve Needle Localization in Procedures with Limited Visibility
Rutgers Inventor(s): Cosmas Mwikirize; John Nosher; Ilker Hacihaliloglu
Awarded: January 2019
Summary: Ultrasound is a popular imaging modality for guidance of minimally invasive procedures such as biopsies and regional anesthesia. However, needle visibility is often lost when advancing needles towards deep targets or complicated anatomy. This reduces procedure efficacy and could result in injury and post-procedure complications. We have developed advanced machine learning algorithms that achieve real-time, automatic and robust localization of needles in procedures where needle visibility to the naked eye is limited. Rutgers TechAdvance award will be used to: 1) integrate the algorithms into a software platform compatible with existing cart-based and portable ultrasound systems, and 2) prepare the technology for clinical use through validation in clinical studies with physician testing in animal models.
Market Applications:
- Localization of needles in percutaneous ultrasound-guided interventional procedures such as regional anesthesia, peripheral vascular interventions, biopsies, fluid aspirations and fetal medicine.
- Localization of other interventional devices such as guidewires and catheters under ultrasound guidance or other imaging modalities such as fluoroscopy and endoscopy.
