Voxel-based evaluation for [¹⁸F] Florbetaben brain β-amyloid positron emission tomography of healthy control, mild cognitive impairment, and Alzheimer’s disease

Introduction

Alzheimer’s disease (AD) is the most common cause of dementia among elderly adults, accounting for 60–80% of dementia cases. AD is characterized by memory loss and cognitive decline as well as behavioral and psychological symptoms. AD poses serious challenges for public health systems, particularly in aging societies. Early diagnosis and timely treatment are crucial to overcoming many of these issues.

A core pathophysiology of AD is the accumulation of β-amyloid (Aβ) in the brain. Positron emission tomography (PET) using the radiotracer, florbetaben (FBB) labeled with the radioactive isotope fluorine-18 ([¹⁸F]) has proven highly effective in detecting Aβ plaques in the brain (1). Visual assessments of PET scans typically involve the comparison of [¹⁸F] FBB uptake in gray matter versus [¹⁸F] FBB uptake in white matter. Gray-matter uptake equal to or exceeding white-matter uptake indicates a positive diagnosis, whereas gray-matter uptake below white-matter uptake leads to a negative diagnosis (2). However, visual assessments depend heavily on the subjective judgment of the observer (3), making equivocal results more likely in cases where the difference between uptake intensity between gray matter and white matter is subtle.

The quantitative assessment of Aβ burden is crucial to obtaining a reliable diagnosis of AD and for monitoring longitudinal changes in Aβ burden during or following the use of anti-Aβ agents for AD treatment (4). Numerous protocols have been established for the use of PET scans to quantify brain Aβ burden (5,6). One approach referred to as the centiloid scale, allows the comparison of PET scans captured using different radiopharmaceutical or imaging modalities (7). Note however that the standardized uptake value ratio (SUVR) and centiloid scale both emphasize the severity of Aβ accumulation rather than its range of its distribution or spatial extent. Ozlen et al. posited that the distribution of accumulated Aβ, as represented as a distribution of radiopharmaceutical uptake in PET images, could be relevant to tau pathology and the selection of anti-Aβ agents (8). Note however that analysis methods based on regions of interest (ROIs), such as SUVR and centiloid scale, focus exclusively on specific targets and reference regions (i.e., disregarding all other regions). As a result, this approach depends heavily on the precision of ROI delineation.

Voxel-based analysis has also emerged as a useful tool in the interpretation of Aβ PET scans (9). In the current study, we employed voxel-based analysis of brain Aβ PET scans to compare [¹⁸F] FBB uptake in gray matter versus uptake in white matter in healthy controls (HCs) versus patients with AD or mild cognitive impairment (MCI). We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1100/rc).

Methods

Ethics

All clinical analysis was performed in accordance with the principles outlined in the Declaration of Helsinki (as revised in 2013). The study was under the approval of Institutional Review Board of Taipei Veterans General Hospital (No. 2022-07-040CC). The requirement for informed consent from study subjects was waived due to the retrospective design of the study and our use of anonymized and deidentified data.

Subject population and selection criteria

This study utilized [¹⁸F] FBB brain Aβ PET scans and brain magnetic resonance imaging (MRI) from HC subjects and patients with a wide range of neurodegenerative disorders. These data were obtained from the publicly available database, the Global Alzheimer’s Association Information Network (GAAIN; https://www.gaain.org/centiloid-project; ¹⁸F-amyloid tracers: ----> Florbetaben). This study also examined patients with neurodegenerative disorders who underwent [¹⁸F] FBB Aβ PET and MRI brain scans in Taipei Veterans General Hospital (TVGH) between October 2019 and December 2021. The inclusion criteria were that patients having [¹⁸F] FBB brain Aβ PET scans and brain MRI images. Subjects neither diagnosed of AD, MCI nor HC were excluded.

[¹⁸F] FBB brain Aβ PET scan and brain MRI imaging protocols

PET scans acquired from the GAAIN database and TVGH were obtained using the same established protocol. Briefly, PET scan acquisition was initiated at 90 min after the intravenous administration of [¹⁸F] FBB at a dosage of 300 MBq (8 mCi), with an imaging duration of 20 min (10). Using three-dimensional (3D) data acquisition mode and after attenuation and scatter correction, raw images were reconstructed iteratively using the ordered-subset expectation maximization algorithm (4 iterations; 16 subsets). For GAAIN database, scanning was performed using either a Philips Allegro PET camera or Philips TF64 PET/CT scanner (11). In TVGH, scanning was performed using a GE Discovery MI DR PET/CT scanner.

