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

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, 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

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 the correct treatment. Recently, 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) imaging has been used for the evaluation of pediatric epilepsy. However, the epileptic focus is very difficult to be identified by visual assessment since it may present either hypo- or hyper-metabolic abnormality with unclear boundary. This study aimed to develop a novel symmetricity-driven deep learning framework of PET imaging for the identification of epileptic foci in pediatric patients with temporal lobe epilepsy (TLE).

Methods: We retrospectively included 201 pediatric patients with TLE and 24 age-matched controls who underwent 18F-FDG PET-CT studies. 18F-FDG PET images were quantitatively investigated using 386 symmetricity features, and a pair-of-cube (PoC)-based Siamese convolutional neural network (CNN) was proposed for precise localization of epileptic focus, and then metabolic abnormality level of the predicted focus was calculated automatically by asymmetric index (AI). Performances of the proposed framework were compared with visual assessment, statistical parametric mapping (SPM) software, and Jensen-Shannon divergence-based logistic regression (JS-LR) analysis.

Results: The proposed deep learning framework could detect the epileptic foci accurately with the dice coefficient of 0.51, which was significantly higher than that of SPM (0.24, P < 0.01) and significantly (or marginally) higher than that of visual assessment (0.31-0.44, P = 0.005-0.27). The area under the curve (AUC) of the PoC classification was higher than that of the JS-LR (0.93 vs. 0.72). The metabolic level detection accuracy of the proposed method was significantly higher than that of visual assessment blinded or unblinded to clinical information (90% vs. 56% or 68%, P < 0.01).

Conclusion: The proposed deep learning framework for 18F-FDG PET imaging could identify epileptic foci accurately and efficiently, which might be applied as a computer-assisted approach for the future diagnosis of epilepsy patients.

Trial registration: NCT04169581. Registered November 13, 2019 Public site: https://ichgcp.net/clinical-trials-registry/NCT04169581.

Keywords: Deep learning; Epilepsy; Glucose metabolism; Pediatrics; Positron emission tomography (PET).

Conflict of interest statement

The authors declare that they have no conflicts of interest.

Figures

Fig. 1
Fig. 1
The proposed framework for epileptic focus detection
Fig. 2
Fig. 2
a Symmetricity feature heat map; b ROC curves of mono- and our selected features; and c ROC curves of visual assessment, SPM analysis, and the proposed method
Fig. 3
Fig. 3
a ROC curves of JS-LR and PoC-Siamese network; b dice coefficients obtained by the physicians with different levels, SPM analysis, and the proposed method
Fig. 4
Fig. 4
Two examples of the proposed epileptic focus localization of a hypometabolism and b hypermetabolism
Fig. 5
Fig. 5
Comparison of metabolic abnormality level determination accuracy between proposed method and physicians blinded to clinical information (a) and unblinded (b)

