An enhanced deep learning architecture for classification of Tuberculosis types from CT lung images
- Submitting institution
-
Middlesex University
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1218
- Type
- E - Conference contribution
- DOI
-
10.1109/ICIP40778.2020.9190815
- Title of conference / published proceedings
- 2020 IEEE International Conference on Image Processing (ICIP)
- First page
- 2486
- Volume
- -
- Issue
- -
- ISSN
- 1522-4880
- Open access status
- Compliant
- Month of publication
- September
- Year of publication
- 2020
- URL
-
http://eprints.mdx.ac.uk/30256/
- 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
-
2
- Research group(s)
-
-
- Citation count
- -
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Identifying tuberculosis types plays a crucial role for clinicians to administer correct antibiotic treatments. However, challenges remain due to infected regions presenting not only similar appearances but varying by depth. Hence many slices of a 3D Computerised Tomography lung images appear normal, leading most of the existing 2D approaches fail. This paper is significant because it proposes an enhanced deep learning architecture to incorporate both 2D and 3D features and achieves state of the art results, contributing to the development of decision-support systems while minimising the invasiveness incurred by conventional means of obtaining biopsies from the lung.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -