{"paper_id":"317bfc5b-c6e5-466a-aad8-ef2c138c01be","body_text":"Test-Retest Reliability of Dopaminergic fPET and fMRI Measures \nDuring Reward Processing  \nSchlosser G1, 2, Handschuh PA1, 2, Murgaš M1, 2, Graf S1, 2, Milz C1, 2, Klug S1, 2, Falb P1, 2,  \nSchmidt C1, 2, Eggerstorfer B1, 2, Briem E1, 2, Mayerweg A1, 2, Artmeier L1, 2,  \nGodbersen GM1,2, Nics L3, Razul S3, Hacker M3, Hahn A1, 2, Lanzenberger R1, 2*, Reed MB1, 2 \n \n1Department of Psychiatry and Psychotherapy, Medical University of Vienna, Austria \n2Comprehensive Center for Clinical Neurosciences and Mental Health (C3NMH) \n3Department of Biomedical Imaging und Image-guided Therapy, Division of Nuclear \nMedicine, Medical University of Vienna, Austria \n \nBioRxiv \n \nMain article \nWord count 200 Abstract \n          4139 Main section  \nTables and figures: 4 \n References: 43 \n \n*Correspondence to:    \nUniv. Prof. PD Dr.med. Rupert Lanzenberger \nTel.: +43-1-40400-35760  \nE-mail: rupert.lanzenberger@meduniwien.ac.at \nDepartment of Psychiatry and Psychotherapy \nMedical University of Vienna \nWaehringer Guertel 18-20, A-1090 Wien \n \nClinicaltrials.gov Identifier: NCT06675851 \nEudraCT Number: 2019-004880-33 \n \n \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\n \nAbstract: \nReward processing is essential to human brain function , with dopamine signalling in the \nnucleus accumbens (NAcc) as key element. The monetary incentive delay task is widely studied \nwith functional magnetic resonance imaging ( fMRI), measuring indirect hemodynamic \nchanges. Functional positron emission tomography (fPET) with 6-[¹⁸F]FDOPA directly quantifies \ndopamine synthesis enabling dynamic assessment during task performance within a single \nscan. We investigated the reliability of 6 -[¹⁸F]FDOPA fPET and  blood oxygenation level \ndependent (BOLD) fMRI during a modified  monetary incentive delay  task in 25 healthy \nparticipants across two PET/MRI sessions. Intraclass correlation coefficients and coefficients of \nvariance were computed for BOLD beta estimates and striatal dopamine synthesis at 30s and \n2s resolutions . fPET showed fair to good reliability in the NAcc and putamen at rest, fair \nreliability during the win condition in the caudate and putamen, but poor reliability across the \nloss condition in all regions. Conversely, fMRI showed good reliability in the NAcc during \nfeedback and in the caudate during feedback loss, but fair reliability elsewhere except for poor \nreliability in the caudate during cue loss. These findings indicate that both methods achieve \ncomparable reliability  but in different target areas , with the molecular speci ficity of fPET \noffering dynamic assessment of dopaminergic function. \n   \nClinicaltrials.gov Identifier: NCT06675851  \n \n \n \n \n \nKeywords:    Functional PET (fPET); 6-[¹⁸F]FDOPA ; Monetary incentive delay (MID); Dopamine \nsynthesis; Test-Retest-Reliability\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nIntroduction: \nDopamine plays a central role in modulating reward and loss processing (1). The dopaminergic \nneurons most relevant to these processes project from the ventral tegmental area (VTA) to the \nnucleus accumbens (NAcc) and ventral striatum (2,3). The NAcc is a abundantly innervated \nregion important for the integration of cortical afferent information and goal -directed \nbehaviour (4). Reward processing can be divided into an anticipation phase and an outcome \nphase, both of which engage dopaminergically innervated regions of the striatum, including \nthe NAcc, caudate, and putamen (5). Disruptions in these processes are common in a variety \nof neuropsychiatric disorders including, major depressive disorders (MDD), addiction, anxiety \nand schizophrenia (6–8), where patients often show reduced motivation and anhedonia (9).  \nThe monetary incentive delay (MID) task is a widely employed paradigm for probing t he \ndifferent phases of reward and loss processing in humans and has been combined extensively \nwith blood oxygen level dependant (BOLD) functional magnetic resonance imaging (fMRI) (5). \nOriginally based on non -human primate  research showing reward anticipation-related \nactivation in VTA dopaminergic neurons (1,10), the MID task has since been applied in more \nthan 200 fMRI studies (5). BOLD fMRI has a number of advantages including a high \nspatiotemporal resolution, high sensitivity to fluctuations in blood oxygenation and the \npossibility to acquire high resolution anatomic scans in the same session for localisation and \nco-registration (11). However the BOLD signal is affected by a  combination of  multiple \nvariables including blood oxygenation, blood flow and blood volume making it only and \nindirect measure of neuronal activation (12). Additionally the BOLD signal may be instable \nduring longer task performance and can be influenced by physical effects like magnetic field \ninhomogeneities, low frequency drifts and heating (13,14). \nTask-induced alterations during MID performance have recently been shown also for \ndopamine synthesis rates (15). Positron emission tomography (PET) with the radiolabelled \ndopamine precursor 6-[¹⁸F]FDOPA is a suitable method for mapping reward-related dopamine \nsynthesis (16). However, conventional PET protocols typically require at least two \nmeasurements. One measurement is needed to capture resting state activity and another one \ncaptures task performance (17–19). The necessity to acquire data on different days introduces \nvariability due to distinctions in daily performance and resting activity. Furthermore, the \nradiation burden is increased by repeated measurements.  