Anestesia e Pesquisa Clínica
Acesso livre

Abstrato

Imaging Performance of a Handheld Ultrasound System for Thoracic Epidural Analgesia Guidance

Amitabh Gulati, Chandni Patel, Ajay Patel, Jason Chang, Rachel Newman, Kathryn Ozgun, Adam J. Dixon4*, F. William Mauldin

The aim of this study was to evaluate the performance of an ultrasound system with Computer-Aided Detection (CAD) of thoracic spinal landmarks and a thoracic epidural needle guidance algorithm. This study was approved by the institutional review board of the Memorial Sloan Kettering Cancer Center, NY, USA. Algorithm performance was evaluated by imaging the thoracic spines of 55 adult volunteers presenting for pain therapy unrelated to the thoracic spine. The accuracy of the CAD for neuraxial landmark detection was assessed by comparing automated measurements of the lamina depth and midline position to ground-truth annotations provided by physician reviewers. The sensitivity and specificity of the needle guidance algorithm was measured by reviewing the proposed needle trajectory in 435 images. The CAD algorithm detected the lamina depth with an error of 2 mm (95% CI, 1.79-2.39 mm). The spinal midline was detected with an error of 0.5 mm (95% CI, 0.32-0.69 mm). The needle guidance algorithm yielded a sensitivity of 92.4% (95% CI, 90.2%-94.6%) and a specificity of 89.7% (95% CI, 86.3%-93.2%). The results of this study demonstrate the feasibility of computer-aided interpretation for ultrasound imaging of the thoracic spine and for guidance of epidural needle insertion.