Computer-aided diagnosis for acute ischaemic stroke using non-contrast computed tomography: a systematic review.
| Abstract: |
- Introduction: Ischaemic stroke is one of the leading causes of morbidity and mortality world-wide. Non-contrast computed tomography (NCCT) remains the initial imaging modality for investigation of acute ischaemic strokes. Computer-aided diagnosis (CAD) and machine learning is an emerging field in medical imaging. The aim in completing this project is to review and critically appraise the current scope of research on CAD and machine learning in acute ischaemic stroke using NCCT.
- Methods: A systematic review of the literature from 7 medical and computer engineering databases was performed for studies assessing CAD in acute ischaemic stroke using NCCT. The non-duplicated articles were reviewed via a two-step process, title/abstract and full text reviews, by two independent reviewers. The included articles were critically analysed using the QUADAS-2 tool and appropriate data and information was extracted for qualitative and narrative review.
- Results: The literature search resulted in 11,235 non-duplicated articles, with 68 articles accepted following the full-text review stage. The majority of the articles presented different CAD/machine learning methods with small test data sets and poor documentation of population demographics. Many studies did not specify the acuity of the scans making up their database. Automated ASPECTS software, particularly Brainomix Ltd., was the only method which was presented in multiple publications.
- Conclusion: The review illustrates that the majority of CAD and machine learning for diagnosis of acute ischaemic strokes with NCCT has not been clinically validated. Further research with larger and more relevant data sets, in addition to collaboration between clinicians and researchers, is required to assist in advancing the field of CAD prior to its adoption into clinical practice.
|
| Project type: |
|
| Imaging keywords: |
|
| Application / disease keywords: |
|
| Supervisor(s): |
|
| Programme: |
|
| Year: |
|
| Associated publication(s): |
|