To capture both resting-state and \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\ntask-induced alterations in a single session functional PET (fPET) was developed to address \nthese limitations (20,21). fPET allows to capture dynamics of radiotracer binding by controlling \nthe delivery of the radiotracer to the blood via continuous infusion. Like fMRI , fPET uses \nrepeated periods of task performance alternated with a control condition, thereby enabling \nthe measurement of task-induced changes in dopamine synthesis across multiple conditions \nin a single scan session (15). To further optimize the fPET framework  6-[¹⁸F]FDOPA may be \napplied using a bolus plus constant infusion protocol, which allows for an increase in signal-to-\nnoise ratio and provides a higher temporal resolution (22). As a further advancement , high-\ntemporal resolution fPET with the glucose analogue [18F]fluorodesoxyglucose ([18F]FDG) and \nreconstruction to 3  s frames enabled a direct comparison between fPET and fMRI  signals \n(23,24). Furthermore, [18F]FDG fPET showed high test-retest reliability during rest and \nmoderate performance during the execution of a cognitive task, which was however still higher \nwhen compared to BOLD fMRI (25). For future scientific and clinical applications any imaging \nparameter must provide robust reliability to ensure detection of subtle task -dependent \nchanges despite variance that may arise from repeated measurements.  However, for 6-\n[¹⁸F]FDOPA dopamine synthesis fPET two crucial aspects for such use have not yet been \ndetermined: identification of task-induced changes at high temporal resolution of seconds and \ntest-retest reliability. Therefore, we aim to assess the test-retest reliability of 2  s and 30  s \nframes of task induced dopamine synthesis measured with 6-[¹⁸F]FDOPA fPET and BOLD fMRI. \nWe further aimed to assess the effects extended dynamic Non-local-Means (edNLM) filtering \nhave on the test-retest reliability (26) of high-temporal fPET data compared to Gaussian 8 mm \nsmoothing.  \n \nMethods and Materials \nSubjects \nTwenty-five (10 female, age 24.6 ± 6.1 years) healthy participants completed two PET/MRI \nmeasurements with the radiotracer 6-[18F]FDOPA. All subjects underwent a routine medical \nevaluation during a screening visit, which included electrocardiography, blood tests, \nneurological and physiological assessments, and a urine drug test.  Female participants \nadditionally took a pregnancy test at the screening visit and before each PET/MRI session.  \nPsychiatric disorders were excluded using the Structured Clinical Interview for DSM -V, \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nadministered by an experienced interviewer. Exclusion criteria included a weight above 100 \nkg, current or past neurological, physiological, or psychiatric disorders, current breastfeeding \nor pregnancy, left-handedness, substance abuse, MRI contraindications, and participation in a \nstudy involving ionizing radiation exposure within the past 10 years. After detailed explanation \nof the study protocol, all subjects gave written informed consent. All subjects were insured \nand reimbursed for their participation. The study was approved by the Ethics Committee of \nthe Medical University of Vienna (ethics number: 2321/2019) and all procedures were carried \nout in accordance with the Declaration of Helsinki. The study was registered in the European \nClinical Trial Database (EudraCT 2019-004880-33). \nMonetary Incentive Delay Task \nParticipants performed a modified version of the monetary incentive delay (MID) task during \nsimultaneous PET/MRI acquisition. They were instructed to maximize monetary gains and \navoid losses by responding as quickly as possible to a target cue. Each trial began with the \npresentation of a cue indicating potential gain or loss amount. A response faster than the \nindividualized reaction time (RT) resulted in the reward being obtained or the loss avoided; \nslower responses yielded no gain or incurred the loss. Reward/loss magnitudes were €0.5, €1, \nor €3, and participants started with  an initial  balance of  €10. To enhance motivation, \nparticipants were informed that their remaining balance at the end of the session would be \npaid out if it was positive. \nThe task was adapted to accommodate the lower temporal resolution of fPET, which currently \nprecludes conventional event -related analysis. Before scanning, participants completed a \npractice session to familiarize themselves with the paradigm and determine individual mean \nRT. During scanning, each participant completed four task blocks (297 s each), with each block \ncontaining 27 trials of fixed duration (11 s). Trials consisted of an anticipation phase (2–5 s, in \n0.5 s increments), a reaction phase (limited t o the individual RT; maximum 1 s), a feedback \nphase (2 s), and a baseline fixation cross of variable duration to maintain a total trial length of \n11 s. \nBlocks were assigned pseudorandomly as either gain or loss conditions. In gain blocks, the RT \nthreshold was increased by 50 ms to enhance the probability of success; in loss blocks, it was \ndecreased by 50 ms to increase the likelihood of failure. RTs were re-estimated at the start and \nmidpoint of each block to adjust for fluctuations in participant attention. Between task blocks, \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nparticipants were instructed to  focus on a fixation cross and let their thoughts wander . \nParticipants were informed regarding the RT threshold manipulation only after completion of \nboth measurements. Full task design details are reported in Hahn et al. [17]. \nPET/MRI data acquisition \nEach subject underwent two 58-minute PET/MR scans using a hybrid PET/MR system (Siemens \nBiograph mMR, Erlangen, Germany  at the Department of Radiology and Nuclear  \nMedicine at the Medical University of Vienna ). As dopamine synthesis rates depend on \ntyrosine plasma levels, participants fasted for at least  four hours before each scan, avoiding \nsweetened beverages and caffeine. One hour before 6-[¹⁸F]FDOPA administration, \nparticipants received 150 mg Carbidopa and 400 mg Entacapone to inhibit peripheral \nmetabolism of the radiotracer  by dopa -decarboxylase and catechol -O-methyl transferase.  \nScans commenced simultaneously with the intravenous administration of 6-[¹⁸F]FDOPA. The \nradiotracer was administered using a bolus 816ml/h for 1 min and constant infusion 39 ml/h \nfor 56 min, via a perfusion pump (Syramed µSP6000 with UniQUE MRI-shield, both Arcomed, \nRegensdorf, Switzerland). \nThe fPET data was acquired in list mode with the examination table moving in a stop -and-go \npattern between brain and thorax field of view (FOV)  as previously published by our group \n(22). This enabled non -invasive quantification of dopamine synthesis rates with a cardiac \nimage-derived input function (IDIF). fPET acquisition started in the thorax F OV to obtain the \nIDIF from the left ventricle, ascending aorta, and descending aorta. After 6 minutes, the \nexamination table was repositioned to capture the brain FOV, collecting baseline and MID task \ndata (4 × 5 min) in a block design. Following each task block, the table moved back to the \nthorax to acquire additional IDIF data points (4 × 30s) (22). This alternating process was \nrepeated several times to ensure the collection of consistent and accurate IDIF, baseline, and \ntask data. fPET scans were obtained at a spatial resolution of (x, y, z) 2.09 × 2.09 × 2.03 mm \nand a matrix size of 344 × 344 × 127 voxels. \nStructural MRI of the brain was acquired before fPET using a T1-weighted MPRAGE sequence \n(TE/TR = 4.21/2200 ms, TI = 900 ms, flip angle = 9°, matrix size = 240 × 256, 160 slices, voxel \nsize = 1 mm isotropic, TA = 7:41 min) for spatial normalization.  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nThe thorax was imaged using a T1 -weighted STARVIBE sequence (TE/TR = 1.44/3050 ms, flip \nangle = 5°, matrix size = 320 × 320, 208 slices, voxel size = 1.19 × 1.19 × 1.2 mm, TA = 5:33 min) \nfor IDIF extraction.  \nMID task functional data were acquired using an  echo planar imaging  (EPI) sequence \n(TE/TR = 30/2000 ms, flip angle = 90°, matrix size = 80 × 80, 34 slices, voxel size = 2.5 × 2.5 × 2.5 \nmm with a 0.825 mm gap). \nImage derived input function (IDIF) \nThe MRI T1 STARVIBE, MRAC and mean PET images were used as templates to manually place \nthree volumes of interest (VOIs) with fixed sizes on the left ventricle, the ascending aorta and \ndescending thoracic aorta. Activity in these VOIs was sampled during our measurement \nprotocol at 5, 18.5, 30, 41.5 and 55.5 min after start of tracer application.  Afterwards, mean \nactivity was extracted from each VOI across time points to derive IDIFs. Composite IDIFs were \nobtained by using the initial peak activity in the left ventricle VOI, as shown to be robust in \nprior studies (22), and by linear fitting the time activity curve in each VOI using 15 time points \nper subject to reduce motion sensitivity. Accuracy of the fit was further improved by \nincorporating venous blood samples collected at 7, 20 .5, 32, 43 .5 and 51.5  minutes, which \nwere temporally aligned with PET frame acquisition via linear interpolation. The resulting IDIFs \nwere then multiplied  with the plasma -to-whole-blood ratio. The latter was fitted as a linear \nfunction. For more information on IDIF acquisition, see Reed et. al. (22). \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\n \n \nFigure 1:  Graphical overview of the fPET/fMRI measurements. The fPET data acquisition \nalways commenced with the PET field of view (FOV) positioned over the thorax (yellow box) for \n6 minutes to capture the initial tracer peak. Afterwards, the table position was then shifted to \nthe brain (green box), where fPET acquisition was initiated while participants fixated on a cross \nand engaged in unconstrained thought. At minute 11  after infusion start , the first MID task \nblock began concurrently with the onset of the fMRI sequence. Manual venous blood samples \nwere obtained at minutes 7, 20.5, 32, 43.5 and 51.5. Upon completion of the brain acquisition, \nthe bed automatically returned to the thorax to collect additional data points for the image -\nderived input function (IDIF) using a stop -and-go motion protocol. This cycle was repeated \nmultiple times to ensure reliable estimation of both IDIF and fPET task-related measures. \nfPET data preprocessing \nfPET list mode data was reconstructed with a Poisson ordered subset expectation \nmaximization algorithm  for each block  (3 iterations, 21 subsets). Concatenation and decay \ncorrection of thorax and brain frames was performed separately at the start of the \nmeasurement. To adjust for the initial tracer kinetics the first thorax block was binned to \n20x5s, 8x10s and 5x30s frames. The later thorax acquisitions were binned to 30s frames. The \nbrain acquisition frames were binned into 30  s or 2 s frames, respectively. Attenuation and \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nscatter correction for the brain scans was performed based on the structural T1 scans in a \npseudo-CT approach (27). The thorax blocks on the other hand were corrected utilizing a \nDIXON MRAC with CAIPIRINHA sampling pattern (28).  \nfPET data was pre -processed as published previously by our team using SPM12 (Welcome \nTrust Centre for Neuroimaging) (29). Brain blocks were corrected for head movement (quality \n= best, registered to mean). After coregistration to the structural T1 scan, the images were \nnormalized to MNI space. Furthermore,  both 30s and 2s frames were smoothed with an \nedNLM Filter (26) with a kernel size of 3 3 x 5 frames  or with a conventional 8mm Gaussian \nkernel. A mask was applied to only include grey matter voxels.  \nTo distinguish baseline metabolism and task dependant effects the fPET toolbox was utilized \n(30). In short, a general linear model was used, including regressors for win - and loss-blocks. \nFor further optimization , the principal components of the six motion parameters which \nexplained the majority of variance , were added as motion regressors. The baseline \nmetabolism was defined as the average activity across grey matter voxels, that were not active \nduring fMRI task performance (contrast success > failure, p < 0.001 uncorrected) and had not \nbeen identified previously as active voxels in a MID -task meta-analysis (31). The single frame \nbefore and after table movement was deweighted to 0.5 to reduce potential movement  \nartifacts and the infusion start was anchored within the GLM  by introducing a value of 0 \nkBq/cm3 at time 0 . The net influx constant (Ki) was estimated using the Gjedde-Patlak plot \nwith the slope fitted from t*=30 min after infusion start. \nBOLD signal changes \nfMRI data was pre-processed as previously published by our team in SPM12 (32). We carried \nout slice timing correction to the middle slice and realignment to the mean image (quality = \n1). Subsequently , BOLD data was normalized to MNI space and smoothed using an 8mm \nGaussian kernel. First level analysis in the general linear model was conducted in a n event-\nrelated design with one regressor for each event (cue win, cue loss, feedback win and \nfeedback loss) and further regressors for head motion, white matter and cerebrospinal fluid. \nAfterwards four contrasts of interest were calculated from the GLM’s beta values: cue win vs \nrest, cue loss vs rest, feedback win vs rest, and feedback loss vs rest. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nRegion of interest definition \nRegions of interest (ROIs) were defined based on our previous 6-[¹⁸F]FDOPA fPET studies and \na large scale meta -analysis of fMRI studies utilising the MID task (5,15). Regions of interest \nchosen for this analysis were the caudate, putamen and NAcc, extracted from the Harvard -\nOxford atlas (33), see figure 2. \n \nFigure 2: Regions of Interest including the nucleus accumbens (red), putamen (pink) and \ncaudate (blue) in coronal and sagittal view, extracted from the Harvard-Oxford atlas (33). \nStatistical analysis \nFor 6-[¹⁸F]FDOPA fPET, the dopamine synthesis rate was estimated by calculating the net influx \nconstant (Ki) to allow comparison with previously published studies. Similarly, fMRI beta \nvalues of the four different contrasts were computed for the same purpose. To quantitatively \nassess the consistency between the two  PET/MRI measurements, the intraclass correlation \ncoefficient (ICC, Equation 1) and the coefficient of variation (CV) was calculated for each \nmodality, region of interest (ROI), and condition. \n \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nResults \nfPET dopamine synthesis \nFor 30 s fPET data after edNLM filtering (see Tables 1 & 2), fair reliability was observed at rest \nin the NAcc (ICC = 0.58; CV = 3.13 ) and the putamen (ICC = 0.47; CV 3.86). Furthermore, fair \nreliability for the 30 s fPET data was observed during the win condition in the caudate (ICC = \n0.54; CV = 23.97) and putamen (ICC = 0.57; CV = 19.24). In contrast, poor reliability was found \nacross all regions during the loss condition  (ICC = 0.30-0.37; CV = 29.83 -37.53), in the NAcc \nduring the win condition (ICC = 0.32; CV = 34.01), and in the caudate at rest (ICC = 0.39; CV = \n7.20).  \nA comparable pattern emerged for the 2 s edNLM fPET data (see Tables 1 & 2). Good reliability \nwas observed at rest in the NAcc (ICC = 0.66; CV = 3.30). Fair reliability was again found during \nthe win condition in the caudate (ICC = 0.57; CV = 23.04 ) and the putamen (ICC = 0.46; CV = \n23.75), as well as at rest in the putamen (ICC = 0.54; CV = 3.79). Poor reliability persisted across \nall regions during the loss condition (ICC = 0.11-0.33; CV = 28.88-40.55) and in the NAcc during \nthe win condition (ICC = 0.29; CV = 35.57). Additionally, poor reliability was observed at rest in \nthe caudate (ICC = 0.39; CV = 5.75). \nFor 2 s fPET data filtered with an 8mm Gaussian kernel, reliability estimates mirrored those of \nthe 2 s edNLM fPET data (see Tables 1 & 2). Fair reliability was seen at rest in the NAcc (ICC = \n0.57; CV = 25.07) and the putamen (ICC = 0.51; CV = 30.23). Further, fair reliability was found \nduring the win condition in the caudate (ICC = 0.54; CV 3.68) and the putamen (ICC = 0.47; CV \n= 43.77). Poor reliability was consistently observed across all regions during the loss condition \n(ICC = 0.08-0.34; CV = 3.82-35.49), in the caudate at rest (ICC = 0.38; CV = 5.78) and in the NAcc \nduring the win condition (ICC = 0.27; CV = 37.41). \nBOLD-derived