EMVision Medical Devices’ interim trial data shows promising results in identifying stroke types

Go to Imelda Cotton author's page
By Imelda Cotton - 
EM Vision ASX EMV Stage 2 AI algorithm stroke detection

Stage 2 interim data from a pre-validation trial by EMVision Medical Devices (ASX: EMV) has confirmed the positive performance of neurodiagnostic artificial intelligence (AI) algorithms in the diagnosis of suspected haemorrhagic or ischaemic strokes.

The data showed the algorithms had an “encouraging ability” to identify patterns and features across complex ischaemic patient data sets, including early onset hyperacute ischaemic strokes that can be challenging to detect on non-contrast computed tomography (CT).

The trial involved 180 patients who had presented with stroke-like symptoms to emergency departments at Royal Melbourne Hospital, Liverpool Hospital in Sydney and Brisbane’s Princess Alexandra Hospital.

75 of those patients had confirmed ischaemic strokes and 105 had non-ischaemic events including 18 haemorrhagic strokes (caused by bleeding due to a ruptured blood vessel), 67 stroke mimics and 20 transient ischaemic attacks.

Brain damage score

Patients were subjected to the Alberta Stroke Program Early CT Score (ASPECTS), which measures the extent of brain damage from ischaemic strokes by quantifying early ischaemic changes detected on CT scans.

The ASPECTS score ranges from 0 to 10, with lower scores indicating that more regions of the brain are affected, resulting in more severe brain damage.

Scores below 7 are associated with worse functional outcomes for patients.

In Stage 2 of the current trial, the average ASPECTS score for the ischaemic stroke cohort was 7.4, representing a diverse range of stroke cases enrolled.

Pattern identification

The Stage 2 data showed EMVision’s AI model could identify patterns and features in the signals across complex ischaemic patient data sets, including cases of very early onset hyperacute ischaemic stroke where there are minimal (or zero) non-diagnostic radiological findings on a traditional CT scan.

The neurodiagnostic capabilities were reported to demonstrate the potential of the company’s technology to significantly improve diagnosis, care and outcomes for haemorrhagic and ischaemic stroke patients.

The data is now being used to enhance EMVision’s AI algorithms.

Cross-validation analysis

Cross-validation interim analysis has been undertaken to prepare for an upcoming validation (sensitivity/specificity confirmation) trial.

The detection and classification performance of the AI algorithm will then be estimated by observing clustering which occurs in the cross-validation data, with the ideal cluster pattern being a distinct separation between ischaemic and non-ischaemic cases.

Promising development

Chief executive officer Scott Kirkland said the interim data was a promising development.

“The ability of our technology to detect hyperacute and acute ischaemic cases in this cross-validation interim analysis is incredibly exciting for our team and our clinical collaborators,” he said.

“Most importantly, [we are excited about] what this may mean for the improvement of care pathways and outcomes for future stroke patients.”

The ability to indicate likely haemorrhagic or ischaemic strokes in an accessible and easy-to-use technology is not generally available to patients.

“As our clinical trial and development progresses, we look forward to seeing how this device will expand our neurodiagnostic capabilities,” he said.

Patient recruitment for the final Stage 3 of the pre-validation clinical trial is on track for completion in the coming months, with over half the target cohort (up to 30 haemorrhages) recruited to date.