CINA-ASPECTS

CINA-ASPECTS is a computer-aided diagnostic software designed to assist clinicians in the rapid and accurate assessment of early ischemic brain injury by automating the generation of the Alberta Stroke Program Early CT (ASPECT) Score from non-contrast head CT scans. Utilizing deep learning algorithms, the device processes NCCT images to identify and quantify ischemic changes within predefined brain regions for patients over 21 with confirmed MCA or ICA occlusions within 12 hours of symptom onset, thereby aiding in the evaluation of disease extent.

CINA-ASPECTS is a computer-aided diagnosis (CADx) software designed to assist clinicians in assessing and characterizing brain tissue abnormalities indicative of early ischemic stroke using non-contrast head CT (NCCT) data. The software is specifically indicated for patients older than 21 years with known Middle Cerebral Artery (MCA) or Internal Carotid Artery (ICA) occlusions, evaluated within 12 hours from the time last known well, to determine the extent of disease by generating an Alberta Stroke Program Early CT (ASPECT) Score. Its methodology involves the automatic processing of NCCT images, including reorientation, tilt-correction, segmentation of predefined ASPECTS Regions of Interest (ROIs), and extraction of image data. This information is then analyzed based on morphological characteristics of potential ischemic changes, with the results, including highlighted ROIs and a numerical score, presented to the clinician for review and potential editing after confirmation of a Large Vessel Occlusion. The core science behind CINA-ASPECTS leverages a deep learning artificial intelligence algorithm to synthesize extracted imaging features from NCCT scans acquired on validated CT systems (Canon, GE, Philips, Siemens) into a single ASPECT score and to visualize voxels contributing to this score via a 'heat map'. Standalone performance testing on 200 anonymized NCCT cases from multiple sites and scanner models affirmed its accuracy. Furthermore, a clinical Multi-Reader Multi-Case (MRMC) study involving 8 readers (including expert neuroradiologists and non-expert clinicians) and 200 cases demonstrated that clinicians using CINA-ASPECTS achieved improved agreement with ground truth for ASPECT scoring (average 4.1% more regions correct per scan), an increase in diagnostic accuracy (ROC AUC improved from 0.75 to 0.79), and a significant reduction in the mean time spent per case compared to unaided assessment.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the deep learning model in CINA-ASPECTS account for subtle variations in early ischemic changes that might be on the threshold of visual detection by human readers, and what is its sensitivity for these borderline cases?; The MRMC study showed an average improvement; what was the variability in performance improvement and reading time reduction across the different types of readers (e.g., expert neuroradiologists versus non-expert stroke neurologists or general radiologists)?; What specific image pre-processing steps, beyond reorientation and tilt-correction, are employed to standardize NCCT data from the different validated scanner vendors (Canon, GE, Philips, Siemens) to ensure robust deep learning analysis?; How does the software's performance, particularly its ASPECT score accuracy and the utility of the 'heat map', vary in patients imaged at the later end of the validated 12-hour window from 'time last known well' compared to those imaged much earlier?; Are there plans to expand the validation of CINA-ASPECTS to include patient populations under 21 years of age or to assess its utility in identifying neurological pathologies other than acute stroke, given its current limitations?