Cardiac Rhythm Device Identification Using Neural Networks

• Published in: JACC. Clinical electrophysiology 2019-05, Vol.5 (5), p.576-586
• Main Authors: Howard, James P., Fisher, Louis, Shun-Shin, Matthew J., Keene, Daniel, Arnold, Ahran D., Ahmad, Yousif, Cook, Christopher M., Moon, James C., Manisty, Charlotte H., Whinnett, Zach I., Cole, Graham D., Rueckert, Daniel, Francis, Darrel P.
• Format: Article
• Language: English
• Subjects: cardiac rhythm devices; Databases, Factual; Humans; Image Processing, Computer-Assisted - methods; machine learning; neural networks; Neural Networks, Computer; pacemaker; Pacemaker, Artificial - classification; Radiography, Thoracic; Thorax - diagnostic imaging
• ISSN: 2405-500X; 2405-5018
• DOI: 10.1016/j.jacep.2019.02.003

This paper reports the development, validation, and public availability of a new neural network-based system which attempts to identify the manufacturer and even the model group of a pacemaker or defibrillator from a chest radiograph. Medical staff often need to determine the model of a pacemaker or defibrillator (cardiac rhythm device) quickly and accurately. Current approaches involve comparing a device’s radiographic appearance with a manual flow chart. In this study, radiographic images of 1,676 devices, comprising 45 models from 5 manufacturers were extracted. A convolutional neural network was developed to classify the images, using a training set of 1,451 images. The testing set contained an additional 225 images consisting of 5 examples of each model. The network’s ability to identify the manufacturer of a device was compared with that of cardiologists, using a published flowchart. The neural network was 99.6% (95% confidence interval [CI]: 97.5% to 100.0%) accurate in identifying the manufacturer of a device from a radiograph and 96.4% (95% CI: 93.1% to 98.5%) accurate in identifying the model group. Among 5 cardiologists who used the flowchart, median identification of manufacturer accuracy was 72.0% (range 62.2% to 88.9%), and model group identification was not possible. The network’s ability to identify the manufacturer of the devices was significantly superior to that of all the cardiologists (p < 0.0001 compared with the median human identification; p < 0.0001 compared with the best human identification). A neural network can accurately identify the manufacturer and even model group of a cardiac rhythm device from a radiograph and exceeds human performance. This system may speed up the diagnosis and treatment of patients with cardiac rhythm devices, and it is publicly accessible online. [Display omitted]