Evaluation of the effectiveness of artificial intelligence for ultrasound guided peripheral nerve and plane blocks in recognizing anatomical structures.

Background: We aimed to assess the accuracy of artificial intelligence (AI) based real-time anatomy identification for ultrasound-guided peripheral nerve and plane block in eight regions in this prospective observational study.

Methods: After obtaining ethics committee approval and written informed consent from 40 healthy volunteers (20 men and 20 women, between 18 and 72 years old), an ultrasound device installed with AI software (Nerveblox, SmartAlfa, Turkey) were used to scan regions of the cervical plexus, brachial plexus, pectoralis (PECS), rectus sheet, femoralis, canalis adductorius, popliteal, and ESP by three anesthesiology trainees. During scanning by a trainee, once software indicates 100 % scan success of associated anatomic landmarks, both raw and labeled ultrasound images were saved, assessed, and validated using a 6-point scale between 0 and 5 by two expert validators. Evaluation scores of the validators for each block were compared according to demographics (gender, age, and BMI) and block type exists.

Results: The scores were not different except ESP, femoralis, and cervical plexus regions between the experts. The mean scores of the experts for the PECS, popliteal and rectus sheath were significant between males and females (p < 0.05). In terms of BMI, significant differences in the scores were observed only in the canalis adductorius, brachial plexus, and ESP regions (p < 0.05).

Conclusions: Ultrasound guided AI-based anatomy identification was performed in commonly used eight block regions by the trainees where AI technology can successfully interpret the anatomical structures in real-time sonography which would be valuable in assisting anesthesiologists.