P15. Evaluation and comparison of 3D-ultrasound of the median nerve with common 2D-ultrasound

Background Ultrasound is still regarded as an imaging modality that considerably depends on the examiner's experience. Additionally, storage of only single images or short video sequences restricts re-evaluation and repeated measurements. While 3D-ultrasound (3D-US) might overcome these limitations, its use is currently mainly hampered by relatively low image quality. Here, we want to compare high-resolution 3D-US with 2D-US for the examination of the median nerve. Methods In 22 healthy probands 3D-US of both median nerves was performed by 2 examiners (Ex1, Ex2) using an 18MHz linear transducer (Toshiba Aplio 500) connected to the Curefab CS (Curefab Technologies GmbH, Munich, Germany). A single virtual 3D-stack of the ventral forearm ranging from the distal wrist crease to 20cm in proximal direction was generated. Subsequently, cross section area (CSA) of each median nerve was determined by common 2D-US in 3 randomly assigned distances (1–20cm) proximal to the distal wrist crease. Finally, for each median nerve CSA was measured in the 3D-stack at the same distances as for 2D-US. Furthermore, quality of 3D-US was evaluated with a semi-quantitative scale (1: good quality, nerve and fibre structures can be depicted axially and horizontally; 2: sufficient quality, CSA measurement is possible, but horizontal visualisation of fibre structures is limited by blurry view or too much artefacts; 3: poor quality, CSA measurement is not possible). While 3D-US of the median nerve was done by 2 examiners, CSA measurements and evaluation of quality were done by 1 examiner. Results CSA of the median nerve from 3D-US was 0.8mm 2 (Ex1) respectively 0.5mm 2 (Ex2) smaller than CSA obtained by 2D-US. Applying intraclass correlation coefficient, interrater agreement was moderate with 0.68. 88% of CSA from 3D-US differed by just ±1mm 2 between examiners. Quality assessment showed a good quality of 3D-data in 55% (Ex1) and 45% (Ex2), a sufficient quality in 39% and 48% and a poor quality with missing CSA analysis in just 6% respectively 7%. Furthermore, for 2D- and 3D-US, there was a significant inverse correlation between the distance from the wrist and the median nerve's CSA (Pearson r =0.32–0.36, each p 0.05 ). Conclusion Although it is technically challenging to scan the median nerve over a distance of 20cm without interruption, 3D-US showed a moderate to good agreement with conventional 2D-US examination. Contrary to 2D-US, 3D-US allows measurements of CSA exactly perpendicular to the nerve's course, thus yielding more accurate results. Therefore, we propose, that 3D-US of nerves is a promising diagnostic tool to minimise time effort by separating data acquisition from subsequent data analysis since the virtual 3D-stack contains all information for a profound offline evaluation; in addition 3D-US facilitates comparability of follow-up examinations.