Multishell diffusion MRI reveals whole-brain white matter changes in HIV.

Diffusion tensor imaging (DTI) and diffusion kurtosis imaging (DKI) have been previously used to explore white matter related to human immunodeficiency virus (HIV) infection. While DTI and DKI suffer from low specificity, the Combined Hindered and Restricted Model of Diffusion (CHARMED) provides additional microstructural specificity. We used these three models to evaluate microstructural differences between 35 HIV-positive patients without neurological impairment and 20 healthy controls who underwent diffusion-weighted imaging using three b-values. While significant group effects were found in all diffusion metrics, CHARMED and DKI analyses uncovered wider involvement (80% vs. 20%) of all white matter tracts in HIV infection compared with DTI. In restricted fraction (FR) analysis, we found significant differences in the left corticospinal tract, middle cerebellar peduncle, right inferior cerebellar peduncle, right corticospinal tract, splenium of the corpus callosum, left superior cerebellar peduncle, left superior cerebellar peduncle, pontine crossing tract, left posterior limb of the internal capsule, and left/right medial lemniscus. These are involved in language, motor, equilibrium, behavior, and proprioception, supporting the functional integration that is frequently impaired in HIV-positivity. Additionally, we employed a machine learning algorithm (XGBoost) to discriminate HIV-positive patients from healthy controls using DTI and CHARMED metrics on an ROIwise basis, and unique contributions to this discrimination were examined using Shapley Explanation values. The CHARMED and DKI estimates produced the best performance. Our results suggest that biophysical multishell imaging, combining additional sensitivity and built-in specificity, provides further information about the brain microstructural changes in multimodal areas involved in attentive, emotional and memory networks often impaired in HIV patients.

Clinical implementation of a log file-based machine and patient QA system for IMRT and VMAT treatment plans.

PURPOSE: To determine the error detection sensitivity of a commercial log file-based system (LINACWatch®, LW) for integration into clinical routine and to compare it with a measurement device (OCTAVIUS 4D, Oct4D) for IMRT and VMAT delivery QA. MATERIALS AND METHODS: 76 VMAT/IMRT plans (H&N, prostate, rectum and breast) preliminarily classified according to their Modulation Complexity Score (MCS) calculated by LW, were considered. Receiver Operating Characteristic (ROC) Curves were used to establish gamma criteria for LW. 12 plans (3 for each site) were intentionally modified in order to introduce delivery errors regarding MLC, jaws, collimator, gantry and MU (for a total set of 168 incorrect plans) and irradiated on Oct4D; the corresponding log files were analysed by LW. Each incorrect plan was compared to the error-free plan using γ-index analysis for MLC, jaws and MU errors investigation and Root-Mean-Square (RMS) values for gantry and collimator errors investigation. RESULTS: MCS ranges values were: 0.10-0.20 for H&N, 0.21-0.40 for prostate and rectum, 0.41-1.00 for breast. From ROC curves, the Gamma Passing Rate (GPR) thresholds were: 87%, 92%, 99% for H&N, prostate and rectum, and breast, respectively. The 1.5%/1.5 mm/local criteria were adopted for the γ-analysis. LW sensitivity in detecting the introduced errors was higher when compared to Oct4D: 48.5% vs 30.4% respectively. CONCLUSIONS: LW can be considered useful complement to phantom-based delivery QA of IMRT/VMAT plans. The MCS tool is effective in detecting over or under modulated plans prior to pre-treatment QA. However, rigorous and routinely machine QCs are recommended.

Classification of real-world pathological phonocardiograms through multi-instance learning.

Heart auscultation is an inexpensive and fundamental technique to effectively to diagnose cardiovascular disease. However, due to relatively high human error rates even when auscultation is performed by an experienced physician, and due to the not universal availability of qualified personnel e.g. in developing countries, a large body of research is attempting to develop automated, computational tools for detecting abnormalities in heart sounds. The large heterogeneity of achievable data quality and devices, the variety o possible heart pathologies, and a generally poor signal-to-noise ratio make this problem extremely challenging. We present an accurate classification strategy for diagnosing heart sounds based on 1) automatic heart phase segmentation, 2) state-of-the art filters drawn from the filed of speech synthesis (mel-frequency cepstral representation), and 3) an ad-hoc multi-branch, multi-instance artificial neural network based on convolutional layers and fully connected neuronal ensembles which separately learns from each heart phase, hence leveraging their different physiological significance. We demonstrate that it is possible to train our architecture to reach very high performances, e.g. an AUC of 0.87 or a sensitivity of 0.97. Our machine-learning-based tool could be employed for heart sound classification, especially as a screening tool in a variety of situations including telemedicine applications.

Compartmental models for diffusion weighted MRI reveal widespread brain changes in HIV-infected patients.

