Page 84 - Contributed Paper Session (CPS) - Volume 4

CPS2131 Philip L.H. Yu et al.

Deep learning the mca dot sign of acute ischemic
stroke on non-contrast ct images
Jia You, Philip L.H. Yu
Department of Statistics and Actuarial Science, The University of Hong Kong

Abstract
The hyperdense middle cerebral artery (MCA) dot sign has been reported as
an important factor in the diagnosis of acute ischemic stroke due to large
vessel occlusion. Interpreting the initial CT brain scan in these patients requires
high level of expertise and has high inter-observer variability. An automated
computerized interpretation of the urgent CT brain image, with an emphasis
to pick up early signs of ischemic stroke will facilitate early patient diagnosis,
triage, and shorten the door-to-revascularization time for these group of
patients. In this paper, we present an automated detection method of
segmenting the MCA dot sign on non-contrast CT brain image scans based
on powerful deep learning technique.

Keywords
Deep learning, Segmentation; Medical imaging; Hyperdense middle cerebral
artery dot sign; Acute stroke.

1. Introduction
Acute ischemic stroke (AIS) has becoming a leading cause of morbidity and
mortality worldwide and recent advances in endovascular thrombectomy
(EVT) for treatment of AIS caused by large vessel occlusion (LVO) have been
widely accepted around the world (Powers et al., 2018; Malhotra & Liebeskind,
2015). The hyperdense middle cerebral artery (MCA) dot sign has been
reported as an important factor in the diagnosis of acute ischemia, especially
in LVO cases (Lim et al., 2018). Fast diagnosis and localization of MCA sign can
largely save patients’ rescue time, thus lower the probability of severe effect.
However, it is fairly challenge to detect MCA sign due to the subtlety of the
pathological intensity changes and low signal to noise ratio (Fig. 1). Available
data on large vessel occlusion stroke is based on western populations and the
respective incidence in Asian countries is largely unknown. So far, this study is
the first application of deep learning with specified interest to the hyperdense
MCA sign.
We adopted deep learning model as a feature extractor in this study, as
well. Recent years saw the availability of large amounts of annotated training
sets and the accessibility of affordable parallel computing resources via
Graphics Processing Units (or GPUs) have made it feasible to train deep neural

73 | I S I W S C 2 0 1 9