StrokeSENS ASPECTS Software Application

StrokeSENS ASPECTS is a computer-aided diagnosis software that analyzes non-contrast CT brain scans to automatically calculate an Alberta Stroke Program Early CT (ASPECTS) score. It is intended to assist clinicians in assessing and characterizing ischemic brain tissue injury in patients with known middle cerebral artery (MCA) or internal carotid artery (ICA) occlusion within 12 hours of symptom onset. The software highlights affected brain regions and provides an editable automated score for physician review.

Both the subject device and predicate use machine learning algorithms to segment ASPECTS regions, highlight affected areas, and generate an automated ASPECTS score from non-contrast head CT scans. Key difference: StrokeSENS uses deep learning (vs. predicate's random forest), is not limited to Siemens Somatom scanners (vs. predicate limitation), and extends the clinical window to 12 hours (vs. predicate's 6 hours). Both include gating conditions requiring confirmation of ICA/MCA occlusion, provide voxel-wise analysis heat maps, and allow manual editing of results.

ISO 13485:2016+A11:2021 (medical device quality management), IEC 62304:2006+A1:2015 (software lifecycle processes), IEC 62366:2015+A1:2020 (usability engineering), ISO 14971:2019+A11:2021 (risk management), and NEMA 3.1-3.20 (2021) for DICOM conformance. Software verification and validation followed FDA guidance on premarket submissions for software and cybersecurity in medical devices.

Both devices have identical intended use as CADx tools for ASPECTS scoring in acute ischemic stroke patients with ICA/MCA occlusion. Both employ substantially similar technological approaches: automated image registration to an atlas, segmentation of ASPECTS regions, machine learning-based identification of ischemic tissue, and presentation of results for clinician verification. Performance data from a 200-patient standalone study demonstrated 90.9% AUC-ROC, and a 100-patient MRMC clinical study showed statistically significant reader improvement of 5.7% in AUC when using StrokeSENS ASPECTS compared to unaided reading. The differences (deep learning vs. random forest algorithm, broader CT scanner compatibility, extended time window to 12 hours) represent product enhancements rather than fundamental changes to the device's mechanism or intended clinical application, and do not raise new questions of safety or effectiveness.