Summary
International Workshop on Smart Info-Media Systems in Asia
2022
Session Number:SS1
Session:
Number:SS1-2
Automatic Thrombus Localization on Patient with Ischemic Stroke on Non-contrast CT Using MCA Location Filter
Donlawit Beesomboon, Natsuda Kaothanthong, Tanapon Chansumpao, Dittapong Songsaeng,
pp.5-10
Publication Date:2022/9/15
Online ISSN:2188-5079
Summary:
Acute Ischemic Stroke (AIS) is commonly caused by thrombus in the middle cerebral artery (MCA). The localization of the thrombus on non-contrast computerized tomography (ncCT) images of an AIS patient is subtle. Additional imaging such as a computed tomography angiography (CTA) is required to locate the thrombus, which results in a delay of the treatment and affects the damage of patient's brain. This work presents an algorithm to locate an MCA region to improve the computation time of the thrombus segmentation of our previous work. Given a brain region of the input ncCT image, an MCA is found by a summation of pixels' weight in the interested region. The MCA is used as a reference for computing thrombus candidates using k-mean clustering. The located thrombus is selected from the candidate regions that are closer to the MCA region. Experiment is conducted on 1154 ncCT images of 10 patients. The result shows that the proposed thrombus localization algorithm is able to locate the thrombus with accuracy 99.93%, precision 1.76%, sensitivity 62.54% and specificity 99.93%. The computation time is averagely improved from 2536.18 second per patient to 305.58 seconds.