Interpretation of SPECT Wall Motion with Deep Learning

Yangmei Zhang; Emma Bos; Owen Clarkin; Tyler Wilson; Gary Small; Glenn Wells; Lijun Lu; Benjamin J W Chow

doi:10.1016/j.nuclcard.2024.101881

Interpretation of SPECT Wall Motion with Deep Learning

J Nucl Cardiol. 2024 May 7:101881. doi: 10.1016/j.nuclcard.2024.101881. Online ahead of print.

Authors

Yangmei Zhang¹, Emma Bos², Owen Clarkin³, Tyler Wilson⁴, Gary Small³, Glenn Wells³, Lijun Lu⁵, Benjamin J W Chow⁶

Affiliations

¹ School of Biomedical Engineering, Southern Medical University, Guangzhou, China; Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, China.
² Department of Physics, Engineering Physics & Astronomy, Queen's University, Canada.
³ Department of Medicine (Cardiology), University of Ottawa Heart Institute, Canada.
⁴ Department of Applied Science in Computer Engineering, Queen's University, Canada.
⁵ School of Biomedical Engineering, Southern Medical University, Guangzhou, China; Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou, China; Pazhou Lab, Guangzhou, China.
⁶ Department of Medicine (Cardiology), University of Ottawa Heart Institute, Canada; Department of Radiology, University of Ottawa, Canada. Electronic address: bchow@ottawaheart.ca.

PMID: 38723886
DOI: 10.1016/j.nuclcard.2024.101881

Abstract

Objectives: We sought to develop a novel deep learning (DL) workflow to interpret single-photon emission computed tomography (SPECT) wall motion.

Background: Wall motion assessment with SPECT is limited by image temporal and spatial resolution. Visual interpretation of wall motion can be subjective and prone to error. Artificial intelligence (AI) may improve accuracy of wall motion assessment.

Methods: A total of 1,038 patients undergoing rest electrocardiogram (ECG)-gated SPECT and echocardiography were included. Using echocardiography as truth, a DL-model (DL-model 1) was trained to predict the probability of abnormal wall motion. Of the 1,038 patients, 317 were used to train a DL-model (DL-model 2) to assess regional wall motion. A 10-fold cross validation was adopted. Diagnostic performance of DL was compared with human readers and quantitative parameter.

Results: The area under the receiver operating characteristic curve (AUC) and accuracy (ACC) of DL model (AUC: 0.82 [95% CI: 0.79-0.85]; ACC: 0.88) were higher than human (AUC: 0.77 [95% CI: 0.73-0.81]; ACC:0.82; P < 0.001) and quantitative parameter (AUC: 0.74 [95% CI: 0.66-0.81]; ACC: 0.78; p< 0.05). The net reclassification index (NRI) was 7.7%. The AUC and accuracy of DL model for per-segment and per-vessel territory diagnosis were also higher than human reader. The DL model generated results within 30s with operable guided user interface (GUI) and therefore could serve preliminary interpretation in areas without sufficient qualified cardiologists.

Conclusions: DL can be used to improve interpretation of rest SPECT wall motion as compared with current human readers and quantitative parameter diagnosis.

Keywords: Deep learning; artificial intelligence; echocardiography; gated SPECT myocardial perfusion imaging; regional wall motion abnormality.