Abstract #3986

Prediction of Amyloid-β Deposition Using Multiple Regression Analysis of Quantitative Susceptibility Mapping

Ryota Sato¹, Kohsuke Kudo², Niki Udo³, Masaaki Matsushima⁴, Ichiro Yabe⁴, Akinori Yamaguchi², Makoto Sasaki⁵, Masafumi Harada⁶, Noriyuki Matsukawa⁷, Tomoki Amemiya¹, Yasuo Kawata¹, Yoshitaka Bito¹, Hisaaki Ochi¹, and Toru Shirai¹

¹Healthcare Business Unit, Hitachi, Ltd., Tokyo, Japan, ²Department of Diagnostic Imaging, Hokkaido University Graduate School of Medicine, Hokkaido, Japan, ³Department of Psychiatry, Hokkaido University Graduate School of Medicine, Hokkaido, Japan, ⁴Department of Neurology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Hokkaido, Japan, ⁵Institute for Biomedical Sciences, Iwate Medical University, Iwate, Japan, ⁶Department of Radiology, Tokushima University, Tokushima, Japan, ⁷Department of Neurology, Nagoya City University, Aichi, Japan

For early diagnosis of Alzheimer’s disease, we created and evaluated a prediction method of amyloid β deposition based on multiple regression analysis of quantitative susceptibility mapping. A multiple regression model to predict standard uptake values (SUVs) of amyloid PET was constructed based on susceptibilities in 47 brain regions with the constraint Aβ deposition and susceptibility being positively correlated. The correlation coefficients between true and predicted SUVs were increased by incorporating the constraint, and the area under the receiver operating characteristics curve to predict Aβ positivity was 70%. The results suggest that the model could predict Aβ positivity at moderate accuracy.

This abstract and the presentation materials are available to members only; a login is required.

Join Here