Antonio Augimeri

Machine learning on brain MRI data for differential diagnosis of Parkinson's disease and Progressive Supranuclear Palsy

BACKGROUND Supervised machine learning has been proposed as a revolutionary approach for identifying sensitive medical image biomarkers (or combination of them) allowing for automatic diagnosis of individual subjects. The aim of this work was to assess the feasibility of a supervised machine learning algorithm for the assisted diagnosis of patients with clinically diagnosed Parkinsons disease (PD) and Progressive Supranuclear Palsy (PSP). METHOD Morphological T1-weighted Magnetic Resonance Images (MRIs) of PD patients (28), PSP patients (28) and healthy control subjects (28) were used by a supervised machine learning algorithm based on the combination of Principal Components Analysis as feature extraction technique and on Support Vector Machines as classification algorithm. The algorithm was able to obtain voxel-based morphological biomarkers of PD and PSP. RESULTS The algorithm allowed individual diagnosis of PD versus controls, PSP versus controls and PSP versus PD with an Accuracy, Specificity and Sensitivity>90%. Voxels influencing classification between PD and PSP patients involved midbrain, pons, corpus callosum and thalamus, four critical regions known to be strongly involved in the pathophysiological mechanisms of PSP. COMPARISON WITH EXISTING METHODS Classification accuracy of individual PSP patients was consistent with previous manual morphological metrics and with other supervised machine learning application to MRI data, whereas accuracy in the detection of individual PD patients was significantly higher with our classification method. CONCLUSIONS The algorithm provides excellent discrimination of PD patients from PSP patients at an individual level, thus encouraging the application of computer-based diagnosis in clinical practice.

Automatic Detection of White Matter Hyperintensities in Healthy Aging and Pathology Using Magnetic Resonance Imaging: A Review

White matter hyperintensities (WMH) are commonly seen in the brain of healthy elderly subjects and patients with several neurological and vascular disorders. A truly reliable and fully automated method for quantitative assessment of WMH on magnetic resonance imaging (MRI) has not yet been identified. In this paper, we review and compare the large number of automated approaches proposed for segmentation of WMH in the elderly and in patients with vascular risk factors. We conclude that, in order to avoid artifacts and exclude the several sources of bias that may influence the analysis, an optimal method should comprise a careful preprocessing of the images, be based on multimodal, complementary data, take into account spatial information about the lesions and correct for false positives. All these features should not exclude computational leanness and adaptability to available data.

Prefrontal thickening in PD with levodopa-induced dyskinesias: New evidence from cortical thickness measurement

PURPOSE Neurodegenerative processes in Parkinsons disease (PD) patients with levodopa-induced dyskinesias (LID) are still a matter of debate. Recently, we demonstrated that this clinical phenotype is associated with an abnormal gray matter increase in the prefrontal cortex when compared to PD without LID. This evidence was found by using voxel-based morphometry (VBM). However, VBM may not be the most appropriate procedure to assess cortical pathology, since its normalization/smoothing steps reduce the ability to anatomically characterize sulci and gyri. The aim of this study is to better delineate the LID-related anatomical abnormalities by using an advanced neuroimaging method that provides a direct and objective measure of the cortical morphology. METHODS Surface-based investigation of cortical mantle (cortical thickness) was carried out by using Freesurfer in two groups of treated PD patients with LID (no 29) and without LID (no 30), and one group of age- and sex-matched controls (no 24). RESULTS Cortical thickness analysis revealed a pronounced increase of thickness in the right inferior frontal sulcus in PD patients with LID with respect to PD patients without LID. DISCUSSION The current study confirms our previous morphological findings on the role of the prefrontal cortex in the pathophysiology of LID and delineates with more precision the anatomical abnormalities characterizing this clinical phenotype.

