COVID-CT-Mask-Net: Prediction of COVID-19 from CT Scans Using Regional Features
- Submitting institution
-
City, University of London
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1274
- Type
- U - Working paper
- Platform
- medRxiv
- Month of publication
- November
- Year of publication
- 2020
- URL
-
https://doi.org/10.1101/2020.10.11.20211052
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
0
- Research group(s)
-
-
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- The model introduced segments instances of lesions (Ground Glass Opacity, Consolidation) in chest CT scans and predicts classes of these scans (COVID-19, Common Pneumonia, Control/Negative) in a single shot. The significance of the methodology is based on the augmentation of the Region of Interest (RoI) layer in Mask R-CNN with a classification branch that constructs a batch of RoIs, from which the class of the image is predicted. Currently under review at Scientific Reports: https://scientific-reports-under-consideration.nature.com/channels/2717-under-consideration. Code and results: https://github.com/AlexTS1980/COVID-Single-Shot-Model
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
