Automated Quantification of Stroke Damage on Brain Computed Tomography Scans: e-ASPECTS

2015 
Emergency radiological diagnosis of acute ischaemic stroke requires the accurate detection and appropriate interpretation of relevant imaging findings. Non-contrast computed tomography (CT) provides fast and low-cost assessment of the early signs of ischaemia and is the most widely used diagnostic modality for acute stroke. The Alberta Stroke Program Early CT Score (ASPECTS) is a quantitative and clinically validated method to measure the extent of ischaemic signs on brain CT scans. The CE-marked electronic-ASPECTS (e-ASPECTS) software automates the ASPECTS score. Anglia Ruskin Clinical Trials Unit (ARCTU) independently carried out a clinical investigation of the e-ASPECTS software, an automated scoring system which can be integrated into the diagnostic pathway of an acute ischaemic stroke patient, thereby assisting the physician with expert interpretation of the brain CT scan. Here we describe a literature review of the clinical importance of reliable assessment of early ischaemic signs on plain CT scans, and of technologies automating these processed scoring systems in ischaemic stroke on CT scans focusing on the e-ASPECTS software. To be suitable for critical appraisal in this evaluation, the published studies needed a sample size of a minimum of 10 cases. All randomised studies were screened and data deemed relevant to demonstration of performance of ASPECTS were appraised. The literature review focused on three domains: i) interpretation of brain CT scans of stroke patients, ii) the application of the ASPECTS score in ischaemic stroke, and iii) automation of brain CT analysis. Finally, the appraised references are discussed in the context of the clinical impact of e-ASPECTS and the expected performance, which will be independently evaluated by a non-inferiority study conducted by the ARCTU.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    27
    References
    7
    Citations
    NaN
    KQI
    []