A Cellular Neural Network methodology for the automated segmentation of multiple sclerosis lesions

We present a new application based on genetic algorithms (GAs) that evolves a Cellular Neural Network (CNN) capable of automatically determining the lesion load in multiple sclerosis (MS) patients from magnetic resonance imaging (MRI). In particular, it seeks to identify brain areas affected by lesions, whose presence is revealed by areas of higher intensity if compared to healthy tissue. The performance of the CNN algorithm has been quantitatively evaluated by comparing the CNN output with the experts manual delineation of MS lesions. The CNN algorithm was run on a data set of 11 MS patients; for each one a single dataset of MRI images (matrix resolution of 256×256 pixels) was acquired. Our automated approach gives satisfactory results showing that after the learning process the CNN is capable of detecting MS lesions with different shapes and intensities (mean DICE coefficient=0.64). The system could provide a useful support tool for the evaluation of lesions in MS patients, although it needs to be evolved and developed in the future.

Role of the Insula and Vestibular System in Patients with Chronic Subjective Dizziness: An fMRI Study Using Sound-Evoked Vestibular Stimulation

Chronic subjective dizziness (CSD) is a common vestibular disorder characterized by persistent non-vertiginous dizziness, unsteadiness, and heightened sensitivity to motion stimuli that may last for months to years after events that cause acute vestibular symptoms or disrupt balance. CSD is not associated with abnormalities of basic vestibular or oculomotor reflexes. Rather, it is thought to arise from persistent use of high-threat postural control strategies and greater reliance on visual cues for spatial orientation (i.e., visual dependence), long after triggering events resolve. Anxiety-related personality traits confer vulnerability to CSD. Anomalous interactions between the central vestibular system and neural structures related to anxiety may sustain it. Vestibular- and anxiety-related processes overlap in the brain, particularly in the insula and hippocampus. Alterations in activity and connectivity in these brain regions in response to vestibular stimuli may be the neural basis of CSD. We examined this hypothesis by comparing brain activity from 18 patients with CSD and 18 healthy controls measured by functional magnetic resonance imaging during loud short tone bursts, which are auditory stimuli that evoke robust vestibular responses. Relative to controls, patients with CSD showed reduced activations to sound-evoked vestibular stimulation in the parieto-insular vestibular cortex (PIVC) including the posterior insula, and in the anterior insula, inferior frontal gyrus, hippocampus, and anterior cingulate cortex. Patients with CSD also showed altered connectivity between the anterior insula and PIVC, anterior insula and middle occipital cortex, hippocampus and PIVC, and anterior cingulate cortex and PIVC. We conclude that reduced activation in PIVC, hippocampus, anterior insula, inferior frontal gyrus, and anterior cingulate cortex, as well as connectivity changes among these regions, may be linked to long-term vestibular symptoms in patients with CSD. Furthermore, altered connectivity between the anterior insula and middle occipital cortex may underlie the greater reliance on visual cues for spatial orientation in CSD patients relative to controls.

Magnetic resonance support vector machine discriminates between Parkinson disease and progressive supranuclear palsy

The aim of the current study was to distinguish patients with Parkinson disease (PD) from those with progressive supranuclear palsy (PSP) at the individual level using pattern recognition of magnetic resonance imaging data.

The relationship between regional microstructural abnormalities of the corpus callosum and physical and cognitive disability in relapsing-remitting multiple sclerosis.

Significant corpus callosum (CC) involvement has been found in relapsing–remitting multiple sclerosis (RRMS), even if conventional magnetic resonance imaging measures have shown poor correlation with clinical disability measures. In this work, we tested the potential of multimodal imaging of the entire CC to explain physical and cognitive disability in 47 patients with RRMS. Values of thickness, fractional anisotropy (FA) and mean diffusivity (MD) were extracted from 50 regions of interest (ROIs) sampled along the bundle. The relationships between clinical, neuropsychological and imaging variables were assessed by using Spearmans correlation. Multiple linear regression analysis was employed in order to identify the relative importance of imaging metrics in modeling different clinical variables. Regional fiber composition of the CC differentially explained the response variables (Expanded Disability Status Scale [EDSS], cognitive impairment). Increases in EDSS were explained by reductions in CC thickness and MD. Cognitive impairment was mainly explained by FA reductions in the genu and splenium. Regional CC imaging properties differentially explained disability within RRMS patients revealing strong, distinct patterns of correlation with clinical and cognitive status of patients affected by this specific clinical phenotype.

