Machine Learning for Early Parkinson's Disease Identification within SWEDD Group Using Clinical and DaTSCAN SPECT Imaging Features.

Early Parkinson's Disease (PD) diagnosis is a critical challenge in the treatment process. Meeting this challenge allows appropriate planning for patients. However, Scan Without Evidence of Dopaminergic Deficit (SWEDD) is a heterogeneous group of PD patients and Healthy Controls (HC) in clinical and imaging features. The application of diagnostic tools based on Machine Learning (ML) comes into play here as they are capable of distinguishing between HC subjects and PD patients within an SWEDD group. In the present study, three ML algorithms were used to separate PD patients from HC within an SWEDD group. Data of 548 subjects were firstly analyzed by Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) techniques. Using the best reduction technique result, we built the following clustering models: Density-Based Spatial (DBSCAN), K-means and Hierarchical Clustering. According to our findings, LDA performs better than PCA; therefore, LDA was used as input for the clustering models. The different models' performances were assessed by comparing the clustering algorithms outcomes with the ground truth after a follow-up. Hierarchical Clustering surpassed DBSCAN and K-means algorithms by 64%, 78.13% and 38.89% in terms of accuracy, sensitivity and specificity. The proposed method demonstrated the suitability of ML models to distinguish PD patients from HC subjects within an SWEDD group.

Principal Component Analysis versus Subject's Residual Profile Analysis for Neuroinflammation Investigation in Parkinson Patients: A PET Brain Imaging Study.

Dysfunction of neurons in the central nervous system is the primary pathological feature of Parkinson's disease (PD). Despite different triggering, emerging evidence indicates that neuroinflammation revealed through microglia activation is critical for PD. Moreover, recent investigations sought a potential relationship between Lrrk2 genetic mutation and microglia activation. In this paper, neuroinflammation in sporadic PD, Lrrk2-PD and unaffected Lrrk2 mutation carriers were investigated. The principal component analysis (PCA) and the subject's residual profile (SRP) techniques were performed on multiple groups and regions of interest in 22 brain-regions. The 11C-PBR28 binding profiles were compared in four genotypes depending on groups, i.e., HC, sPD, Lrrk2-PD and UC, using the PCA and SPR scores. The genotype effect was found as a principal feature of group-dependent 11C-PBR28 binding, and preliminary evidence of a MAB-Lrrk2 mutation interaction in manifest Parkinson's and subjects at risk was found.

Investigation of serotonergic Parkinson's disease-related covariance pattern using [11C]-DASB/PET.

We used positron emission tomography imaging with [11C]-3-amino-4-(2-dimethylaminomethylphenylsulfanyl)- benzonitrile (DASB) and principal component analysis to investigate whether a specific Parkinson's disease (PD)-related spatial covariance pattern could be identified for the serotonergic system. We also explored if non-manifesting leucine-rich repeat kinase 2 (LRRK2) mutation carriers, with normal striatal dopaminergic innervation as measured with [11C]-dihydrotetrabenazine (DTBZ), exhibit a distinct spatial covariance pattern compared to healthy controls and subjects with manifest PD. 15 subjects with sporadic PD, eight subjects with LRRK2 mutation-associated PD, nine LRRK2 non-manifesting mutation carriers, and nine healthy controls participated in the study. The analysis was applied to the DASB non-displaceable binding potential values evaluated in 42 pre-defined regions of interest. PD was found to be associated with a specific spatial covariance pattern, comprising relatively decreased DASB binding in the caudate, putamen and substantia nigra and relatively preserved binding in the hypothalamus and hippocampus; the expression of this pattern in PD subjects was significantly higher than in healthy controls (P < 0.001) and correlated significantly with disease duration (P < 0.01) and with DTBZ binding in the more affected putamen (P < 0.01). The LRRK2 non-manifesting mutation carriers expressed a different pattern, also significantly different from healthy controls (P < 0.001), comprising relatively decreased DASB binding in the pons, pedunculopontine nucleus, thalamus and rostral raphe nucleus, and with relatively preserved binding in the hypothalamus, amygdala, hippocampus and substantia nigra. This pattern was not present in either sporadic or LRRK2 mutation-associated PD subjects. These findings, although obtained with a relatively limited number of subjects, suggest that specific and overall distinct spatial serotonergic patterns may be associated with PD and LRRK2 mutations. Alterations in regions where relative upregulation is observed in both patterns may be indicative of compensatory mechanisms preceding or protecting from disease manifestation.

