This study evaluated the impact of TOF reconstruction on SUVmax values in several selected brain regions which remarks importance in diagnosing AD/PD among three radioactive tracers, 11C-PIB, 11C-CFT and 18F-FDG, with hybrid PET/MR. It revealed that TOF reconstructions significantly affect on SUVmax values compared to non-TOF, and further improved the image contrast and % SNR. TOF and non-TOF SUVmax shows a strong correlation in most of the brain VOIs for all three radiotracers. To the best of our knowledge, it is the first time that different tracers for brain quantification imaging were performed with TOF and non-TOF PET/MRI. The results suggested that TOF PET/MR brain imaging could be recommended for diagnosing AD/PD rather than using of non-TOF PET/MR modalities, however, multi-centre clinical trials should be carried out in future for further confirmation of our findings.
Although simulating phantom experiments  and initial patient studies with 18F-FDG have been reported,  the quantitative effect would be different due to variances in radiotracer uptake and biodistribution within brain compartments, which also has benefitted in PET/MR for diagnosing neurodegenerative diseases by improving spatial resolution and SNR. Early phantom studies demonstrated the improved contrast of the smallest spheres with TOF compared with non-TOF and confirmed the faster convergence of contrast with TOF. [12-14] These gains are evident from visual assessment of the images as well as a quantitative evaluation of contrast recovery of the spheres and noise in the background. The improvements with TOF are higher for larger objects. Further above outcomes correlate with patient studies in which lesions (smaller VOIs) are seen more clearly and with higher uptake at comparable noise for TOF than with non-TOF. Armstrong et al. explain the use of TOF and/or PSF (point spread function) reduced partial volume effects may increase lesion detectability.  Moreover TOF capable to minimize PET quantification bias substantially and significantly improve the quantitative accuracy of standard MRAC (Magnetic Resonance Attenuation correction) methods in PET/MR. 
Results showed significantly increased SUVmax of 18F-FDG in TOF compared to conventional non-TOF reconstruction in all the brain VOIs. This is due to the higher uptake qualities of 18F-FDG which can cause a higher scatter fraction in both TOF and non-TOF. Further % RAD-SUVmax remains positively in all the brain VOIs. All the basal ganglionic regions (CN, PU, NA, PA and SN) showed higher % RAD-SUVmax in 18F-FDG compared to PIB which indicated the normal higher uptake influence of 18F-FDG in basal ganglionic regions. [17, 18] In cerebellar, cerebellar cortex showed slightly higher % RAD-SUVmax in 18F-FDG (6.62%) compared to 11C-PIB (5.26%). The commonly cerebellar cortex is considered as the reference region of normalizing the quantification values of these radiotracers in nuclear medicine in previous literature, due to its lack of affinity with this region and having homogeneity uptake appearance.[20, 29, 30]
Consequently, significantly increased SUVmax values were observed in all brain regions except the PU and CM in 11C-PIB brain images in TOF compared to conventional non-TOF. Yet, CM is close to the significance level of P<0.05 which shown in Table 2. This result is due to low uptake patterns of 11C-PIB in the putamen and cerebellar medulla regions, which cause relatively low scatter fraction contribution within the VOI, and could not benefit from improved quantitative accuracy using TOF technology. Increased % RAD-SUVmax in all brain cortex regions, (FL, TL, PL, OL, ACC and PCC) in 11C-PIB compared to 18F-FDG, this is due to the higher amyloid deposition in grey matter in cortex regions in AD patients. Though cerebellar medulla showed a higher RAD-SUVmax % difference in 11C-PIB (4.36%) compared to 18F-FDG (4.03%), this result is caused due to non-specific 11C-PIB retention in white matter and cerebrospinal fluid in CM .