References

    1. Aaberg KM, Gunnes N, Bakken IJ, Søraas CL, Berntsen A, Magnus P, et al. Incidence and prevalence of childhood epilepsy: a nationwide cohort study. Pediatrics. 2017;139(5).
    1. Berg AT, Coryell J, Saneto RP, Grinspan ZM, Alexander JJ, Kekis M, et al. Early-life epilepsies and the emerging role of genetic testing. JAMA Pediatr. 2017;171(9):863–871. doi: 10.1001/jamapediatrics.2017.1743.
    1. Joo EY, Seo DW, Hong S-C, Hong SB. Functional neuroimaging findings in patients with lateral and mesio-lateral temporal lobe epilepsy; FDG-PET and ictal SPECT studies. J Neurol. 2015;262(5):1120–1129. doi: 10.1007/s00415-014-7625-z.
    1. Theodore WH. Presurgical focus localization in epilepsy: PET and SPECT. Semin Nucl Med. 2017;47(1):44–53. doi: 10.1053/j.semnuclmed.2016.09.008.
    1. Spencer SSJE. The relative contributions of MRI, SPECT, and PET imaging in epilepsy. Epilepsia. 1994;35:S72–S89. doi: 10.1111/j.1528-1157.1994.tb05990.x.
    1. Verger A, Lagarde S, Maillard L, Bartolomei F, Guedj E. Brain molecular imaging in pharmacoresistant focal epilepsy: Current practice and perspectives. Rev Neurol. 2018;174(1-2):16–27. doi: 10.1016/j.neurol.2017.05.001.
    1. Willmann O, Wennberg R, May T, Woermann F, Pohlmann-Eden BJS. The contribution of 18F-FDG PET in preoperative epilepsy surgery evaluation for patients with temporal lobe epilepsy: a meta-analysis. Seizure. 2007;16(6):509–520. doi: 10.1016/j.seizure.2007.04.001.
    1. Rheims S, Jung J, Ryvlin P. Combination of PET and magnetoencephalography in the presurgical assessment of MRI-negative epilepsy. Front Neurol. 2013;4:188. doi: 10.3389/fneur.2013.00188.
    1. Won HJ, Chang K-H, Cheon J-E, Kim HD, Lee DS, Han MH, et al. Comparison of MR imaging with PET and IctalSPECT in 118 patients with intractable epilepsy. Am J Neuroradiol. 1999;20(4):593–599.
    1. Ohta Y, Nariai T, Ishii K, Ishiwata K, Mishina M, Senda M, et al. Voxel-and ROI-based statistical analyses of PET parameters for guidance in the surgical treatment of intractable mesial temporal lobe epilepsy. Ann Nucl Med. 2008;22(6):495. doi: 10.1007/s12149-008-0140-5.
    1. Alkonyi B, Chugani HT, Juhász CJE. Transient focal cortical increase of interictal glucose metabolism in Sturge-Weber syndrome: Implications for epileptogenesis. Epilepsia. 2011;52(7):1265–1272. doi: 10.1111/j.1528-1167.2011.03066.x.
    1. Muzik O, Chugani DC, Shen C, Da Silva EA, Shah J, Shah A, et al. Objective method for localization of cortical asymmetries using positron emission tomography to aid surgical resection of epileptic foci. Comput Aided Surg. 1998;3(2):74–82. doi: 10.3109/10929089809148132.
    1. Archambaud F, Bouilleret V, Hertz-Pannier L, Chaumet-Riffaud P, Rodrigo S, Dulac O, et al. Optimizing statistical parametric mapping analysis of 18 F-FDG PET in children. EJNMMI Res. 2013;3(1):2. doi: 10.1186/2191-219X-3-2.
    1. Kim YK, Lee DS, Lee SK, Chung CK, Chung J-K, Lee MC. 18F-FDG PET in localization of frontal lobe epilepsy: comparison of visual and SPM analysis. J Nucl Med. 2002;43(9):1167–1174.
    1. Kumar A, Juhász C, Asano E, Sood S, Muzik O. Chugani HTJJoNM. Objective detection of epileptic foci by 18F-FDG PET in children undergoing epilepsy surgery. J Nucl Med. 2010;51(12):1901–1907. doi: 10.2967/jnumed.110.075390.
    1. Scarpazza C, Tognin S, Frisciata S, Sartori G, Mechelli AJN, Reviews B. False positive rates in voxel-based morphometry studies of the human brain: should we be worried? Neurosci Biobehav Rev. 2015;52:49–55. doi: 10.1016/j.neubiorev.2015.02.008.
    1. Gillies RJ, Kinahan PE, Hricak HJR. Radiomics: images are more than pictures, they are data. Radiology. 2016;278(2):563–577. doi: 10.1148/radiol.2015151169.
    1. Ardila D, Kiraly AP, Bharadwaj S, Choi B, Reicher JJ, Peng L, et al. End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nat Med. 2019;25(6):954–961. doi: 10.1038/s41591-019-0447-x.
    1. Ding Y, Sohn JH, Kawczynski MG, Trivedi H, Harnish R, Jenkins NW, et al. A deep learning model to predict a diagnosis of Alzheimer disease by using 18F-FDG PET of the brain. Radiology. 2019;290(2):456–464. doi: 10.1148/radiol.2018180958.
    1. Zeng L-L, Wang H, Hu P, Yang B, Pu W, Shen H, et al. Multi-site diagnostic classification of schizophrenia using discriminant deep learning with functional connectivity MRI. EBioMedicine. 2018;30:74–85. doi: 10.1016/j.ebiom.2018.03.017.
