Evaluation of machine learning algorithms performance for the prediction of early multiple sclerosis from resting-state FMRI connectivity data.

Machine Learning application on clinical data in order to support diagnosis and prognostic evaluation arouses growing interest in scientific community. However, choice of right algorithm to use was fundamental to perform reliable and robust classification. Our study aimed to explore if different kinds of Machine Learning technique could be effective to support early diagnosis of Multiple Sclerosis and which of them presented best performance in distinguishing Multiple Sclerosis patients from control subjects. We selected following algorithms: Random Forest, Support Vector Machine, Naïve-Bayes, K-nearest-neighbor and Artificial Neural Network. We applied the Independent Component Analysis to resting-state functional-MRI sequence to identify brain networks. We found 15 networks, from which we extracted the mean signals used into classification. We performed feature selection tasks in all algorithms to obtain the most important variables. We showed that best discriminant network between controls and early Multiple Sclerosis, was the sensori-motor I, according to early manifestation of motor/sensorial deficits in Multiple Sclerosis. Moreover, in classification performance, Random Forest and Support Vector Machine showed same 5-fold cross-validation accuracies (85.7%) using only this network, resulting to be best approaches. We believe that these findings could represent encouraging step toward the translation to clinical diagnosis and prognosis.

Multimodal assessment of normal-appearing corpus callosum is a useful marker of disability in relapsing-remitting multiple sclerosis: an MRI cluster analysis study.

BACKGROUND AND PURPOSE: Corpus callosum (CC) is frequently involved in relapsing-remitting multiple sclerosis (RRMS). Magnetic resonance imaging (MRI) and diffusion tensor imaging (DTI) allow to study CC macrostructural and microstructural tissue integrity. Here, we applied a data-driven approach to MRI and DTI data of normal-appearing CC in RRMS subjects, and subsequently evaluated if differences in tissue integrity corresponded to different levels of physical disability and cognitive impairment. METHODS: 74 RRMS patients and 20 healthy controls (HC) underwent 3 T MRI and DTI. Thickness and fractional anisotropy (FA) along midsagittal CC were extracted, and values from RRMS patients were fed to a hierarchical clustering algorithm. We then used ANOVA to test for differences in clinical and cognitive variables across the imaging-based clusters and HC. RESULTS: We found three distinct MRI-based subgroups of RRMS patients with increasing severity of CC damage. The first subgroup showed callosal integrity similar to HC (Cluster 1); Cluster 2 had milder callosal damage; a third subgroup showed the most severe callosal damage (Cluster 3). Cluster 3 included patients with longer disease duration and worst scores in Expanded Disability Status Scale. Cognitive domains of verbal memory, executive functions and processing speed were impaired in Cluster 3 and Cluster 2 compared to Cluster 1 and HC. CONCLUSIONS: Within the same homogeneous cohort of patients, we could identify three neuroimaging RRMS clusters characterized by different involvement of normal-appearing CC. Interestingly, these corresponded to three distinct levels of clinical and cognitive disability.

Hepatic microabscesses during CMV reactivation in a multiple sclerosis patient after alemtuzumab treatment.

The anti-CD52 monoclonal antibody alemtuzumab is a highly active treatment for multiple sclerosis (MS) causing rapid depletion of B and T lymphocytes with nadir one month after last infusion. Opportunistic Cytomegalovirus (CMV) infections have been reported in MS patients treated with this drug. We report one patient who developed a CMV reactivation with hepatic involvement three weeks after the first cycle of alemtuzumab. This patient, promptly diagnosed and treated, achieved a complete recovery with valganciclovir. The possibility of this treatable opportunistic infection should be considered by neurologists in febrile patients with hepatic markers alteration after treatment with alemtuzumab.

Diffusion tensor MRI changes in gray structures of the frontal-subcortical circuits in amyotrophic lateral sclerosis.

In this study, we used an automated segmentation of regions of interest and co-registration to diffusion tensor imaging (DTI) images to investigate whether microstructural abnormalities occur in gray structures of the frontal-subcortical circuits in patients with amyotrophic lateral sclerosis (ALS). Twenty-four patients with probable or definite sporadic ALS and 22 healthy controls were enrolled in the study. Thirteen out of 24 ALS patients and all of the control subjects underwent a detailed neuropsychological evaluation. DTI was performed to measure mean diffusivity (MD) and fractional anisotropy in the frontal cortex, caudate, putamen, globus pallidus, thalamus, amygdala and hippocampus. MD values of ALS patients were significantly higher in the frontal cortex (P = 0.023), caudate (P = 0.01), thalamus (P = 0.019), amygdala (P = 0.012) and hippocampus (P = 0.002) compared to controls. MD of these structures significantly correlated to a variable degree with neurological disability and neuropsychological dysfunctions. The increased MD values in several cortical and subcortical gray structures and their correlations with neuropsychological variables substantiate a multisystemic degeneration in ALS and suggest that dysfunctions of frontal-subcortical circuits could play a pivotal role in frontal impairment and behavioral symptoms in ALS patients.

MR imaging and cognitive correlates of relapsing-remitting multiple sclerosis patients with cerebellar symptoms.

Multiple sclerosis (MS) is a demyelinating disease affecting the central nervous system, frequently associated with cognitive impairments. Damages of the cerebellum are very common features of patients with MS, although the impact of this clinical factor is generally neglected. Recent evidence from our group demonstrated that MS patients with cerebellar damages are characterized by selective cognitive dysfunctions related to attention and language abilities. Here, we aimed at investigating the presence of neuroanatomical abnormalities in relapsing-remitting MS patients with (RR-MSc) and without (RR-MSnc) cerebellar signs. Twelve RR-MSc patients, 14 demographically, clinically, and radiologically, matched RR-MSnc patients and 20 controls were investigated. All patients underwent neuropsychological assessment. After refilling of FLAIR lesions on the 3D T1-weighted images, VBM was performed using SPM8 and DARTEL. A correlation analysis was performed between VBM results and neuropsychological variables characterizing RR-MSc patients. Despite a similar clinical status, RR-MSc patients were characterized by more severe cognitive damages in attention and language domains with respect to RR-MSnc and controls. With respect to controls, RR-MSnc patients were characterized by a specific atrophy of the bilateral thalami that became more widespread (including motor cortex) in the RR-MSc group (FWE < 0.05). However, consistent with their well-defined neuropsychological deficits, RR-MSc group showed atrophies in the prefrontal and temporal cortical areas when directly compared with RR-MSnc group. Our results demonstrated that RR-MS patients having cerebellar signs were characterized by a distinct neuroanatomical profile, mainly involving cortical regions underpinning executive functions and verbal fluency.