Nuclear Techniques ›› 2020, Vol. 43 ›› Issue (5): 50301-050301.doi: 10.11889/j.0253-3219.2020.hjs.43.050301


A comparative study of PET-MRI brain quantitative accuracy: the effect of MRI based segmentation and PET based segmentation on SUVR calculation

Zaisheng LI1,2,3,Shuangshuang SONG5,6,Tianyi ZENG1,3,Jie LU5,6,7,Lingzhi HU4,Qun CHEN2,4()   

  1. 1.Shanghai Advanced Research Institute, Chinese Academy of Sciences, Shanghai 201210, China
    2.School of Information Science and Technology, ShanghaiTech University, Shanghai 201210, China
    3.University of Chinese Academy of Sciences, Beijing 100049, China
    4.Shanghai United Imaging Healthcare Co. , Ltd, Shanghai 201807, China
    5.Department of Radiology, Xuanwu Hospital, Capital Medical University, Beijing 100053, China
    6.Beijing Key Laboratory of Magnetic Resonance Imaging and Brain Informatics, Beijing 100053, China
    7.Department of Nuclear Medicine, Xuanwu Hospital, Capital Medical University, Beijing 100053, China
  • Received:2020-02-06 Revised:2020-03-04 Online:2020-05-15 Published:2020-05-07
  • Contact: Qun CHEN
  • About author:LI Zaisheng, male, born in 1994, graduated from Nanjing University of Aeronautics and Astronautics in 2016, master student, focusing on medical imaging processing
  • Supported by:
    National Key Research and Development Program Digital Diagnostic Equipment R&D Pilot(2016YFC0103900)

Abstract: Background

In clinical application and scientific research of positron emission tomography (PET), the standard uptake value (SUV) is commonly used for disease assessment. In order to reduce the deviations of SUV among clinical PET images, standard uptake value ratio (SUVR) based on a specific reference region is substituted for SUV. Accurate SUVR is hardly achieved directly from PET images due to the difficulty of accurately segmenting reference regions from PET images with low resolution and high noise. The integrated positron emission tomography-magnetic resonance imaging (PET-MRI) can simultaneously acquire MRI images and PET images, hence the combination of MRI image and PET image is expected to improve the accuracy of SUVR.


This paper study aims to investigate the effect of MRI segmentation and PET segmentation on SURV calculation to improve the PET-MRI brain quantitative accuracy.


First of all, integrated PET-MRI was used to acquire MRI and PET images of 12-age-controlled healthy volunteers simultaneously. Then the MRI images and PET images were segmented into reference regions by the atlas-based method. Based on reference regions derived from MRI and PET respectively, the SUVRs of left / right cerebrum, cerebral cortex, cerebral white matter, frontal lobe, parietal lobe, temporal lobe and occipital lobe were calculated. For each group of SUVRs, the coefficient of variation (CV) of SUVRs was calculated to measure consistency level. Finally, quantitative analysis was performed to evaluate which segmentation method offered better SUVR consistency among volunteers.


CV of SUVRs based on cerebellar gray matter from MRI images and PET images were 0.020/0.021, 0.023/0.026, 0.031/0.028, 0.028/0.031, 0.033/0.036, 0.022/0.024, 0.044/0.045 and 0.052/0.055, 0.054/0.055, 0.058/0.063, 0.053/0.058, 0.063/0.065, 0.048/0.050, 0.059/0.065 respectively. When the cerebellum was chosen as reference region, MRI results and PET results were 0.017/0.019, 0.021/0.025, 0.029/0.025, 0.026/0.030, 0.032/0.036, 0.020/0.022, 0.044/0.045 and 0.046/0.050,0.049/0.051, 0.052/0.057, 0.048/0.054, 0.059/0.061, 0.044/0.045, 0.056/0.062 respectively.


Compared with SUVRs based on reference region segmented from PET images, SUVRs based on reference region segmented from MRI images have better consistency with smaller coefficient of variation.

Key words: SUVR, Quantification, PET-MRI, 18F-FDG

CLC Number: 

  • R445,TL99