Improved interpretation of 18F-florzolotau PET in progressive supranuclear palsy using a normalization-free deep-learning classifier

Jiaying Lu; Christoph Clement; Jimin Hong; Min Wang; Xinyi Li; Lara Cavinato; Tzu-Chen Yen; Fangyang Jiao; Ping Wu; Jianjun Wu; Jingjie Ge; Yimin Sun; Matthias Brendel; Leonor Lopes; Axel Rominger; Jian Wang; Fengtao Liu; Chuantao Zuo; Yihui Guan; Qianhua Zhao; Kuangyu Shi; Progressive Supranuclear Palsy Neuroimage Initiative (PSPNI)

doi:10.1016/j.isci.2023.107426

Improved interpretation of ¹⁸F-florzolotau PET in progressive supranuclear palsy using a normalization-free deep-learning classifier

iScience. 2023 Jul 20;26(8):107426. doi: 10.1016/j.isci.2023.107426. eCollection 2023 Aug 18.

Authors

Jiaying Lu^{1

2}, Christoph Clement², Jimin Hong², Min Wang^{3

4}, Xinyi Li⁵, Lara Cavinato^{2

6}, Tzu-Chen Yen⁷, Fangyang Jiao¹, Ping Wu¹, Jianjun Wu⁵, Jingjie Ge¹, Yimin Sun⁵, Matthias Brendel⁸, Leonor Lopes², Axel Rominger², Jian Wang⁵, Fengtao Liu⁵, Chuantao Zuo^{1

9}, Yihui Guan¹, Qianhua Zhao⁵, Kuangyu Shi^{2

4}; Progressive Supranuclear Palsy Neuroimage Initiative (PSPNI)

Affiliations

¹ Department of Nuclear Medicine & PET Center & National Center for Neurological Disorders & National Clinical Research Center for Aging and Medicine, Huashan Hospital, Fudan University, Shanghai 200235, China.
² Department of Nuclear Medicine, Inselspital, Bern University Hospital, University of Bern, 3010 Bern, Switzerland.
³ Institute of Biomedical Engineering, School of Life Science, Shanghai University, Shanghai 200444, China.
⁴ Department of Informatics, Technical University of Munich, 80333 Munich, Germany.
⁵ Department of Neurology & National Center for Neurological Disorders & National Clinical Research Center for Aging and Medicine, Huashan Hospital, Fudan University, Shanghai 200400, China.
⁶ MOX - Modeling and Scientific Computing, Department of Mathematics, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milan, Italy.
⁷ APRINOIA Therapeutics Co., Ltd, Suzhou 215122, China.
⁸ Department of Nuclear Medicine, University of Munich, 80539 Munich, Germany.
⁹ Human Phenome Institute, Fudan University, Shanghai 200433, China.

Abstract

While ¹⁸F-florzolotau tau PET is an emerging biomarker for progressive supranuclear palsy (PSP), its interpretation has been hindered by a lack of consensus on visual reading and potential biases in conventional semi-quantitative analysis. As clinical manifestations and regions of elevated ¹⁸F-florzolotau binding are highly overlapping in PSP and the Parkinsonian type of multiple system atrophy (MSA-P), developing a reliable discriminative classifier for ¹⁸F-florzolotau PET is urgently needed. Herein, we developed a normalization-free deep-learning (NFDL) model for ¹⁸F-florzolotau PET, which achieved significantly higher accuracy for both PSP and MSA-P compared to semi-quantitative classifiers. Regions driving the NFDL classifier's decision were consistent with disease-specific topographies. NFDL-guided radiomic features correlated with clinical severity of PSP. This suggests that the NFDL model has the potential for early and accurate differentiation of atypical parkinsonism and that it can be applied in various scenarios due to not requiring subjective interpretation, MR-dependent, and reference-based preprocessing.

Keywords: Clinical neuroscience; Health informatics; Medical imaging.