    1. Avants BB, Epstein CL, Grossman M. Gee JCJMia. Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain. Med Image Anal. 2008;12(1):26–41. doi: 10.1016/j.media.2007.06.004.
    1. Zhang Q, Xiao Y, Suo J, Shi J, Yu J, Guo Y, et al. Sonoelastomics for breast tumor classification: a radiomics approach with clustering-based feature selection on sonoelastography. Ultrasound Med Biol. 2017;43(5):1058–1069. doi: 10.1016/j.ultrasmedbio.2016.12.016.
    1. Shen X, Finn ES, Scheinost D, Rosenberg MD, Chun MM, Papademetris X, et al. Using connectome-based predictive modeling to predict individual behavior from brain connectivity. Nat Protoc. 2017;12(3):506–518. doi: 10.1038/nprot.2016.178.
    1. Koch G, Zemel R, Salakhutdinov R. Siamese neural networks for one-shot image recognition. ICML Deep Learn Work. 2015;2.
    1. Zhu Y, Feng J, Wu S, Hou H, Ji J, Zhang K, et al. Glucose metabolic profile by visual assessment combined with statistical parametric mapping analysis in pediatric patients with epilepsy. J Nucl Med. 2017;58(8):1293–1299. doi: 10.2967/jnumed.116.187492.
    1. Chen S, Ma K, Zheng YJ. Med3d: Transfer learning for 3d medical image analysis. arXiv preprint arXiv:190400625. 2019.
    1. Didelot A, Mauguière F, Redouté J, Bouvard S, Lothe A, Reilhac A, et al. Voxel-based analysis of asymmetry index maps increases the specificity of 18F-MPPF PET abnormalities for localizing the epileptogenic zone in temporal lobe epilepsies. J Nucl Med. 2010;51(11):1732–1739. doi: 10.2967/jnumed.109.070938.
    1. De Blasi B, Barnes A, Galazzo IB, Hua C-h, Shulkin B, Koepp M, et al. Age-Specific 18F-FDG Image processing pipelines and analysis are essential for individual mapping of seizure foci in pediatric patients with intractable epilepsy. J Nucl Med. 2018;59(10):1590–1596. doi: 10.2967/jnumed.117.203950.
    1. Liao W, Zhang Z, Pan Z, Mantini D, Ding J, Duan X, et al. Default mode network abnormalities in mesial temporal lobe epilepsy: a study combining fMRI and DTI. Hum Brain Mapp. 2011;32(6):883–895. doi: 10.1002/hbm.21076.
    1. van Diessen E, Diederen SJ, Braun KP, Jansen FE, Stam CJJE. Functional and structural brain networks in epilepsy: what have we learned? Epilepsia. 2013;54(11):1855–1865. doi: 10.1111/epi.12350.
    1. Bonilha L, Kobayashi E, Castellano G, Coelho G, Tinois E, Cendes F, et al. Texture analysis of hippocampal sclerosis. Epilepsia. 2003;44(12):1546–1550. doi: 10.1111/j.0013-9580.2003.27103.x.
    1. Bettus G, Guedj E, Joyeux F, Confort-Gouny S, Soulier E, Laguitton V, et al. Decreased basal fMRI functional connectivity in epileptogenic networks and contralateral compensatory mechanisms. Hum Brain Mapp. 2009;30(5):1580–1591. doi: 10.1002/hbm.20625.
    1. Chassoux F, Artiges E, Semah F, Desarnaud S, Laurent A, Landre E, et al. Determinants of brain metabolism changes in mesial temporal lobe epilepsy. Epilepsia. 2016;57(6):907–919. doi: 10.1111/epi.13377.
    1. de Oliveira MS, Betting LE, Mory SB, Cendes F, Castellano GJE. Behavior. Texture analysis of magnetic resonance images of patients with juvenile myoclonic epilepsy. Epilepsy Behav. 2013;27(1):22–28. doi: 10.1016/j.yebeh.2012.12.009.
    1. del Toro OAJ, Foncubierta-Rodriguez A, Gómez MIV, Müller H, Depeursinge A. Epileptogenic lesion quantification in MRI using contralateral 3D texture comparisons. MICCAI. 2013:353–60.
    1. Focke NK, Yogarajah M, Symms MR, Gruber O, Paulus W, Duncan JSJN. Automated MR image classification in temporal lobe epilepsy. Neuroimage. 2012;59(1):356–362. doi: 10.1016/j.neuroimage.2011.07.068.
    1. Cantor-Rivera D, Khan AR, Goubran M, Mirsattari SM, Peters TM, Graphics Detection of temporal lobe epilepsy using support vector machines in multi-parametric quantitative MR imaging. Comput Med Imaging Graph. 2015;41:14–28. doi: 10.1016/j.compmedimag.2014.07.002.
    1. Chen S, Zhang J, Ruan X, Deng K, Zhang J, Zou D, et al. Voxel-based morphometry analysis and machine learning based classification in pediatric mesial temporal lobe epilepsy with hippocampal sclerosis. Brain Imaging Behav. 2019:1–10.
    1. Wook KD, Kun LS, Hye-Jin M, Ki-Young J, Kon C, Chun-Ki CJJN. Surgical treatment of nonlesional neocortical epilepsy long-term longitudinal study. JAMA Neurol. 2017;74(3):1–8.

Source: PubMed

赞助者和合作者

医疗条件

药物干预

3
订阅