Automated PET-only quantification of amyloid deposition with adaptive template and empirically pre-defined ROI

G. Akamatsu, Y. Ikari, A. Ohnishi, H. Nishida, K. Aita, M. Sasaki, Y. Yamamoto, M. Sasaki, M. Senda

Research output: Contribution to journalArticlepeer-review

13 Citations (Scopus)

Abstract

Amyloid PET is useful for early and/or differential diagnosis of Alzheimer's disease (AD). Quantification of amyloid deposition using PET has been employed to improve diagnosis and to monitor AD therapy, particularly in research. Although MRI is often used for segmentation of gray matter and for spatial normalization into standard Montreal Neurological Institute (MNI) space where region-of-interest (ROI) template is defined, 3D MRI is not always available in clinical practice. The purpose of this study was to examine the feasibility of PET-only amyloid quantification with an adaptive template and a pre-defined standard ROI template that has been empirically generated from typical cases. A total of 68 subjects who underwent brain 11C-PiB PET were examined. The 11C-PiB images were non-linearly spatially normalized to the standard MNI T1 atlas using the same transformation parameters of MRI-based normalization. The automatic-anatomical-labeling-ROI (AAL-ROI) template was applied to the PET images. All voxel values were normalized by the mean value of cerebellar cortex to generate the SUVR-scaled images. Eleven typical positive images and eight typical negative images were normalized and averaged, respectively, and were used as the positive and negative template. Positive and negative masks which consist of voxels with SUVR 1.7 were extracted from both templates. Empirical PiB-prone ROI (EPP-ROI) was generated by subtracting the negative mask from the positive mask. The 11C-PiB image of each subject was non-rigidly normalized to the positive and negative template, respectively, and the one with higher cross-correlation was adopted. The EPP-ROI was then inversely transformed to individual PET images. We evaluated differences of SUVR between standard MRI-based method and PET-only method. We additionally evaluated whether the PET-only method would correctly categorize 11C-PiB scans as positive or negative. Significant correlation was observed between the SUVRs obtained with AAL-ROI and those with EPP-ROI when MRI-based normalization was used, the latter providing higher SUVR. When EPP-ROI was used, MRI-based method and PET-only method provided almost identical SUVR. All 11C-PiB scans were correctly categorized into positive and negative using a cutoff value of 1.7 as compared to visual interpretation. The 11C-PiB SUVR were 2.30 ± 0.24 and 1.25 ± 0.11 for the positive and negative images. PET-only amyloid quantification method with adaptive templates and EPP-ROI can provide accurate, robust and simple amyloid quantification without MRI.

Original languageEnglish
Pages (from-to)5768-5780
Number of pages13
JournalPhysics in Medicine and Biology
Volume61
Issue number15
DOIs
Publication statusPublished - Jul 13 2016

All Science Journal Classification (ASJC) codes

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging

Fingerprint Dive into the research topics of 'Automated PET-only quantification of amyloid deposition with adaptive template and empirically pre-defined ROI'. Together they form a unique fingerprint.

Cite this