Table 1 Study characteristics of included articles
Citation | Study Design | Target Condition | Sample Size (Diseased/Healthy) | Index Tests | Assessment Method (Visual/Quantitative) | Diagnostic Accuracy Measures | Findings |
|---|---|---|---|---|---|---|---|
Anazodo et al. [2017] [27] | Case-control | Frontotemporal dementia (FTD) | 10/10 | ASL-MRI /FDG-PET | Visual | Sensitivity/specificity | ASL showed lower sensitivity (0.6667), specificity (0.6212), and inter-rater reliability (0.2) than FDG-PET (being 0.8843/0.9091/0.61 respectively) |
Quantitative | t-score maps | Hypometabolism areas exceeded that of hypoperfused ones | |||||
Ceccarini et al. [2020] [19] | Case-control | Different types of dementia | 27/30 | Enhanced multiplane tagging ASL-MRI/ FDG-PET | Visual | Sensitivity, specificity, and interobserver agreement | Similar specificity (0.7) was registered among both modalities, but higher sensitivity (0.93) and interobserver agreement (0.64) for FDG-PET |
Quantitative | Volume of interest (VOI)–based analysis/z-scores | FDG-PET revealed more volume and intensity abnormalities than ASL-MRI. FDG-PET-based VOI reported higher z-scores with varying brain regions | |||||
Chen et al. [2011] [28] | Case-control | Alzheimer’s disease (AD) | 15/19 | Pseudo continuous ASL-MRI /FDG-PET | Quantitative | Statistical parametric mapping and regions of interest (ROI) analysis | Perfusion and metabolism maps showed an overlap in various brain regions including bilateral angular gyri and posterior cingulate ROI results revealed similar abnormalities in diseased and healthy patients |
Linear correlation maps of perfusion and metabolism to neuropsychological test scores | Both modalities were able to correctly discriminate neural networks related to psychological tests. Positive correlations were observed in the dorsolateral prefrontal cortex and bilateral inferior parietal lobes | ||||||
Corouge et al. [2012] [29] | Case-control | Semantic dementia (SD) | 6/9 | Pulsed ASL-MRI /FDG-PET-CT | Visual | Partial volumes effects (PVE) maps for inspection of brain regions | Good agreement between the two modalities; hypoperfusion and hypometabolism were observed in areas including basifrontal, anterior temporal lobe, left posterior part of the temporal lobe, and left parietal lobe |
Dolui et al. [2020] [30] | Case-control | Mild cognitive impairment (MCI) | 50/35 | Pseudo continuous ASL-MRI /FDG-PET-CT | Quantitative | t-score maps | ASL-MRI and FDG-PET have demonstrated similar areas of abnormalities including the medial temporoparietal regions |
Receiver-operator characteristic curve (ROC-curve) analysis | Similar area under the curve (AUC) between ASL and FDG | ||||||
Fällmar et al. [2017] [20] | Case-control | Frontotemporal dementia (FTD) | 20/38 | Pseudo continuous ASL-MRI /FDG-PET | Visual | Sensitivity and specificity | ASL-MRI had higher specificity (0.84), but significantly lower sensitivity (0.53 vs. 0.96) than FDG-PET |
Alzheimer’s disease (AD) | 25/38 | ||||||
Musiek et al. [2012] [31] | Case-control | Alzheimer’s disease (AD) | 15/19 (qualitative assessment) | Pseudo continuous ASL-MRI /FDG-PET | Visual | Receiver-Operator curve (ROC) analysis, and sensitivity/specificity analysis | Readers were more confident with FDG-PET images than ASL-MRI ones; both modalities displayed comparable sensitivity and specificity measures |
13/18 (quantitative assessment) | Quantitative | Both modalities demonstrated excellent diagnostic accuracy, with FDG-PET having AUC of 0.9 and that of ASL-MRI being 0.91 | |||||
