sablier petit

Fast prediction of infarct growth

from diffusion weighted MRI imaging during acute stroke


The core methodology of NEURiNFARCT consists of the prediction of infarct growth from MR-DWI data acquired during the acute stage of stroke.Our basic research has demonstrated that the non invasive prediction in the acute stage of the final growth of the ischemic lesion would reach very good agreement with retrospective measures.

NEURiNFARCT is a software prototype developed in our group under Matlab. The software workflow was designed following to clinical standards and is highly interactive its graphical user interface.
The workflow consists of 3 consecutive steps:
1) Segmentation of the initial lesion using a large palette of automatic/manually-assisted tools.
2) Registration of the initial mask of the initial lesion to ADC maps.
3) Prediction of final lesion using a region-growing infarct model.
Growth prediction follows an iterative process which is updated in real time and therefore, can be controlled by the clinical expert. The final characteristics of the ischemic lesion (location, volume, ADC statistics, etc.) are summarized in a report sheet which includes visualization of the estimate of the final lesion using 2D image stacks, or even 3D renderings.
See video of prediction


NEURiNFARCT has been developed as a software prototype and is now available for collaborative evaluation and testing at a larger scale. The usage of the tool has been documented and can be proficiently comprehended by clinical neurologists and radiologists with minimum training. Please contact us is you are interested in trying out NEURiNFARCT!