Brain MRI scans in TVGH were performed using a Signa HDxt 3 T GE system with an 8-channel phased-array neurovascular coil. The parameters used for 3D magnetization-prepared rapid acquisition with gradient echo (MPRAGE) imaging were as follows: repetition time (9.16 ms); echo time (3.39 ms); flip angle (20 degrees); field of view (384–320 mm); number of excitations [2]; and slice thickness (1 mm). For quantitative evaluation of [¹⁸F] FBB Brain Aβ PET, the SUVR was calculated using whole cerebellum as a reference region.

Intergroup comparison of gray matter voxel numbers

[¹⁸F] FBB Aβ PET scans and MPRAGE sequence MRI images were processed using MATLAB (MathWorks, Inc., Natick, MA, USA) and the statistical parametric mapping program SPM12 (Functional Imaging Laboratory, Institute of Neurology, University College London, London, United Kingdom). We developed an analytic pipeline to highlight gray matter voxels with uptake intensity exceeding 98% of the maximal uptake intensity of white matter (i.e., positive gray matter voxels). Our reason for setting a threshold of 98% (rather than 100%) was to decrease the influence of outliers. Note that this approach has also been employed in previous studies (12,13). We also calculated the number of positive gray matter voxels for cluster analysis (Figure 1).

Figure 1 Proposed image processing pipeline for [¹⁸F] FBB brain Aβ PET scans and MRI (MPRAGE) images for calculating the number of gray matter voxels with intensity exceeding 98% of the maximal intensity of white matter voxels as an indicator of Aβ accumulation. The yellow areas superimposed on the brain MRI indicated the location of cluster of gray matter voxels with intensity exceeding 98% of the maximal intensity of white matter voxels. MPRAGE, magnetization-prepared rapid acquisition with gradient echo; GMD, gray matter density maps; WMD, white matter density; PET, positron emission tomography; FBB, Florbetaben; Aβ, β-amyloid; MRI, magnetic resonance imaging.

[¹⁸F] FBB Aβ PET scans and MPRAGE images were converted from the original digital imaging and communications in medicine (DICOM) file format into the 3D NIfTI-1 (Neuroimaging Informatics Technology Initiative) file format to enable image preprocessing in SPM12. The origin of the [¹⁸F] FBB Aβ PET scans and MPRAGE images was shifted to roughly align with the anterior commissure of the individual brain space to facilitate re-registration, segmentation, and spatial normalization in SPM12. [¹⁸F] FBB Aβ PET scans were registered to MPRAGE images via 3D voxel registration using the normalized mutual information method. MPRAGE images were segmented to generate gray matter and white matter density maps, in accordance with the methods outlined in a previous study (14). 3D MPRAGE images were spatially normalized to the Montreal Neurological Institute (MNI) space with a voxel size of 3×3×3 mm³ via diffeomorphic anatomical registration and the exponentiated lie algebra (DARTEL) module in SPM12 (15). Spatially normalized [¹⁸F] FBB Aβ PET scans underwent spatial smoothing using a Gaussian smoothing kernel with full width at half maximum (FWHM) of 4 mm. Whole-brain gray and white matter masks were generated by applying a threshold of 0.5 to binarized probabilistic gray and white matter density maps. WM_max values were defined as 98% of the maximum intensity in the white matter mask, and calculated on a per-voxel basis. Clusters of voxels in the gray matter mask with intensities exceeding WM_max were highlighted.

Intergroup comparison of voxel uptake intensity

Whole-brain voxel-based analysis was employed to investigate intergroup differences in Aβ deposition. Spatially normalized [¹⁸F] FBB Aβ PET scans with a voxel size 3×3×3 mm³ were smoothed using a Gaussian smoothing kernel with an FWHM of 4 mm to facilitate a comparison of the various clinical diagnosis groups. Voxels served as the smallest independent unit of analysis. The mean uptake intensities of voxels at a given coordinate were compared across the following groups: (I) AD vs. HC; (II) MCI vs. HC; (III) AD vs. MCI.