The motor inhibition system in Parkinson's disease with levodopa-induced dyskinesias.

Parkinsons disease is primarily a disorder of response initiation characterized by an excessive motor inhibition, whereas levodopa‐induced dyskinesias are clearly a clinical expression of disinhibition of movements.

Hippocampal Subfield Atrophies in Converted and Not-Converted Mild Cognitive Impairments Patients by a Markov Random Fields Algorithm

Although measurement of total hippocampal volume is considered as an important hallmark of Alzheimers disease (AD), recent evidence demonstrated that atrophies of hippocampal subregions might be more sensitive in predicting this neurodegenerative disease. The vast majority of neuroimaging papers investigating this topic are focused on the difference between AD and patients with mild cognitive impairment (MCI), not considering the impact of MCI patients who will or not convert in AD. For this reason, the aim of this study was to determine if measurements of hippocampal subfields provide advantages over total hippocampal volume for discriminating these groups. Hippocampal subfields volumetry was extracted in 55 AD, 32 converted and 89 not-converted MCI (c/nc-MCI) and 47 healthy controls, using an atlas-based automatic algorithm based on Markov random fields embedded in the Freesurfer framework. To evaluate the impact of hippocampal atrophy in discriminating the insurgence of AD-like phenotypes we used three classification methods: Support Vector Machine, Naïve Bayesian Classifier and Neural Networks Classifier. Taking into account only the total hippocampal volume, all classification models, reached a sensitivity of about 66% in discriminating between c-MCI and nc-MCI. Otherwise, classification analysis considering all segmenting subfields increased accuracy to diagnose c-MCI from 68% to 72%. This effect resulted to be strongly dependent upon atrophies of the subiculum and presubiculum. Our multivariate analysis revealed that the magnitude of the difference considering hippocampal subfield volumetry, as segmented by the considered atlas-based automatic algorithm, offers an advantage over hippocampal volume in distinguishing early AD from nc-MCI.

Neuroticism modulates brain visuo-vestibular and anxiety systems during a virtual rollercoaster task

Different lines of research suggest that anxiety‐related personality traits may influence the visual and vestibular control of balance, although the brain mechanisms underlying this effect remain unclear. To our knowledge, this is the first functional magnetic resonance imaging (fMRI) study that investigates how individual differences in neuroticism and introversion, two key personality traits linked to anxiety, modulate brain regional responses and functional connectivity patterns during a fMRI task simulating self‐motion. Twenty‐four healthy individuals with variable levels of neuroticism and introversion underwent fMRI while performing a virtual reality rollercoaster task that included two main types of trials: (1) trials simulating downward or upward self‐motion (vertical motion), and (2) trials simulating self‐motion in horizontal planes (horizontal motion). Regional brain activity and functional connectivity patterns when comparing vertical versus horizontal motion trials were correlated with personality traits of the Five Factor Model (i.e., neuroticism, extraversion‐introversion, openness, agreeableness, and conscientiousness). When comparing vertical to horizontal motion trials, we found a positive correlation between neuroticism scores and regional activity in the left parieto‐insular vestibular cortex (PIVC). For the same contrast, increased functional connectivity between the left PIVC and right amygdala was also detected as a function of higher neuroticism scores. Together, these findings provide new evidence that individual differences in personality traits linked to anxiety are significantly associated with changes in the activity and functional connectivity patterns within visuo‐vestibular and anxiety‐related systems during simulated vertical self‐motion. Hum Brain Mapp 38:715–726, 2017.