The effect of LRRK2 mutations on the cholinergic system in manifest and premanifest stages of Parkinson's disease: a cross-sectional PET study.

BACKGROUND: Markers of neuroinflammation are increased in some patients with LRRK2 Parkinson's disease compared with individuals with idiopathic Parkinson's disease, suggesting possible differences in disease pathogenesis. Previous PET studies have suggested amplified dopamine turnover and preserved serotonergic innervation in LRRK2 mutation carriers. We postulated that patients with LRRK2 mutations might show abnormalities of central cholinergic activity, even before the diagnosis of Parkinson's disease. METHODS: Between June, 2009, and December, 2015, we recruited participants from four movement disorder clinics in Canada, Norway, and the USA. Patients with Parkinson's disease were diagnosed by movement disorder neurologists on the basis of the UK Parkinson's Disease Society Brain Bank criteria. LRRK2 carrier status was confirmed by bidirectional Sanger sequencing. We used the PET tracer N-11C-methyl-piperidin-4-yl propionate to scan for acetylcholinesterase activity. The primary outcome measure was rate of acetylcholinesterase hydrolysis, calculated using the striatal input method. We compared acetylcholinesterase hydrolysis rates between groups using ANCOVA, with adjustment for age based on the results of linear regression analysis. FINDINGS: We recruited 14 patients with LRRK2 Parkinson's disease, 16 LRRK2 mutation carriers without Parkinson's disease, eight patients with idiopathic Parkinson's disease, and 11 healthy controls. We noted significant between-group differences in rates of acetylcholinesterase hydrolysis in cortical regions (average cortex p=0·009, default mode network-related regions p=0·006, limbic network-related regions p=0·020) and the thalamus (p=0·008). LRRK2 mutation carriers without Parkinson's disease had increased acetylcholinesterase hydrolysis rates compared with healthy controls in the cortex (average cortex, p=0·046). Patients with LRRK2 Parkinson's disease had significantly higher acetylcholinesterase activity in some cortical regions (average cortex p=0·043, default mode network-related regions p=0·021) and the thalamus (thalamus p=0·004) compared with individuals with idiopathic disease. Acetylcholinesterase hydrolysis rates in healthy controls were correlated inversely with age. INTERPRETATION: LRRK2 mutations are associated with significantly increased cholinergic activity in the brain in mutation carriers without Parkinson's disease compared with healthy controls and in LRRK2 mutation carriers with Parkinson's disease compared with individuals with idiopathic disease. Changes in cholinergic activity might represent early and sustained attempts to compensate for LRRK2-related dysfunction, or alteration of acetylcholinesterase in non-neuronal cells. FUNDING: Michael J Fox Foundation, National Institutes of Health, and Pacific Alzheimer Research Foundation.

Feasibility study of TSPO quantification with [18F]FEPPA using population-based input function.

PURPOSE: The input function (IF) is a core element in the quantification of Translocator protein 18 kDa with positron emission tomography (PET), as no suitable reference region with negligible binding has been identified. Arterial blood sampling is indeed needed to create the IF (ASIF). In the present manuscript we study individualization of a population based input function (PBIF) with a single arterial manual sample to estimate total distribution volume (VT) for [18F]FEPPA and to replicate previously published clinical studies in which the ASIF was used. METHODS: The data of 3 previous [18F]FEPPA studies (39 of healthy controls (HC), 16 patients with Parkinson's disease (PD) and 18 with Alzheimer's disease (AD)) was reanalyzed with the new approach. PBIF was used with the Logan graphical analysis (GA) neglecting the vascular contribution to estimate VT. Time of linearization of the GA was determined with the maximum error criteria. The optimal calibration of the PBIF was determined based on the area under the curve (AUC) of the IF and the agreement range of VT between methods. The shape of the IF between groups was studied while taking into account genotyping of the polymorphism (rs6971). RESULTS: PBIF scaled with a single value of activity due to unmetabolized radioligand in arterial plasma, calculated as the average of a sample taken at 60 min and a sample taken at 90 min post-injection, yielded a good interval of agreement between methods and optimized the area under the curve of IF. In HC, gray matter VTs estimated by PBIF highly correlated with those using the standard method (r2 = 0.82, p = 0.0001). Bland-Altman plots revealed PBIF slightly underestimates (~1 mL/cm3) VT calculated by ASIF (including a vascular contribution). It was verified that the AUC of the ASIF were independent of genotype and disease (HC, PD, and AD). Previous clinical results were replicated using PBIF but with lower statistical power. CONCLUSION: A single arterial blood sample taken 75 minute post-injection contains enough information to individualize the IF in the groups of subjects studied; however, the higher variability produced requires an increase in sample size to reach the same effect size.