Statistical analysis

Continuous variables were expressed as median [interquartile range (IQR) 25, 75]. The normality of continuous variables was examined using the Shapiro-Wilk test. ANOVA (Kruskal-Wallis) was used for comparisons of continuous variables (e.g., gray matter voxel number) across the three groups (AD, MCI, and HC), while controlling for age and gender, with P<0.05 as significant difference. Group comparisons were performed using two-sample t-tests based on data from the two groups of interest. Prior to parametric statistical analysis, the voxel uptake intensity of [¹⁸F] FBB Aβ PET was transformed using a Fisher r-to-z transformation to approximate a normal distribution. This enabled the use of a double statistical threshold, which involved a combined height threshold of P<0.01 and a minimum cluster size of 19 voxels, as determined via AlphaSim correction (16,17). Note that AlphaSim correction was used to evaluate the likelihood of a false detection (type I error) across the whole brain in Monte Carlo simulations (18). All statistical analyses were two-sided, and calculated using MATLAB R2022a (MathWorks Inc., Natick, Massachusetts, USA).

Results

This study initially comprised 40 subjects: 35 from the GAAIN database and 5 from TVGH. Note that [¹⁸F] FBB brain PET scans and brain MRI images were available for all subjects. Two of the 35 subjects from the GAAIN database were excluded due to a diagnosis of frontotemporal dementia. One of the 5 subjects from TVGH was excluded due to the lack of a definite diagnosis and 1 subject was excluded due to a diagnosis of hereditary spastic paraparesis. Note that exclusions were necessary to prevent the skewing of results in this small group. The final study group comprised 36 subjects, including 11 patients with AD, 9 patients with MCI, and 16 HC subjects (Figure 2). Among the 16 HC subjects, 10 were below the age of 60 years (referred to as young HC) and the other 6 subjects older than 60 years were referred to as elderly HC. The classification of young and elderly HC was based on annotations from the GAAIN database. The basic demographic characteristics of the 36 subjects are listed in Table 1.

Figure 2 Flowchart showing the process of participant selection from the database of the Global Alzheimer’s Association Information Network and Taipei Veterans General Hospital. PET, positron emission tomography; MRI, magnetic resonance imaging; FTD, frontotemporal dementia; HSP, hereditary spastic paraparesis.

Cerebral SUVR among HC (young and elderly), AD, and MCI

The number of positive gray matter voxels and SUVR of all 36 subjects (11 AD, 9 MCI, and 16 HC) are listed in the Table S1. The median SUVR values were as follows: 16 HC subjects [1.03 (IQR 1, 1.05)]; 11 AD subjects [1.38 (IQR 1.18, 1.81)]; and 9 MCI subjects [1.61 (IQR 1.3, 1.84)].

Intergroup comparison of gray matter voxel numbers

HC (including young and elderly) vs. AD vs. MCI

The number of positive gray matter voxels was significantly higher in the AD group than in the HC group [861 (IQR 61.5, 1,647) for AD vs. 0 (IQR 0, 0) for HC; P<0.001]. The number of positive gray matter voxels was significantly higher in the MCI group than in the HC group [1,016 (IQR 42, 1,589) for MCI vs. 0 (IQR 0, 0) for HC, P=0.002]. The difference in the number of positive gray matter voxels between AD and MCI patients did not reach the level of significance [861 (IQR 61.5, 1,647) for AD vs. 1,016 (IQR 42, 1,589) for MCI; P>0.99]. No significant differences in age or gender were observed between the AD, MCI, and elderly HC groups (Table 1). Significant differences in the number of positive gray matter voxels were observed between the AD, MCI, and elderly HC groups (P=0.0281).

Elderly HC vs. AD

No significant differences in gender or age were observed between the elderly HC group and AD group. The number of positive gray matter voxels was significantly higher among AD patients than among elderly HC subjects [861 (IQR 61.5, 1,647) for AD vs. 0 (IQR 0, 0) for elderly HC, P=0.036].