In the 11C-CFT brain, caudate nucleus and putamen regions were shown slightly higher SUVmax values in TOF compared to non-TOF, though there was no significance in results. (Table 2) Higher % RAD-SUVmax (10.7%) of 11C-CFT in the caudate nucleus region was observed compared to 18F-FDG (8.96%) and 11C-PIB (8.47%), which may cause due to the higher uptake affinity of 11C-CFT in the caudate nucleus. However, in putamen it showed slightly lower RAD-SUVmax % in 11C-CFT (7.74%) compared to 18F-FDG (8.96%), this results may be due to the presence of PD positive patients within the suspected sample and it cause decrease uptake of 11C-CFT in the putamen region which cause a low impact of TOF effect compared to 18F-FDG. However, the above-mentioned insignificance of several brain regions among different radiotracers’ occurred due to lower uptake patterns and inefficient biodistribution were well explained in previous studies. [16, 17, 21, 22]
Previous studies of TOF PET/MR imaging has proved that TOF improves the contrast and signal to noise ratio. [35, 42] In this study, we measured the contrast of brain VOIs’ SUV mean as a ratio with respect to the SUV mean of cerebellar cortex VOI as mentioned in formula 01 in the method. Results proved that the TOF reconstructions significantly increased the average contrast of all brain cortex regions plus caudate nucleus, putamen and thalamus in 18F-FDG. Though nucleus accumbens, palladium, substantia nigra, brainstem and cerebellar medulla did not show contrast enhancement in TOF compared to non-TOF in 18F-FDG as shown in Figure 3A. This is due to the equal or higher uptake patterns in mentioned regions per the reference cerebellum cortex which cause not evident effect in TOF as we expected. In 11C-PIB, significantly increased contrast was observed in all the brain VOIs with TOF except the frontal lobe, though the frontal lobe showed equal contrast gain in TOF and non-TOF. The outcome of the frontal lobe shows the similar uptake patterns do exist in the frontal lobe and cerebellar cortex. For 11C-CFT, we observed higher contrast in both caudate nucleus and putamen regions, only putamen showed significance due to its high uptake properties of 11C-CFT within, compared to caudate nuclei.
Figures of Pearson’s correlation illustration shows TOF and non-TOF SUVmax measurements are strongly (R2≥0.7) positively correlated with significant difference among all the VOIs in the brain where reconstruction methods and SUVmax tends to increase together, which demonstrate TOF and non-TOF reconstructions equally impact towards the SUVmax value in the brain for 11C-PIB, 11C-CFT and 18F-FDG. (Figure S1-3) Further, we performed the Bland-Altman analysis for these two reconstruction methods where it evaluates the clinical applicability of the both methods, it reveals that both reconstruction methods are fundamentally equal for the whole brain according to the results. (Figure 4)
One previous study revealed that new generation PET systems, which utilize new reconstruction method/s such as OSEM, PSF, and TOF could be misinterpreted the SUVs.  Though according to our study, TOF can enhance the SUVmax in several ways and make it closer to the actual SUVmax value, the quantitative accuracy of TOF and its impact in quantitative measurements such as SUV mean, SUVmax, SUVSD and SUV peak of brain compartments in the assessment of brain pathologies should be investigated further in-depth.
Limitations of the study
Our study has several limitations. First, the smaller number of cases are included due to the retrospective study design, novelty of technology and time limitation for data collection of 11C-PIB, 11C-CFT and 18F-FDG scans of suspected AD/PD patients which were archived in the PACS (Picture Archive and Communication System). A larger sample would possibly be needed to generalize these findings to a considerable population (e.g. a wider range of patient BMI (Kg/m2) and wider range of age). Second, for outlining most of cortical structures as the brain VOIs, PET-based MPA (Maximum Probability Atlas) was used due to avoid slowness or interruption of the segmentation process of the PNEURO module in PMOD 3.9 software, though T1 MR based parcellation is preferred to VOI outlining in deep nuclei and thalamus by PMOD team, thus the quality of the VOI definition in above-mentioned areas are reduced. For the effective use of PNEURO with high-resolution data, a high-end workstation (e.g. 8 core, 16GB or more RAM) is required. Third, our study did not address the clinical relevance of these quantitative measures, therefore a subsequent study with correlation of imaging parameters with patient diagnosis would be preferred. Fourth, the PET acquisition and reconstruction parameters used in our study were not optimized, required parameters such as iterations, subsets and FWHM selected for the reconstructions were the same used in our clinical setting. The same iteration was selected to reduce the noise generated by increasing iteration unnecessarily. Yet, these parameters concurrently provided a clinically detectable enhancement in contrast and SNR for SUVmax measurements in most regions of brain VOIs.