Vall d'Hebron, a pioneer in applying artificial intelligence to detect serious strokes

Endovascular thrombectomy is indicated in case of ischemic stroke caused by a large vessel occlusion. In its effectiveness, the speed of diagnosis is crucial, since the treatment has to be done in the first hours. Vall d'Hebron is a pioneer in applying artificial intelligence software, MethinksLVO, to patients suspected of suffering a stroke, which detects where the occlusion of the large vessel that caused it is located. According to a study published in the journal Stroke, carried out from the observation of 1,453 patients from three Catalan centers (Vall d'Hebron, Germans Trias i Pujol and Hospital Clínic), this software can predict with a degree of accuracy of 87% occlusions of large vessels in patients with suspected stroke. The percentage increases to 91% if patient clinical data is incorporated into the software.

MethinksLVO, which Vall d'Hebron has been using since May 2020, is based on a machine learning algorithm that distinguishes occlusion patterns of a large vessel without the need to make a Computed Axial Tomography (CAT) scan with contrast. "In the diagnosis for stroke imaging, time is crucial. Generally, we perform a cranial CAT without contrast and then a CAT with contrast to see where the arterial occlusion is", summarizes Dr. Marc Ribó, neurologist at the Stroke Unit: "This artificial intelligence software allows us to save the step of acquiring X-ray images with contrast, a technology that is not available in many hospitals where patients go first: artificial intelligence avoids unnecessary transfers of patients to larger centers, with the consequent delay in performing endovascular thrombectomy".

In hospitals like Vall d'Hebron, the diagnosis of stroke can be accelerated by around 15 minutes thanks to artificial intelligence, calculates Dr. Marc Ribó, but in smaller health centers they can save up to two hours in applying the treatment. "The interpretation of images in real time by artificial intelligence algorithms has great potential to favor geographic equity in access to rapid diagnosis and treatment in regional centers or developing countries", concludes Dr. Marc Ribó.