Ada June Zhang BS, Presenter: Nothing to Disclose

John Galeotti PhD, Abstract Co-Author: Nothing to Disclose

Vikas Shivaprabhu, Abstract Co-Author: Nothing to Disclose

George Dewitt Stetten MD,PhD, Abstract Co-Author: Institutional license agreement, Insituvue, Inc

Vijay Saradhi Gorantla MD, PhD, Abstract Co-Author: Nothing to Disclose

Peripheral nerve injury (PNI) is a worldwide problem that results in significant disability. Objective, non-invasive diagnosis of the precise site, nature and degree of PNI could enable timely intervention and appropriate management. Currently, magnetic resonance imaging (MRI) can be used to image peripheral nerves. However, MRI is expensive, non-portable, and cannot be used in patients with shrapnel. The recent advent of phased-array ultrasound devices at frequencies of up to 70 MHz offers an attractive alternative. This work explores for the first time the use of high-resolution ultrasound as a basis for automated analysis of nerve fascicles.

Cross-sectional ultrasound of the human medial nerve was performed using a VisualSonics Vevo 2100 system at 50 MHz. Penetration depth was approx. 5mm with approx. 30 micron resolution. Sequential scans were taken starting at the wrist and progressing proximally by 1 inch. Median nerve fascicles were easily identified on scans. The nerve was tracked using the Lucas-Kanade algorithm for sequential image alignment and temporal averaging to attenuate speckling. Based on these data, we have implemented several algorithms aiming to segment fascicles from high-resolution ultrasound. For example, classification of filter responses using a support vector machine (SVM).

Our results confirm that 50 MHz ultrasound is capable of imaging nerve fascicles at high resolution. Results from the SVM classifier indicate that adjunct incorporation of computer vision and machine learning algorithms show potential towards automated segmentation and 3-D reconstruction of nerve fascicles from high-resolution ultrasound.

A high-resolution, inexpensive and portable imaging modality for sequential or real time evaluation and monitoring of PNIs could greatly improve the management of these injuries. Accurate identification of changes in nerve anatomy such as edema, myelin debris or fascicular/axonal changes could help objectively diagnose nerve injury or monitor nerve regeneration after trauma. This work presents preliminary findings in automated tracking and segmentation of peripheral nerve anatomy using high-resolution ultrasound.

High-resolution ultrasound is capable of imaging nerve fascicles and could thus help evaluate and monitor patients undergoing regenerative, repair, or transplant strategies for nerve injuries. 

Zhang, A, Galeotti, J, Shivaprabhu, V, Stetten, G, Gorantla, V, Automated Analysis of Nerve Fascicular Anatomy Using High-Resolution Ultrasound Imaging.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12043455.html