Keywords: CT Perfusion, Transformers, Ischemic stroke, Infarct core segmentation.
TL;DR: We propose the use of transformers to encode sequential CT perfusion scans in a spatial attention map, which we subsequently use to perform infarct core segmentation for ischemic stroke patients.
Abstract: The infarct core size is a crucial biomarker for treatment selection for ischemic stroke patients. For this purpose, we present a novel approach to perform infarct core segmentation using CT perfusion (CTP) source data, without ordinary deconvolution analysis. We propose the use of transformers to encode sequential CTP scans in spatial attention maps, which we subsequently use for infarct core segmentation. We report new top results on the ISLES 2018 challenge test data set for infarct core segmentation. This work presents a primary benchmark for infarct core segmentation from CTP source data using transformers.
Paper Type: both
Primary Subject Area: Segmentation
Secondary Subject Area: Application: Radiology
Paper Status: original work, not submitted yet
Source Code Url: I am not able to share the source code, because the rights are at Amsterdam UMC, as the software might be used in-house.
Data Set Url: http://www.isles-challenge.org
Registration: I acknowledge that publication of this at MIDL and in the proceedings requires at least one of the authors to register and present the work during the conference.
Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.