Elderly HC vs. MCI

No significant differences in gender or age were observed between the elderly HC group and MCI group. The number of positive gray matter voxels was higher among MCI patients than among elderly HC subjects; however, the difference did not meet the level of significance [1,016 (IQR 42, 1,589) for MCI vs. 0 (IQR 0, 0) for elderly HC, P=0.051].

Intergroup comparison of uptake patterns using voxel-based analysis

AD vs. elderly HC

The voxels that presented higher [¹⁸F] FBB uptake intensity among AD patients than among elderly HC subjects were distributed in the following brain regions: right middle temporal gyrus (P<0.01, cluster size =38 voxels, peak t-score =3.789, MNI coordinate =51, 3, −21), right superior temporal gyrus (P<0.01, cluster size =29 voxels, peak t-score =3.606, MNI coordinate =63, −24, 3; P<0.01, cluster size =29 voxels, peak t-score =3.342, MNI coordinate =60, −12, 0), right inferior frontal gyrus pars orbitalis (P<0.01, cluster size =44 voxels, peak t-score =3.766, MNI coordinate =42, 33, −6), right putamen (P<0.01, cluster size =60 voxels, peak t-score =3.638, MNI coordinate =18, 9, −6), right middle occipital gyrus (P<0.01, cluster size =1,871 voxels, peak t-score =5.708, MNI coordinate =33, −75, 39), right middle frontal gyrus (P<0.01, cluster size =1,199 voxels, peak t-score =5.998, MNI coordinate =36, 24, 42), right supramarginal gyrus (P<0.01, cluster size =19 voxels, peak t-score =3.398, MNI coordinate =57, −24, 30), right postcentral gyrus (P<0.01, cluster size =218 voxels, peak t-score =4.288, MNI coordinate =36, −36, 54), right precentral gyrus (P<0.01, cluster size =25 voxels, peak t-score =4.134, MNI coordinate =48, −6, 48), left inferior temporal gyrus (P<0.01, cluster size =27 voxels, peak t-score =3.809, MNI coordinate =−57, −39, −21), left putamen (P<0.01, cluster size =32 voxels, peak t-score =3.913, MNI coordinate =−15, 12, −12), left middle temporal gyrus (P<0.01, cluster size =38 voxels, peak t-score =3.581, MNI coordinate =−54, −54, −3), and left superior frontal gyrus (dorsolateral) (P<0.01, cluster size =22 voxels, peak t-score =3.357, MNI coordinate =−15, 21, 54) (Table 2, Figure 3A). No voxels presented higher [¹⁸F] FBB uptake intensity among elderly HC subjects than among AD patients.

Figure 3 Voxel clusters highlighting significant differences in [¹⁸F] FBB uptake intensity and Aβ accumulation: (A) AD vs. elderly HC subjects; and (B) MCI vs. elderly HC subjects. A two-sample t-test group comparison was performed between elderly HCs and each of the other groups. This was achieved using a double statistical threshold with a height threshold of P<0.01 and a minimum cluster size of 19 voxels, as determined using AlphaSim correction. The scale of the color bar indicates the T-score. FBB, Florbetaben; Aβ, β-amyloid; AD, Alzheimer’s disease; HC, healthy control; MCI, mild cognitive impairment.

MCI vs. elderly HC

Voxels presenting higher [¹⁸F] FBB uptake intensity among MCI patients than among elderly HC subjects were distributed in the following brain regions: right inferior temporal gyrus (P<0.01, cluster size =44 voxels, peak t-score =3.945, MNI coordinate =48, −9, −36; P<0.01, cluster size =74 voxels, peak t-score =3.682, MNI coordinate =60, −42, −24), right middle frontal gyrus (P<0.01, cluster size =457 voxels, peak t-score =5.314, MNI coordinate =42, 36, 33), right middle temporal gyrus (P<0.01, cluster size =27 voxels, peak t-score =3.688, MNI coordinate =45, −66, 15), right supramarginal gyrus (P<0.01, cluster size =37 voxels, peak t-score =3.135, MNI coordinate =60, −18, 21; P<0.01, cluster size =115 voxels, peak t-score =3.916, MNI coordinate =57, −48, 24), left middle temporal gyrus (P<0.01, cluster size =35 voxels, peak t-score =4.026, MNI coordinate =−48, −21, −15), left superior temporal gyrus (P<0.01, cluster size =61 voxels, peak t-score =4.467, MNI coordinate =−48, −18, 3), left anterior cingulate and paracingulate gyri (P<0.01, cluster size =370 voxels, peak t-score =4.610, MNI coordinate =−12, 36, 21), left superior frontal gyrus (medial) (P<0.01, cluster size =24 voxels, peak t-score =3.603, MNI coordinate =−9, 54, 27), left precuneus (P<0.01, cluster size =179 voxels, peak t-score =3.964, MNI coordinate =−15, −51, 45), left inferior parietal gyrus (P<0.01, cluster size =79 voxels, peak t-score =3.678, MNI coordinate =−39, −39, 36) (Table 3, Figure 3B). No voxels presented higher [¹⁸F] FBB uptake intensity among elderly HC subjects than among MCI patients.