Nedelska et al. [2018] [32] | Case-control | Dementia with Lewey bodies (DLB) | 19/76 | Pseudo continuous ASL-MRI /FDG-PET and tau-PET | Quantitative | Quantitative maps, like voxel-wise ones, derived from Statistical Parametric Mapping | Hypometabolism and hypoperfusion patterns of the cortex showed a clear similarity among DLB patients in areas including precuneus, cuneus and posterior parieto-occipital cortices |
Alzheimer’s disease (AD) | 19/76 | ROC-curve analysis | FDG-PET performed better with AUC of 0.91, but ASL-MRI also had good accuracy with AUC of 0.8 | ||||
Riederer et al. [2018] [33] | Case-control | Mild cognitive impairment (MCI) | 20/11 | Pulsed ASL-MRI /FDG-PET | Quantitative | Voxel-wise analyses of variance, volume of interest (VOI), and independent component analyses | For MCI patients, FDG-PET revealed hypometabolism patterns in the precuneus unlike ASL-MRI |
Alzheimer’s disease (AD) | 45/11 | For AD patients, both ASL-MRI and FDG-PET showed uniform patterns of hypoperfusion and hypometabolism respectively in areas including precuneus, parietal, temporal, and occipital cortex | |||||
Tosun & Jagust et al. [2016] [34] | Case-control | Early mild cognitive impairment (early-MCI) | 30/34 | ASL-MRI /FDG-PET | Quantitative | Partial least squares (PLS) method generated maps | Differences in the whole-brain perfusion and metabolism patterns |
Late mild cognitive impairment (late-MCI) | 25/34 | Sensitivity and specificity measures | Insignificant differences in sensitivity and specificity; however, FDG-PET had shown better measures in diagnosing AD and late MCI | ||||
Alzheimer’s disease (AD) | 20/34 | ||||||
Tosun & Rabinovici et al. [2016] [35] | Case-control | Behavioral variant of frontotemporal dementia (bvFTD) | 32/15 | ASL-MRI /FDG-PET | Quantitative | Partial least squares (PLS) method generated maps | Spatial regions differentiating each disorder as derived from ASL-MRI and FDG-PET are similar |
Alzheimer’s disease (AD) | 28/15 | Sensitivity and specificity | Sensitivity and specificity measures of ASL and FDG-PET were uniform in discriminating AD, bvFTD, and control subjects. In differentiating AD from control participants, sensitivity and specificity of ASL was 0.86 and 0.92, whereas that of FDG-PET was 0.78 and 1 respectively | ||||
Verclytte et al. [2016] [25] | Prospective | Early-onset Alzheimer’s disease (EOAD) | 37/0 | Pseudo continuous ASL-MRI /FDG-PET-CT | Quantitative | P-maps | Hypometabolic regions were more extensive than hypoperfused ones. ASL maps highlighted changes in frontal lobes unlike FDG-PET |
Verfaillie et al. [2015] [36] | Case-control | Frontotemporal dementia (FTD) | 12/10 | Pseudo continuous ASL-MRI /FDG-PET | Quantitative | Voxel-wise comparison Region-of-interest (ROI) and correlation | Perfusion and metabolism maps showed significant decrease in signal for AD patients in both precuneus and inferior parietal lobule As for both AD and FTD, there was a decrease in metabolism and perfusion in medial prefrontal cortex |
Alzheimer’s disease (AD) | 18/10 | ROC curve analyses | Area under curve were somehow comparable between FDG-PET and ASL-MRI pertaining the precuneus (0.74 versus 0.72) and inferior parietal lobule (0.94 versus 0.85), but not the medial prefrontal cortex region (both 0.68) | ||||
Weyts et al. [2017] [26] | Retrospective | Different types of dementia | 9 demented patients/0 | Pseudo continuous ASL-MRI /FDG-PET-CT | Visual | Inter- and intramodality agreement | Inter and intramodality agreements were insignificantly different among both modalities |
Regional agreement | Between modalities, intermodality agreement was similar in some brain regions including the precuneus | ||||||
Diagnostic accuracy | For ASL-MRI diagnostic accuracy is 5/9, and that of FDG-ASL is 7/9 |