Showing 1–1 of 1 results for author: Badhe, S

Automated Segmentation of Vertebrae on Lateral Chest Radiography Using Deep Learning

Authors: Sanket Badhe, Varun Singh, Joy Li, Paras Lakhani

Abstract: The purpose of this study is to develop an automated algorithm for thoracic vertebral segmentation on chest radiography using deep learning. 124 de-identified lateral chest radiographs on unique patients were obtained. Segmentations of visible vertebrae were manually performed by a medical student and verified by a board-certified radiologist. 74 images were used for training, 10 for validation, a… ▽ More The purpose of this study is to develop an automated algorithm for thoracic vertebral segmentation on chest radiography using deep learning. 124 de-identified lateral chest radiographs on unique patients were obtained. Segmentations of visible vertebrae were manually performed by a medical student and verified by a board-certified radiologist. 74 images were used for training, 10 for validation, and 40 were held out for testing. A U-Net deep convolutional neural network was employed for segmentation, using the sum of dice coefficient and binary cross-entropy as the loss function. On the test set, the algorithm demonstrated an average dice coefficient value of 90.5 and an average intersection-over-union (IoU) of 81.75. Deep learning demonstrates promise in the segmentation of vertebrae on lateral chest radiography. △ Less

Submitted 5 January, 2020; originally announced January 2020.

Comments: 10 pages, Accepted Poster presentation at Conference on Machine Intelligence in Medical Imaging 2018