AD vs. MCI

No voxels presented a significant difference in [¹⁸F] FBB uptake intensity between AD patients and MCI patients.

Discussion

The effectiveness of ROI-based image analysis is constrained by the heterogeneous accumulation of Aβ (19). Thus, it is likely that voxel-based analysis of the whole brain cortex could outperform ROI-based analysis, by eliminating the need for accurate ROI selection (20,21). Nonetheless, voxel-based analysis is prone to interference due to non-specific uptake from adjacent white matter and the skull, which can skew estimates pertaining to cortical uptake (9).

In the current study, we adopted a novel voxel-based analysis method, which involved calculating the number of positive gray matter voxels throughout the entire cerebrum to determine the distribution of accumulated Aβ. Our findings revealed that the distribution of Aβ (based on the number of positive gray matter voxels) was more extensive among MCI and AD patients than among HC subjects. Note that 15 of the 16 HC subjects did not present any positive gray matter voxels. Note also that there was wide variation in the number of positive gray matter voxels observed among AD and MCI subjects. We attribute the large variation between AD and MCI subjects to the small study population. Moreover, amyloid accumulation in the brain can vary significantly even among patients with the same diagnosis, reflecting differences in the severity of the disease.

It has been suggested that even among healthy individuals, Aβ accumulation in the brain increases with age (22,23). Thus, it is reasonable to assume that the existence of positive gray matter voxels in one of the elderly HC subjects was at least partially due to age. To account for age, we conducted further analysis comparing only elderly HC subjects versus MCI and AD patients. Consistent with previous findings, the number of positive gray matter voxels was higher among AD subjects than among elderly HC subjects. Note that the number of positive gray matter voxels was higher among MCI than among elderly HC subjects; however, the difference did not meet the level of significance (P=0.051). These results were likely influenced by the small number of elderly HC subjects (6 subjects) and the fact that 1 of those 6 subjects exhibited a relatively high number of positive gray matter voxels (voxel number =93). A more accurate comparison will no doubt require further analysis with a larger number of elderly HC subjects.

A voxel-to-voxel comparison of [¹⁸F] FBB uptake revealed higher uptake intensity in the AD group (Table 2, Figure 3A) or MCI (Table 3, Figure 3B) than in elderly HC. Moreover, these effects were particularly evident in the frontal, parietal, temporal, cingulate gyrus, and precuneus regions. Note that similar Aβ accumulation patterns have been reported in previous studies (24-26). Based on PET scans obtained using C-11 Pittsburgh Compound-B as a radiopharmaceutical agent, Villeneuve et al. reported that most of the Aβ accumulated in the medial frontal cortex, followed by the precuneus, lateral frontal, parietal lobes, and lateral temporal lobe (24). Based on Aβ PET scans obtained using [¹⁸F] FBB as a radiopharmaceutical agent, Chung et al. reported that most of the Aβ accumulated in the posterior cingulate precuneus, followed by the lateral temporal cortex and the lateral parietal cortex (25). Also based on Aβ PET scans obtained using [¹⁸F] FBB, Bullich et al. reported that the earliest accumulation of Aβ occurs in the cingulate cortices and precuneus (26).

