TY - JOUR
AU - Rohling, Robert
AB - Int J CARS (2017) 12:1189–1198 DOI 10.1007/s11548-017-1575-8 ORIGINAL ARTICLE SLIDE: automatic spine level identification system using a deep convolutional neural network 1 3 4 1 Jorden Hetherington · Victoria Lessoway · Vit Gunka · Purang Abolmaesumi · 1,2 Robert Rohling Received: 30 January 2017 / Accepted: 20 March 2017 / Published online: 30 March 2017 © CARS 2017 Abstract Results The proposed CNN successfully discriminates Purpose Percutaneous spinal needle insertion procedures ultrasound images of the sacrum, intervertebral gaps, and often require proper identification of the vertebral level to vertebral bones, achieving 88% 20-fold cross-validation effectively and safely deliver analgesic agents. The current accuracy. Seventeen of 20 test ultrasound scans had success- clinical method involves “blind” identification of the verte- ful identification of all vertebral levels, processed at real-time bral level through manual palpation of the spine, which has speed (40 frames/s). only 30% reported accuracy. Therefore, there is a need for Conclusion A machine learning system is presented that better anatomical identification prior to needle insertion. successfully identifies lumbar vertebral levels. The small Methods A real-time system was developed to identify the study on human subjects demonstrated real-time perfor- vertebral level from a sequence of ultrasound images, fol- mance. 
TI - SLIDE: automatic spine level identification system using a deep convolutional neural network
JF - International Journal of Computer Assisted Radiology and Surgery
DO - 10.1007/s11548-017-1575-8
DA - 2017-03-30
UR - https://www.deepdyve.com/lp/springer-journals/slide-automatic-spine-level-identification-system-using-a-deep-0B4mN5UTkV
SP - 1189
EP - 1198
VL - 12
IS - 7
DP - DeepDyve
ER -