Identification of Anterior Cervical Spinal Instrumentation Using a Smartphone Application Powered by Machine Learning

Cross-sectional study. The purpose of this study is to develop and validate a machine learning algorithm for the automated identification of anterior cervical discectomy and fusion (ACDF) plates from smartphone images of anterior-posterior (AP) cervical spine radiographs. Identification of existing...

Full description

Saved in:
Bibliographic Details
Published in:Spine (Philadelphia, Pa. 1976) Pa. 1976), 2022-05, Vol.47 (9), p.E407-E414
Main Authors: Schwartz, John T, Valliani, Aly A, Arvind, Varun, Cho, Brian H, Geng, Eric, Henson, Philip, Riew, K Daniel, Lehman, Ronald A, Lenke, Lawrence G, Cho, Samuel K, Kim, Jun S
Format: Article
Language:English
Subjects:
Bone Plates
Cervical Vertebrae - diagnostic imaging
Cervical Vertebrae - surgery
Cross-Sectional Studies
Diskectomy - methods
Humans
Machine Learning
Retrospective Studies
Smartphone
Spinal Fusion - methods
Treatment Outcome
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Cross-sectional study. The purpose of this study is to develop and validate a machine learning algorithm for the automated identification of anterior cervical discectomy and fusion (ACDF) plates from smartphone images of anterior-posterior (AP) cervical spine radiographs. Identification of existing instrumentation is a critical step in planning revision surgery for ACDF. Machine learning algorithms that are known to be adept at image classification may be applied to the problem of ACDF plate identification. A total of 402 smartphone images containing 15 different types of ACDF plates were gathered. Two hundred seventy-five images (∼70%) were used to train and validate a convolution neural network (CNN) for classification of images from radiographs. One hundred twenty-seven (∼30%) images were held out to test algorithm performance. The algorithm performed with an overall accuracy of 94.4% and 85.8% for top-3 and top-1 accuracy, respectively. Overall positive predictive value, sensitivity, and f1-scores were 0.873, 0.858, and 0.855, respectively. This algorithm demonstrates strong performance in the classification of ACDF plates from smartphone images and will be deployed as an accessible smartphone application for further evaluation, improvement, and eventual widespread use.Level of Evidence: 3.
ISSN:0362-2436
1528-1159
DOI:10.1097/BRS.0000000000004172