This study was subject to a number of limitations, which should be considered in the interpretation of our findings. First, the study population included only 36 subjects, which inevitably led to selection bias. This work should be considered preliminary, proof-of-concept research. The clinical value of our results in promoting the visual assessment of PET images will require confirmation in further studies on a larger study population. Second, we set the maximal intensity of uptake in white matter at 98%, which is a relatively high threshold by which to define positive gray matter voxels and may have had a detrimental effect on sensitivity in identifying positive images. Nonetheless, this approach gave us confidence that the extracted positive gray matter voxels (i.e., above this threshold) represented significant Aβ deposits. Third, researchers are uncertain as to whether MCI is a presentation of early AD or some other type of dementia. It is also possible that Aβ accumulation in the brain is associated with conditions other than AD (27). In the current study, we did not observe a significant difference between AD and MCI subjects in terms of the distribution of Aβ accumulation or positive gray matter voxel numbers (Table 1); however, we did not control for the assumption that the etiology of MCI was caused by AD. Further research with control over the etiology of MCI should be conducted.

Conclusions

In this preliminary study on a limited sample size, we compared the uptake intensity of gray matter vs. white matter in [¹⁸F] FBB brain Aβ PET scans. Voxel-based analysis of positive gray matter voxels revealed significant differences between AD, MCI and HC patients. Nonetheless, our findings pertaining to the clinical value of voxel-based analysis will require verification in studies with a larger population.

Acknowledgments

Funding: This work was financially supported in part by the National Science and Technology Council, under the project NSTC 112-2628-E-038-001-MY3 and also in part by the Higher Education Sprout Project by the Ministry of Education under the project DP2-TMU-113-A-10.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-24-1100/rc