neuronal activation  \nFor fMRI beta values  (see Tables 1 & 2), good reliability was noted during the feedback win \ncondition in the NAcc (ICC = 0.65; CV = 88.17) and during the feedback loss cond ition in the \nNAcc (ICC = 0.67; CV = 244.48) and caudate (ICC = 0.61; CV = 134.15). While fair reliability was \nfound in the putamen across all conditions (ICC = 0.40-0.45; CV = 46.48-379.33), in the caudate \nduring the feedback win condition (ICC = 0.56; CV = -143.33) and cue win condition (ICC = 0.41; \nCV = 76.70), as well as in the NACC during the cue win condition (ICC = 0.46; CV = 75.76) and \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nthe cue loss condition (ICC = 0.46; CV = 82.74). Poor reliability was observed in the caudate \nduring the cue loss condition (ICC = 0.26; CV = 77.22). \n \nICC  Caudate Putamen N. accumbens \nfMRI Cue win 0.41 0.45 0.46 \n Feedback win 0.56 0.44 0.65 \n Cue loss 0.26 0.40 0.46 \n Feedback loss 0.61 0.44 0.67 \n30s edNLM fPET Rest 0.39 0.47 0.58 \n Win 0.54 0.57 0.32 \n Loss 0.37 0.30 0.31 \n2s edNLM fPET  Rest 0.39 0.54 0.66 \n Win 0.57 0.46 0.29 \n Loss 0.33 0.26 0.11 \n2 s 8 mm  Rest 0.38 0.51 0.57 \nGaussian fPET Win 0.54 0.47 0.27 \n Loss 0.34 0.26 0.08 \nTable 1: Intraclass correlation coefficient (ICC) Values for fMRI beta values  and fPET data for \neach region of interest and condition \nCV  Caudate Putamen N. accumbens \nfMRI Cue win 76.70 47.26 75.76 \n Feedback win -143.33 379.33 88.17 \n Cue loss 77.22 46.48 82.74 \n Feedback loss 134.15 668.63 244.48 \n30s edNLM fPET Rest 7.20 3.86 3.13 \n Win 23.97 19.24 34.01 \n Loss 29.83 32.38 37.53 \n2s edNLM fPET  Rest 5.75 3.79 3.30 \n Win 23.04 23.75 35.57 \n Loss 28.88 33.17 40.55 \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\n2 s 8 mm  Rest 5.77 30.23 25.07 \nGaussian fPET Win 3.68 43.77 37.41 \n Loss 3.82 35.49 25.34 \nTable 2: Within-subject coefficient of variation ( CV) Values for fMRI and fPET data for each \nregion of interest and condition \nDiscussion \nThis study examined the test -retest reliability of task -induced changes in simultaneously \nacquired dopamine synthesis measured with 6-[¹⁸F]FDOPA fPET and BOLD fMRI during the MID \ntask in healthy participants. Our findings reveal region-, condition-, and processing-dependent \nvariability in test -retest measures.  Fair to good  reliability was observed for task -induced \ndopamine synthesis in the  NAcc and putamen during rest , as well as in the  caudate and \nputamen during the win condition, whereas the reliability of task-induced dopamine synthesis \nduring the loss condition was consistently poor. On the other hand, task-induced betas derived \nfrom BOLD fMRI showed good reliability in the NAcc during the feedback conditions and in the \ncaudate during the feedback loss condition. Fair reliability was observed in all other regions \nacross all conditions, apart from poor reliability, which was found in the caudate during the \ncue loss condition. These findings highlight the challenges of assessing reward-related circuitry \nand dopami nergic function during MID task performance but also emphasize the \ncomplementary nature of fMRI and fPET (34). \nThis complementarity arises from the fundamental differences between the two modalities. \n6-[¹⁸F]FDOPA fPET measures dopamine synthesis  at the  molecular level  (16,35,36), \ndistinguishing it from the composite and indirect fMRI BOLD signal, which reflects  \nhemodynamic changes, blood oxygenation, and vascular factors (11,12,37). Yet, this specificity \nis reliant on ionising radiation as well as a more complex and expensive acquisition and analysis \npipeline. Contrary to fPET , BOLD fMRI offers excellent spatial and temporal resolution  \n(11,12,38). However, it i s more sus ceptible to motion-artifacts, physiological noise, and \nscanner-related instabilities (13,14). Furthermore, traditional 6-[¹⁸F]FDOPA fPET captures data \nin 30 s frames  in a block design , potentially limiting the sensitivity to rapid changes in \ndopamine synthesis (15). In our analysis we increased the temporal resolution of 6-[¹⁸F]FDOPA \nfPET to 2 s based on previous work on high temporal resolution fPET with [18F]FDG (23). A key \nfinding of this study is the slight difference in reliability between the 30  s and 2  s edNLM-\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nfiltered fPET data. This indicates that 6-[¹⁸F]FDOPA fPET retains robust reproducibility even at \nhigh temporal resolution, without a significant loss of signal quality or an increase in noise. \nThis capability represents a significant methodological advantage of fPET over classical PET \nimaging approaches, which typically require longer acquisition windows (35,36). The ability to \nresolve dopamine dynamics on the scale of seconds opens new avenues for studying fast task-\nrelated changes in dopaminergic function while maintaining quantitative reliability. This \nobservation will enable the examination of dopamine synthesis in an event related manner in \nthe future enabling to investigate dopamine dynamics during different phases of reward \nprocessing. However, our high temporal resolution 2  s fPET frames were  also analysed in a \nblock design to allow a direct comparison between 30 s and 2 s fPET data after edNLM filtering. \nfMRI on the other hand was analysed in an event related manner allowing to distinguish \nbetween the anticipation and outcome phases of reward processing (31). Our findings suggest \nthat 6-[¹⁸F]FDOPA fPET can achieve reliability comparable to fMRI in reward -related regions \n(e.g., caudate) while offering