An optimized protocol for correlation of PET-FDG and pCASL data in mild cognitive impairment (MCI) patients

Introduction: Recent development of anti-βamyloid drugs for dementigenic neurodegenerative diseases treatment necessitates identification of target patients in MCI stage. Clinical framing also relies on the identification of brain hypometabolism patterns suggestive of neurodegeneration detected by PET-FDG. New MRI perfusion methods such pseudo Continuous Arterial Spin Labeling (pCASL) that do not require contrast medium make it possible adding perfusion data to biomarkers useful for diagnosis. A study to define an optimized protocol analysis was conducted, to investigate the correlation between FDG-PET and pCASL information. Material and Methods: FDG-PET, pCASL and structural MR images were acquired from 15 patients (mean age 76 yr.). To validate the present analysis with a standard group analysis, a sample of 111 healthy subjects FDG-PET scans underwent the same processing. Starting from a standard SPM12 pipeline and using customized Matlab routines, an optimized protocol of preprocessing and ROI-based analysis were defined. Results: PET, pCASL and MR were normalized in MNI space and smoothed. A better spatial normalization for pCASL scans was obtained using the source image. Considering PET/pCASL low spatial resolution, different methods of Gray Matter (GM) segmentation were compared. A mean tissue probability map of GM, obtained from 662 healthy elderly subjects in the same patients range of age, was considered and used also for PVE correction in PET scans. For patients, GM obtained using the template was compared with the segmented one from the T1-weighted image to evaluate the eventual atrophy degree or possible cortex reduction due to incomplete atlas-patient cortex superimposition. Standardized Uptake Values (SUVr) from FDG-PET and Cerebral Blood Flow values (CBFr) values form pCASL, both normalized on cerebellar mean values, were calculated for all the 88 cortical regions of the Automated Anatomic Labeling template version 3 (AAL3) and for the 22 macro-regions obtained by combining sub-areas of AAL3. Discussion: According to this optimized protocol, further analysis on images are now being conducted. They will lead to the investigation of correlation between SUVr and CBFr values in the 88 regions and in the 22 macro-regions through Pearson coefficients, histograms and pattern similarities. A validation of the present analysis with the standard spm12 one was also conducted.