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1100/coif). All authors report that this work was partially supported by the National Science and Technology Council under project number NSTC 112-2628-E-038-001-MY3 and also by the Higher Education Sprout Project by the Ministry of Education under project number DP2-TMU-113-A-10. Additionally, a U.S. patent is pending for a “BRAIN AMYLOID PET PROCESSING SYSTEM AND OPERATION METHOD THEREOF AND NON-TRANSITORY COMPUTER READABLE MEDIUM” (application date: October 24, 2023, application number 18/493,080). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was under the approval of Institutional Review Board of Taipei Veterans General Hospital (No. 2022-07-040CC). Given the nature and design of the study, the committee granted a waiver for the requirement of informed consent from participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Jack CR Jr, Bennett DA, Blennow K, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease. Alzheimers Dement 2018;14:535-62. [Crossref] [PubMed]
2. Sabri O, Seibyl J, Rowe C, et al. Beta-amyloid imaging with florbetaben. Clin Transl Imaging 2015;3:13-26. [Crossref] [PubMed]
3. Kim JY, Oh D, Sung K, Choi H, Paeng JC, Cheon GJ, Kang KW, Lee DY, Lee DS. Visual interpretation of [18F]Florbetaben PET supported by deep learning-based estimation of amyloid burden. Eur J Nucl Med Mol Imaging 2021;48:1116-23.
4. Pemberton HG, Collij LE, Heeman F, et al. Quantification of amyloid PET for future clinical use: a state-of-the-art review. Eur J Nucl Med Mol Imaging 2022;49:3508-28. [Crossref] [PubMed]
5. van Dyck CH, Swanson CJ, Aisen P, et al. Lecanemab in Early Alzheimer's Disease. N Engl J Med 2023;388:9-21. [Crossref] [PubMed]
6. Tampi RR, Forester BP, Agronin M. Aducanumab: evidence from clinical trial data and controversies. Drugs Context 2021;2021: [Crossref] [PubMed]
7. Klunk WE, Koeppe RA, Price JC, et al. The Centiloid Project: standardizing quantitative amyloid plaque estimation by PET. Alzheimers Dement 2015;11:1-15.e1-4.
8. Ozlen H, Pichet Binette A, Köbe T, et al. Spatial Extent of Amyloid-β Levels and Associations With Tau-PET and Cognition. JAMA Neurol 2022;79:1025-35. [Crossref] [PubMed]
9. Arai A, Kaneta T, Okamura N, et al. Pitfalls of voxel-based amyloid PET analyses for diagnosis of Alzheimer's disease: artifacts due to non-specific uptake in the white matter and the skull. Tohoku J Exp Med 2014;234:175-81. [Crossref] [PubMed]
10. Minoshima S, Drzezga AE, Barthel H, et al. SNMMI Procedure Standard/EANM Practice Guideline for Amyloid PET Imaging of the Brain 1.0. J Nucl Med 2016;57:1316-22. [Crossref] [PubMed]
11. Rowe CC, Doré V, Jones G, et al. (18)F-Florbetaben PET beta-amyloid binding expressed in Centiloids. Eur J Nucl Med Mol Imaging 2017;44:2053-9. [Crossref] [PubMed]
12. WahidKAHeRMcDonaldBAAndersonBMSalzilloTMulderSWangJSharafiCSMcCoyLANaserMAAhmedSSandersKLMohamedASRDingYWangJHutchesonKLaiSYFullerCDvan DijkLV. MRI Intensity Standardization Evaluation Design for Head and Neck Quantitative Imaging Applications.medRxiv 2021:2021.02.24.21252322.
13. Duarte KTN, Gobbi DG, Sidhu AS, McCreary CR, Saad F, Camicioli R, Smith EE, Frayne R. Segmenting white matter hyperintensities in brain magnetic resonance images using convolution neural networks. Pattern Recognition Letters 2023;175:90-4.
14. Ashburner J, Friston KJ. Unified segmentation. Neuroimage 2005;26:839-51. [Crossref] [PubMed]
15. Ashburner J. A fast diffeomorphic image registration algorithm. Neuroimage 2007;38:95-113. [Crossref] [PubMed]
16. Song XW, Dong ZY, Long XY, et al. REST: a toolkit for resting-state functional magnetic resonance imaging data processing. PLoS One 2011;6:e25031. [Crossref] [PubMed]
17. Hung SC, Lee CC, Chen HH, et al. Early recovery of interhemispheric functional connectivity after corpus callosotomy. Epilepsia 2019;60:1126-36. [Crossref] [PubMed]
18. Ledberg A, Akerman S, Roland PE. Estimation of the probabilities of 3D clusters in functional brain images. Neuroimage 1998;8:113-28. [Crossref] [PubMed]
19. Tahmi M, Bou-Zeid W, Razlighi QR. A Fully Automatic Technique for Precise Localization and Quantification of Amyloid-β PET Scans. J Nucl Med 2019;60:1771-9. [Crossref] [PubMed]
20. Astrakas LG, Argyropoulou MI. Shifting from region of interest (ROI) to voxel-based analysis in human brain mapping. Pediatr Radiol 2010;40:1857-67. [Crossref] [PubMed]
21. Seyedi S, Jafari R, Talaei A, Naseri S, Momennezhad M, Moghaddam MD, Akbari-Lalimi H. Comparing VBM and ROI analyses for detection of gray matter abnormalities in patients with bipolar disorder using MRI. Middle East Current Psychiatry 2020;27.
22. Ossenkoppele R, Jansen WJ, Rabinovici GD, et al. Prevalence of amyloid PET positivity in dementia syndromes: a meta-analysis. JAMA 2015;313:1939-49. [Crossref] [PubMed]
23. Xie Y, Yang Q, Liu C, et al. Exploring the Pattern Associated With Longitudinal Changes of β-Amyloid Deposition During Cognitively Normal Healthy Aging. Front Med (Lausanne) 2020;7:617173. [Crossref] [PubMed]
24. Villeneuve S, Rabinovici GD, Cohn-Sheehy BI, et al. Existing Pittsburgh Compound-B positron emission tomography thresholds are too high: statistical and pathological evaluation. Brain 2015;138:2020-33. [Crossref] [PubMed]
25. Chung SE, Kim HJ, Jo S, et al. Patterns of Focal Amyloid Deposition Using (18)F-Florbetaben PET in Patients with Cognitive Impairment. Diagnostics (Basel) 2022;12:1357. [Crossref] [PubMed]
26. Bullich S, Roé-Vellvé N, Marquié M, et al. Early detection of amyloid load using (18)F-florbetaben PET. Alzheimers Res Ther 2021;13:67. [Crossref] [PubMed]
27. Villemagne VL, Ong K, Mulligan RS, et al. Amyloid imaging with (18)F-florbetaben in Alzheimer disease and other dementias. J Nucl Med 2011;52:1210-7. [Crossref] [PubMed]