Semi-supervised labelling of the femur in a whole-body post-mortem CT database using deep learning. (July 2020)

• Record Type: Journal Article
• Authors: Peña-Solórzano, C.A.; Albrecht, D.W.; Bassed, R.B.; Gillam, J.; Harris, P.C.; Dimmock, M.R.
• Is Part Of: Computers in biology and medicine. Volume 122(2020)
• Publication Date: 2020-07
• Subjects: CT -- Deep learning -- Autoencoder -- Semi-supervised -- Machine learning -- Femur localization -- Femoral head representation -- Knee representation -- Post-mortem -- Forensic; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2020.103797
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: A deep learning pipeline was developed and used to localize and classify a variety of implants in the femur contained in whole-body post-mortem computed tomography (PMCT) scans. The results provide a proof-of-principle approach for labelling content not described in medical/autopsy reports. The pipeline, which incorporated residual networks and an autoencoder, was trained and tested using n = 450 full-body PMCT scans. For the localization component, Dice scores of 0.99, 0.96, and 0.98 and mean absolute errors of 3.2, 7.1, and 4.2 mm were obtained in the axial, coronal, and sagittal views, respectively. A regression analysis found the orientation of the implant to the scanner axis and also the relative positioning of extremities to be statistically significant factors. For the classification component, test cases were properly labelled as nail (N + ), hip replacement (H + ), knee replacement (K + ) or without-implant (I − ) with an accuracy >97%. The recall for I − and H + cases was 1.00, but fell to 0.82 and 0.65 for cases with K + and N + . This semi-automatic approach provides a generalized structure for image-based labelling of features, without requiring time-consuming segmentation. Highlights: Femur localization is not significantly affected by decedent biological variables. Unsupervised ML can be readily used to differentiate long bone metallic implants. Interplay between scanner geometry, presence of metal and non-standard positioning limit the performance of ML techniques. A ML pipeline can facilitate structuring retrospective databases of images to improve accessibility for academics.