“AI for better workflow, better outcomes in medical imaging answering unmet medical needs”
The AI4CMR solution is ”software as a medical device” (SaMD) that aims to automate the interpretation of cardiac magnetic resonance imaging (CMR) using artificial intelligence.
The AI4CMR software (SaMD) performs a fully automatic cardiac segmentation allowing instant quantification of ventricular volumes, myocardial mass, and ejection fraction. Besides, it automatically identifies cases with normal left ventricle wall motion (regional contractility).
This safety net solution combined saves significant time and allows the physician to dedicate more time to complex or abnormal cases.
How it works
CMR Reporting is Time Consuming and Subjective
CMR reporting is based on two pillars:
-Fully automatic cardiac segmentation for volume quantification
(zero human interaction)
-Visual interpretation of different sequences.
Significantly decreases CMR reporting time by more than 50%, allowing more reports.
Fully automatic segmentation
Fully automatic and reliable segmentation using state of the art algorithms, with high overlap with experts manual segmentation (Dice coefficient >= 0.8).
Fully automatic quantification
Fully automatic left and right ventricles volumes quantification: End-diastolic volume (EDV); End-systolic volume (ESV); Stroke volume (SV); Ejection fraction (EF); Myocardial mass (LVM); Cardiac output (CO); Respective BSA indexed volumes. Excellent reliability between AI4CMR and experts.
Automatic identification of cases with normal vs suspicious wall motion (100% Sensitivity, 77% Accuracy, 100% Negative Predictive Value, 70% Positive Predictive Value and 50% Specificity).
Significant reduction of the cine reporting time: around 10 times decrease on the segmentation time (average response time of 160 seconds) when compared with manual segmentation (20 minutes).
MRI Scanner Agnostic
Compatible with different MRI manufacturers: Siemens; GE Healthcare Systems; Phillips Medical Systems.
Cloud-based service: high availability, scalability and reliability.
Easy communication with other devices via a standard protocol.
Better Clinical Workflow
Improves clinical workflow when integrated with PACS or VNAs Viewers (no need for a dedicated viewer software).
What experts say
"Cardiac Magnetic Resonance imaging (CMR) is the gold-standard with respect to assessment of cardiac anatomy, pathology and function. However, currently available software is cumbersome to use, time-consuming and insufficiently accurate. Although the introduction of artificial intelligence (AI) / machine learning (ML) algorithms has improved performance, available solutions are still under par. The AI4CMR software provides a cloud-based AI-solution for cardiac MRI image analysis with unsurpassed speed, automaticity, and accuracy. At this point fully automated analysis of left ventricular function (including LV volume and mass) is warranted within a timeframe of a few minutes (LV within 49 seconds, in combination with RV function within 2 minutes), with a sensitivity in excess of 99%. Subsequently, results undergo automated triage with respect to normality (0% False Negatives). Studies already have shown an excellent inter-rater reliability. Further research and training of the AI algorithm is needed to increase specificity and further expand anatomic segmentation, functional analysis, perfusion analysis and late-gadolinium enhancement (LGE), will enable fully automated computer-aided diagnosis following triage. Additional features will include automated LGE quantification and classification, myocardial strain-analysis, and valvular-/vascular flow assessment."