A deep learning framework for 18F-FDG PET imaging diagnosis in pediatric patients with temporal lobe epilepsy

• DOI: 10.1007/s00259-020-05108-y
• Publication Date: 2021-01-09T10:54:07Z
• Authors: Qinming Zhang; Yi Liao; Xiawan Wang; Teng Zhang; Jianhua Feng; Jianing Deng; Kexin Shi; Lin Chen; Liu Feng; Mindi Ma; Le Xue; Haifeng Hou; Xiaofeng Dou; Congcong Yu; Lei Ren; Yao Ding; Yufei Chen; Shuang Wu; Zexin Chen; Hong Zhang; Cheng Zhuo; Mei Tian
• Keywords: Abnormality; Statistical parametric mapping; Pediatric epilepsy

Abstract Purpose Epilepsy is one of the most disabling neurological disorders, which affects all age groups and often results in severe consequences. Since misdiagnoses are common, many pediatric patients fail to receive correct treatment. Recently, 18 F-fluorodeoxyglucose positron emission tomography ( F-FDG PET) imaging has been used for evaluation epilepsy. However, epileptic focus very difficult be identified by visual assessment since it may present either hypo- or hyper-metabolic abnormality with unclear boundary. This study aimed develop a novel symmetricity-driven deep learning framework PET identification foci temporal lobe epilepsy (TLE). Methods We retrospectively included 201 TLE 24 age-matched controls who underwent PET-CT studies. images were quantitatively investigated using 386 symmetricity features, pair-of-cube (PoC)-based Siamese convolutional neural network (CNN) was proposed precise localization focus, then metabolic level predicted calculated automatically asymmetric index (AI). Performances compared assessment, statistical parametric mapping (SPM) software, Jensen-Shannon divergence-based logistic regression (JS-LR) analysis. Results The could detect accurately dice coefficient 0.51, significantly higher than that SPM (0.24, P < 0.01) (or marginally) (0.31–0.44, = 0.005–0.27). area under curve (AUC) PoC classification JS-LR (0.93 vs. 0.72). detection accuracy method blinded unblinded clinical information (90% 56% 68%, 0.01). Conclusion identify efficiently, might applied as computer-assisted approach future diagnosis patients. Trial registration NCT04169581. Registered November 13, 2019 Public site: https://clinicaltrials.gov/ct2/show/NCT04169581