Imaging Striatal Microglial Activation in Patients with Parkinson's Disease.

This study investigated whether the second-generation translocator protein 18kDa (TSPO) radioligand, [18F]-FEPPA, could be used in neurodegenerative parkinsonian disorders as a biomarker for detecting neuroinflammation in the striatum. Neuroinflammation has been implicated as a potential mechanism for the progression of Parkinson's disease (PD). Positron Emission Tomography (PET) radioligand targeting for TSPO allows for the quantification of neuroinflammation in vivo. Based on genotype of the rs6791 polymorphism in the TSPO gene, 16 mixed-affinity binders (MABs) (8 PD and age-matched 8 healthy controls (HCs)), 16 high-affinity binders (HABs) (8 PD and age-matched 8 HCs) and 4 low-affinity binders (LABs) (3 PD and 1 HCs) were identified. Total distribution volume (VT) values in the striatum were derived from a two-tissue compartment model with arterial plasma as an input function. There was a significant main effect of genotype on [18F]-FEPPA VT values in the caudate nucleus (p = 0.001) and putamen (p < 0.001), but no main effect of disease or disease x genotype interaction in either ROI. In the HAB group, the percentage difference between PD and HC was 16% in both caudate nucleus and putamen; in the MAB group, it was -8% and 3%, respectively. While this PET study showed no evidence of increased striatal TSPO expression in PD patients, the current findings provide some insights on the possible interactions between rs6791 polymorphism and neuroinflammation in PD.

Image derived input function for [18F]-FEPPA: application to quantify translocator protein (18 kDa) in the human brain.

In [18F]-FEPPA positron emission topography (PET) imaging, automatic blood sampling system (ABSS) is currently the gold standard to obtain the blood time activity curve (TAC) required to extract the input function (IF). Here, we compare the performance of two image-based methods of IF extraction to the ABSS gold standard method for the quantification of translocator protein (TSPO) in the human brain. The IFs were obtained from a direct delineation of the internal carotid signal (CS) and a new concept of independent component analysis (ICA). PET scans were obtained from 18 healthy volunteers. The estimated total distribution volume (V(T)) by CS-IF and ICA-IF were compared to the reference V(T) obtained by ABSS-IF in the frontal and temporal cortex, cerebellum, striatum and thalamus regions. The V(T) values estimated using ICA-IF were more reliable than CS-IF for all brain regions. Specifically, the slope regression in the frontal cortex with ICA-IF was r²â€Š= 0.91 (p<0.05), and r²â€Š= 0.71 (p<0.05) using CS-IF.

Dynamic cardiac PET imaging: extraction of time-activity curves using ICA and a generalized Gaussian distribution model.

Kinetic modeling of metabolic and physiologic cardiac processes in small animals requires an input function (IF) and a tissue time-activity curves (TACs). In this paper, we present a mathematical method based on independent component analysis (ICA) to extract the IF and the myocardium's TACs directly from dynamic positron emission tomography (PET) images. The method assumes a super-Gaussian distribution model for the blood activity, and a sub-Gaussian distribution model for the tissue activity. Our appreach was applied on 22 PET measurement sets of small animals, which were obtained from the three most frequently used cardiac radiotracers, namely: desoxy-fluoro-glucose ((18)F-FDG), [(13)N]-ammonia, and [(11)C]-acetate. Our study was extended to PET human measurements obtained with the Rubidium-82 ((82) Rb) radiotracer. The resolved mathematical IF values compare favorably to those derived from curves extracted from regions of interest (ROI), suggesting that the procedure presents a reliable alternative to serial blood sampling for small-animal cardiac PET studies.