more direct insight into the underlying neurotransmitter \ndynamics. \nTest–retest reliability of 30 s and 2 s 6-[¹⁸F]FDOPA fPET data processed with edNLM filtering \nwas comparable to that of [¹⁸F] FDG fPET during a cognitive task (25). However, [¹⁸F]FDG fPET \nacquired at rest showed higher reliability than both 30 s and 2 s 6 -[¹⁸F]FDOPA fPET data (25). \nFor fMRI, our reliability estimates were consistent with prior studies using the MID task with \nlarger samples (39), as well as with test-retest findings from other task-based fMRI paradigms \nsuch as finger tapping (40) and decision making under risk (41). \nComparison of edNLM vs. Gaussian Kernel Filtering \nFurthermore, we compared the reliability of different filtering strategies. We previously \nreported that edNLM filtering enhances reliability for task -specific fPET (26). In our current \nanalysis, data filtered with edNLM yielded similar reliability when compared to filtering with a \ntraditional 8mm Gaussian kernel.  The gaussian smoothing is a computationally efficient, \nthough less sophisticated denoising method while edNLM filtering offers a more advanced and \nadaptive approach but requires longer processing time. The comparable reliability of both \nfiltering strategies suggests that for high temporal resolution 6-[¹⁸F]FDOPA fPET the choice of \nfiltering method may be guided by computational resources and research goals rather than by \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nabsolute performance. Here the 8mm Gaussian kernel might offer a simple yet effective \nsolution for large-scale studies or where processing speed is critical. \nLimitations and Future Directions \nOur findings should be interpreted considering certain limitations. Our sample size  of 25 \nhealthy subject was limited, which may have decreased the observed  reliability, particularly \nfor the more variable task conditions.  Furthermore, inter subject variability must be \nconsidered. Subjects were measured approximately 8 weeks apart. Previous literature \nsuggests that multiple variables which may change over the course of days or weeks like sleep \nquality, affective state and stress influence reward-related behaviour (42).  \nFurthermore, a potential practice effect must be considered, since subjects completed the MID \ntask twice. The first time the reaction to the novel task might have elicited a stronger or \ndifferent neural response, potentially reducing the reproducibility of our analysis. However, a \nprevious fMRI study using the MID task found no improvement in MID task performance across \nsessions in healthy subjects, contrary to this prediction (43).  \nAdditionally, it is important to note that 6-[¹⁸F]FDOPA fPET Ki values were analysed using a \nblock design rather than an event -related approach. Event-related analysis of the 30 -second \nframes is not possible because the different phases of reward processing during the MID task, \nsuch as the anticipation and outcome phases, are  much shorter. Nevertheless, since the 2  s \nframes show reliability comparable to that of the 30 s frames, future advancements may allow \nto assess these distinct phases also with fPET in an event-related manner. \nConclusion \nOur study demonstrates moderate region- and condition dependent test-retest reliability of 6-\n[¹⁸F]FDOPA fPET and BOLD fMRI during MID task performance. Neither BOLD fMRI nor 6-\n[¹⁸F]FDOPA fPET showed a clear superior test -retest reliability across all regions of interest \n(caudate, putamen and NAcc). However, 6-[¹⁸F]FDOPA fPET offers molecular specifi city for \ndopamine synthesis  (15,35,36), whereas BOLD fMRI measures a composite signal  (11,12). \nNotably, 6-[¹⁸F]FDOPA fPET retained comparable reliability also for high temporal resolution 2s \nframes suggesting the possibility of further event -related analysis and direct comparison to \nBOLD fMRI at a similar temporal resolution  (11). Thus, o ur findings highlight the \ncomplementary nature of BOLD fMRI and 6-[¹⁸F]FDOPA fPET for investigating reward \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nprocessing and underscore the chance of using high temporal resolution 6-[¹⁸F]FDOPA fPET to \nstudy dopamine synthesis dynamics even closer. \nAcknowledgments \nWe thank the graduated team members and  the diploma students  of the Neuroimaging Lab \n(NIL, headed by R. Lanzenberger)  as well as  the clinical colleagues from the Department of \nPsychiatry and Psychotherapy for clinical and/or administrative support. G. Schlosser, E. Briem, \nA. Mayerweg, L. Artmeier and S. Klug were supported by the MDPhD Excellence Program of \nthe Medical University of Vienna. M.Reed and C. Milz are recipients of a DOC Fellowship of the \nAustrian Academy of Sciences at the Department of Psychiatry an d Psychotherapy, Medical \nUniversity of Vienna. \nAuthor contribution statement \nR.L., M.B.R, A.H., P .A.H, M.H. designed the study. G.S., P .A.H., S.G., S.K., B.E., E.B., A.M., L.A., \nG.M.G., C.S.,  M.M., C.M., P .F., L.N., S.R.  acquired the data. M. B.R. performed data \npreprocessing and analysis. M. B.R and G.S. interpreted the results. G.S. drafted the \nmanuscript. All authors reviewed, edited, and approved the final manuscript. \nEthical considerations \nThe study was approved by the Ethics Committee of the Medical University of Vienna (ethics \nnumber: 2321/2019) and all procedures were carried out in accordance with the Declaration \nof Helsinki.  \nConsent to participate \nAfter detailed explanation of the study protocol, all subjects gave written informed consent. \nAll subjects were insured and reimbursed for their participation. \nConsent for publication \nNot applicable. \nConflict of Interest \nR. Lanzenberger received  investigator-initiated research funding  from Siemens Healthcare \nregarding clinical research using PET/MR.  In the past 3 years he received a travel grant \nfrom Janssen-Cilag Pharma GmbH.  He is a shareholder of the start -up company B M Health \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\nGmbH, Austria since 2019.  M. Hacker received consulting fees and/or honoraria from Bayer \nHealthcare BMS, Eli Lilly, EZAG, GE Healthcare, Ipsen, ITM, Janssen, Roche, Siemens \nHealthineers. All other authors declare no potential conflicts of interest with respect to the \nresearch, authorship, and/or publication of this article. \nFunding \nThis research was funded in whole, or in part, by the Austrian Science Fund (FWF) [grant DOI: \n10.55776/KLI1006 and DOI: 10.55776/PAT6608924, PI: R. Lanzenberger;  DOI: \n10.55776/DOC33; Co-PI: R. Lanzenberger]. For open access purposes, the author has applied \na CC BY public copyright license to any author accepted manuscript version arising from this \nsubmission. \nData availability \n Raw data will not be publicly available due to reasons of data protection. Processed data and \ncustom code can be obtained from the corresponding author with a data sharing agreement, \napproved by the departments of legal affairs and data clearing of  the Medical University of \nVienna. \nReferences \n1. Schultz W, Dayan P , Montague PR. A neural substrate of prediction and reward. Science. 1997 \nMar 14;275(5306):1593–9.  \n2. Beier KT, Steinberg EE, DeLoach KE, Xie S, Miyamichi K, Schwarz L, et al. Circuit Architecture of \nVTA Dopamine Neurons Revealed by Systematic Input–Output Mapping. Cell. 2015 July \n30;162(3):622–34.  \n3. Morales M, Margolis EB. Ventral tegmental area: cellular heterogeneity, connectivity and \nbehaviour. Nat Rev Neurosci. 2017 Feb;18(2):73–85.  \n4. Sesack SR, Grace AA. Cortico-Basal Ganglia Reward Network: Microcircuitry. \nNeuropsychopharmacology. 2010 Jan;35(1):27–47.  \n5. Oldham S, Murawski C, Fornito A, Youssef G, Yücel M, Lorenzetti V. The anticipation and outcome \nphases of reward and loss processing: A neuroimaging meta-analysis of the monetary incentive \ndelay task. Hum Brain Mapp. 2018 Aug;39(8):3398–418.  \n6. García-García I, Horstmann A, Jurado MA, Garolera M, Chaudhry SJ, Margulies DS, et al. Reward \nprocessing in obesity, substance addiction and non-substance addiction. Obes Rev Off J Int Assoc \nStudy Obes. 2014 Nov;15(11):853–69.  \n7. Keren H, O’Callaghan G, Vidal-Ribas P , Buzzell GA, Brotman MA, Leibenluft E, et al. Reward \nProcessing in Depression: A Conceptual and Meta-Analytic Review Across fMRI and EEG Studies. \nAm J Psychiatry. 2018 Nov 1;175(11):1111–20.  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\n8. Bodell LP , Racine SE. A mechanistic staging model of reward processing alterations in individuals \nwith binge-type eating disorders. Int J Eat Disord. 2023 Mar;56(3):516–22.  \n9. Whitton AE, Treadway MT, Pizzagalli DA. Reward processing dysfunction in major depression, \nbipolar disorder and schizophrenia. Curr Opin Psychiatry. 2015 Jan;28(1):7–12.  \n10. Knutson B, Westdorp A, Kaiser E, Hommer D. FMRI visualization of brain activity during a \nmonetary incentive delay task. NeuroImage. 2000 July;12(1):20–7.  \n11. Glover GH. Overview of functional magnetic resonance imaging. Neurosurg Clin N Am. 2011 \nApr;22(2):133–9, vii.  \n12. Logothetis NK, Pfeuffer J. On the nature of the BOLD fMRI contrast mechanism. Magn Reson \nImaging. 2004 Dec;22(10):1517–31.  \n13. Bollmann S, Kasper L, Vannesjo SJ, Diaconescu AO, Dietrich BE, Gross S, et al. Analysis and \ncorrection of field fluctuations in fMRI data using field monitoring. NeuroImage. 2017 July \n1;154:92–105.  \n14. Smith AM, Lewis BK, Ruttimann UE, Ye FQ, Sinnwell TM, Yang Y , et al. Investigation of low \nfrequency drift in fMRI signal. NeuroImage. 1999 May;9(5):526–33.  \n15. Hahn A, Reed MB, Pichler V, Michenthaler P , Rischka L, Godbersen GM, et al. Functional dynamics \nof dopamine synthesis during monetary reward and punishment processing. J Cereb Blood Flow \nMetab Off J Int Soc Cereb Blood Flow Metab. 2021 Nov;41(11):2973–85.  \n16. Cumming P , Léger GC, Kuwabara H, Gjedde A. Pharmacokinetics of plasma 6-[18F]fluoro-L-3,4-\ndihydroxyphenylalanine ([18F]Fdopa) in humans. J Cereb Blood Flow Metab Off J Int Soc Cereb \nBlood Flow Metab. 1993 July;13(4):668–75.  \n17. Molina V, Solera S, Sanz J, Sarramea F, Luque R, Rodríguez R, et al. Association between cerebral \nmetabolic and structural abnormalities and cognitive performance in schizophrenia. Psychiatry \nRes. 2009 Aug 30;173(2):88–93.  \n18. Yehuda R, Harvey PD, Golier JA, Newmark RE, Bowie CR, Wohltmann JJ, et al. Changes in relative \nglucose metabolic rate following cortisol administration in aging veterans with posttraumatic \nstress disorder: an FDG-PET neuroimaging study. J Neuropsychiatry Clin Neurosci. \n2009;21(2):132–43.  \n19. Vlassenko AG, Rundle MM, Mintun MA. Human brain glucose metabolism may evolve during \nactivation: findings from a modified FDG PET paradigm. NeuroImage. 2006 Dec;33(4):1036–41.  \n20. Villien M, Wey HY , Mandeville JB, Catana C, Polimeni JR, Sander CY , et al. Dynamic Functional \nImaging of Brain Glucose Utilization using fPET-FDG. NeuroImage. 2014 Oct 15;100:192–9.  \n21. Hahn A, Gryglewski G, Nics L, Hienert M, Rischka L, Vraka C, et al. Quantification of Task-Specific \nGlucose Metabolism with Constant Infusion of 18F-FDG. J Nucl Med Off Publ Soc Nucl Med. 2016 \nDec;57(12):1933–40.  \n22. Reed MB, Handschuh PA, Schmidt C, Murgaš M, Gomola D, Milz C, et al. Validation of cardiac \nimage-derived input functions for functional PET quantification. Eur J Nucl Med Mol Imaging. \n2024 July;51(9):2625–37.  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\n23. Hahn A, Reed MB, Vraka C, Godbersen GM, Klug S, Komorowski A, et al. High-temporal resolution \nfunctional PET/MRI reveals coupling between human metabolic and hemodynamic brain \nresponse. Eur J Nucl Med Mol Imaging. 2024 Apr 1;51(5):1310–22.  \n24. Hahn A, Reed MB, Pichler V, Michenthaler P , Rischka L, Godbersen GM, et al. Functional dynamics \nof dopamine synthesis during monetary reward and punishment processing. J Cereb Blood Flow \nMetab Off J Int Soc Cereb Blood Flow Metab. 2021 Nov;41(11):2973–85.  \n25. Rischka L, Godbersen GM, Pichler V, Michenthaler P , Klug S, Klöbl M, et al. Reliability of task-\nspecific neuronal activation assessed with functional PET, ASL and BOLD imaging. J Cereb Blood \nFlow Metab Off J Int Soc Cereb Blood Flow Metab. 2021 Nov;41(11):2986–99.  \n26. Reed MB, León MP de, Klug S, Milz C, Silberbauer LR, Falb P , et al. Optimal filtering strategies for \ntask-specific functional PET imaging [Internet]. bioRxiv; 2024 [cited 2024 Aug 12]. p. \n2024.04.25.591053. Available from: \nhttps://www.biorxiv.org/content/10.1101/2024.04.25.591053v1 \n27. Burgos N, Cardoso MJ, Thielemans K, Modat M, Pedemonte S, Dickson J, et al. Attenuation \ncorrection synthesis for hybrid PET-MR scanners: application to brain studies. IEEE Trans Med \nImaging. 2014 Dec;33(12):2332–41.  \n28. Wright KL, Harrell MW, Jesberger JA, Landeras L, Nakamoto DA, Thomas S, et al. Clinical \nevaluation of CAIPIRINHA: comparison against a GRAPPA standard. J Magn Reson Imaging JMRI. \n2014 Jan;39(1):189–94.  \n29. Klug S, Godbersen GM, Rischka L, Wadsak W, Pichler V, Klöbl M, et al. Learning induces \ncoordinated neuronal plasticity of metabolic demands and functional brain networks. Commun \nBiol. 2022 May 9;5(1):428.  \n30. Hahn A, Reed MB, Milz C, Falb P , Murgaš M, Lanzenberger R. A Unified Approach for Identifying \nPET-based Neuronal Activation and Molecular Connectivity with the functional PET toolbox \n[Internet]. bioRxiv; 2024 [cited 2025 Aug 13]. p. 2024.11.13.623377. Available from: \nhttps://www.biorxiv.org/content/10.1101/2024.11.13.623377v1 \n31. Oldham S, Murawski C, Fornito A, Youssef G, Yücel M, Lorenzetti V. The anticipation and outcome \nphases of reward and loss processing: A neuroimaging meta-analysis of the monetary incentive \ndelay task. Hum Brain Mapp. 2018 Aug;39(8):3398–418.  \n32. Rischka L, Gryglewski G, Pfaff S, Vanicek T, Hienert M, Klöbl M, et al. Reduced task durations in \nfunctional PET imaging with [18F]FDG approaching that of functional MRI. NeuroImage. 2018 \nNov 1;181:323–30.  \n33. Rushmore RJ, Bouix S, Kubicki M, Rathi Y , Yeterian E, Makris N. HOA2.0-ComPaRe: A next \ngeneration Harvard-Oxford Atlas comparative parcellation reasoning method for human and \nmacaque individual brain parcellation and atlases of the cerebral cortex. Front Neuroanat. \n2022;16:1035420.  \n34. Ehman EC, Johnson GB, Villanueva-Meyer JE, Cha S, Leynes AP , Larson PEZ, et al. PET/MRI: Where \nmight it replace PET/CT? J Magn Reson Imaging JMRI. 2017 Nov;46(5):1247–62.  \n35. van Hooijdonk CFM, van der Pluijm M, Smith C, Yaqub M, van Velden FHP , Horga G, et al. Striatal \ndopamine synthesis capacity and neuromelanin in the substantia nigra: A multimodal imaging \nstudy in schizophrenia and healthy controls. Neurosci Appl. 2023;2:101134.  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint \n\n36. Sayalı C, van den Bosch R, Määttä JI, Hofmans L, Papadopetraki D, Booij J, et al. Methylphenidate \nundermines or enhances divergent creativity depending on baseline dopamine synthesis \ncapacity. Neuropsychopharmacol Off Publ Am Coll Neuropsychopharmacol. 2023 \nDec;48(13):1849–58.  \n37. Uludağ K. Physiological modeling of the BOLD signal and implications for effective connectivity: A \nprimer. NeuroImage. 2023 Aug 15;277:120249.  \n38. Goense J, Bohraus Y , Logothetis NK. fMRI at High Spatial Resolution: Implications for BOLD-\nModels. Front Comput Neurosci. 2016;10:66.  \n39. Demidenko MI, Mumford JA, Poldrack RA. Impact of analytic decisions on test-retest reliability of \nindividual and group estimates in functional magnetic resonance imaging: a multiverse analysis \nusing the monetary incentive delay task. BioRxiv Prepr Serv Biol. 2024 July 9;2024.03.19.585755.  \n40. Wüthrich F, Lefebvre S, Nadesalingam N, Bernard JA, Mittal VA, Shankman SA, et al. Test-retest \nreliability of a finger-tapping fMRI task in a healthy population. Eur J Neurosci. 2023 \nJan;57(1):78–90.  \n41. Korucuoglu O, Harms MP , Astafiev SV, Kennedy JT , Golosheykin S, Barch DM, et al. Test-retest \nreliability of fMRI-measured brain activity during decision making under risk. NeuroImage. 2020 \nJuly 1;214:116759.  \n42. Wieman ST, Arditte Hall KA, MacDonald HZ, Gallagher MW, Suvak MK, Rando AA, et al. \nRelationships Among Sleep Disturbance, Reward System Functioning, Anhedonia, and Depressive \nSymptoms. Behav Ther. 2022 Jan;53(1):105–18.  \n43. Rapuano KM, Conley MI, Juliano AC, Conan GM, Maza MT, Woodman K, et al. An open-access \naccelerated adult equivalent of the ABCD Study neuroimaging dataset (a-ABCD). NeuroImage. \n2022 July 15;255:119215.  \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 4, 2025. ; https://doi.org/10.1101/2025.11.03.686252doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}