Classification of cardioembolic stroke based on a deep neural network using chest radiographs

• Published in: EBioMedicine 2021-07, Vol.69, p.103466-103466, Article 103466
• Main Authors: Jeong, Han-Gil, Kim, Beom Joon, Kim, Tackeun, Kang, Jihoon, Kim, Jun Yup, Kim, Joonghee, Kim, Joon-Tae, Park, Jong-Moo, Kim, Jae Guk, Hong, Jeong-Ho, Lee, Kyung Bok, Park, Tai Hwan, Kim, Dae-Hyun, Oh, Chang Wan, Han, Moon-Ku, Bae, Hee-Joon
• Format: Article
• Language: English
• Subjects: Aged; Aged, 80 and over; Cardioembolism; Chest radiograph; Classification; Deep Learning; Embolic Stroke - classification; Embolic Stroke - diagnostic imaging; Female; Humans; Male; Middle Aged; Radiographic Image Interpretation, Computer-Assisted - methods; Radiography, Thoracic - methods; Research paper; Stroke
• ISSN: 2352-3964; 2352-3964
• DOI: 10.1016/j.ebiom.2021.103466

Background: Although chest radiographs have not been utilised well for classifying stroke subtypes, they could provide a plethora of information on cardioembolic stroke. This study aimed to develop a deep convolutional neural network that could diagnose cardioembolic stroke based on chest radiographs. Methods: Overall, 4,064 chest radiographs of consecutive patients with acute ischaemic stroke were collected from a prospectively maintained stroke registry. Chest radiographs were randomly partitioned into training/validation (n = 3,255) and internal test (n = 809) datasets in an 8:2 ratio. A densely connected convolutional network (ASTRO-X) was trained to diagnose cardioembolic stroke based on chest radiographs. The performance of ASTRO-X was evaluated using the area under the receiver operating characteristic curve. Gradient-weighted class activation mapping was used to evaluate the region of focus of ASTRO-X. External testing was performed with 750 chest radiographs of patients with acute ischaemic stroke from 7 hospitals. Findings: The areas under the receiver operating characteristic curve of ASTRO-X were 0.86 (95% confidence interval [CI], 0.83–0.89) and 0.82 (95% CI, 0.79–0.85) during the internal and multicentre external testing, respectively. The gradient-weighted class activation map demonstrated that ASTRO-X was focused on the area where the left atrium was located. Compared with cases predicted as non-cardioembolism by ASTRO-X, cases predicted as cardioembolism by ASTRO-X had higher left atrial volume index and lower left ventricular ejection fraction in echocardiography. Interpretation: ASTRO-X, a deep neural network developed to diagnose cardioembolic stroke based on chest radiographs, demonstrated good classification performance and biological plausibility. Grant No. 14–2020–046 and 08–2016–051 from the Seoul National University Bundang Research Fund and NRF-2020M3E5D9079768 from the National Research Foundation of Korea.