Diffusion tensor imaging (DTI) has been used to explore changes in the brain of subjects with human immunodeficiency virus (HIV) infection. However, DTI notoriously suffers from low specificity. Neurite orientation dispersion and density imaging (NODDI) is a compartmental model able to provide specific microstructural information with additional sensitivity/specificity. In this study we use both the NODDI and the DTI models to evaluate microstructural differences between 35 HIV-positive patients and 20 healthy controls. Diffusion-weighted imaging was acquired using three b-values (0, 1000 and 2500 s/mm2). Both DTI and NODDI models were fitted to the data, obtaining estimates for fractional anisotropy (FA), mean diffusivity (MD), radial diffusivity (RD), axial diffusivity (AD), neurite density index (NDI) and orientation dispersion index (ODI), after which we performed group comparisons using Tract-based spatial statistics (TBSS). While significant group effects were found in in FA, MD, RD, AD and NDI, NDI analysis uncovered a much wider involvement of brain tissue in HIV infection as compared to DTI. In region-of interest (ROI)-based analysis, NDI estimates from the right corticospinal tract produced excellent performance in discriminating the two groups (AUC = 0.974, sensitivity = 90%; specificity =97%).

Comparison of rigid and deformable coregistration between mpMRI and CT images in radiotherapy of prostate bed cancer recurrence.

PURPOSE: To evaluate the accuracy of rigid coregistration between multiparametric magnetic resonance (mpMR) and computed tomography (CT) images for radiotherapy of prostate bed cancer recurrence. MATERIALS AND METHOD: Fifty-three patients (59 nodules) accrued in a prospective study on salvage radiotherapy for prostatic bed recurrence were suitable for the analysis. Patients underwent a pre radiotherapy mpMR exam and a planning CT in the same treatment position and with control of organ filling. The site of recurrence was delineated on mpMR images and contours transferred on planning CT images using both rigid and deformable registrations. Coregistrations were evaluated by mathematical operators that quantify deformation (Jacobian determinant and vector curl) and similarity indices (Dice and Jaccard coefficients). Dose coverage was evaluated. RESULTS: Deformable registration did not change volumes, (p = 0.92 MW test). The Jacobian coefficient and the vector curl revealed no important image deformations. Dice and Jaccard coefficients indicated dislocation of the nodule volumes. Dislocation magnitude was d = (5.6 ± 3.1) mm. Organ filling was not correlated with deformation or dislocation. Volumes were covered by the 95% isodose in 96% of cases when rigid registration was performed versus 75% of cases when deformed. CONCLUSIONS: Rigid image coregistration is sufficiently accurate in this setting. The results indicate that the deformable registration tends to shrink the voxels and to dislocate the ROI, the adopted expansion for the recurrence volume adequately accounts for the observed deformation and dislocation, provided that organ filling is controlled.

Echo state network models for nonlinear Granger causality.

While Granger causality (GC) has been often employed in network neuroscience, most GC applications are based on linear multivariate autoregressive (MVAR) models. However, real-life systems like biological networks exhibit notable nonlinear behaviour, hence undermining the validity of MVAR-based GC (MVAR-GC). Most nonlinear GC estimators only cater for additive nonlinearities or, alternatively, are based on recurrent neural networks or long short-term memory networks, which present considerable training difficulties and tailoring needs. We reformulate the GC framework in terms of echo-state networks-based models for arbitrarily complex networks, and characterize its ability to capture nonlinear causal relations in a network of noisy Duffing oscillators, showing a net advantage of echo state GC (ES-GC) in detecting nonlinear, causal links. We then explore the structure of ES-GC networks in the human brain employing functional MRI data from 1003 healthy subjects drawn from the human connectome project, demonstrating the existence of previously unknown directed within-brain interactions. In addition, we examine joint brain-heart signals in 15 subjects where we explore directed interaction between brain networks and central vagal cardiac control in order to investigate the so-called central autonomic network in a causal manner. This article is part of the theme issue 'Advanced computation in cardiovascular physiology: new challenges and opportunities'.

A novel multi-branch architecture for state of the art robust detection of pathological phonocardiograms.

Heart auscultation is an inexpensive and fundamental technique to effectively diagnose cardiovascular disease. However, due to relatively high human error rates even when auscultation is performed by an experienced physician, and due to the not universal availability of qualified personnel, e.g. in developing countries, many efforts are made worldwide to propose computational tools for detecting abnormalities in heart sounds. The large heterogeneity of achievable data quality and devices, the variety of possible heart pathologies, and a generally poor signal-to-noise ratio make this problem very challenging. We present an accurate classification strategy for diagnosing heart sounds based on (1) automatic heart phase segmentation, (2) state-of-the art filters drawn from the field of speech synthesis (mel-frequency cepstral representation) and (3) an ad hoc multi-branch, multi-instance artificial neural network based on convolutional layers and fully connected neuronal ensembles which separately learns from each heart phase hence implicitly leveraging their different physiological significance. We demonstrate that it is possible to train our architecture to reach very high performances, e.g. an area under the curve of 0.87 or a sensitivity of 0.97. Our machine-learning-based tool could be employed for heartsound classification, especially as a screening tool in a variety of situations including telemedicine applications. This article is part of the theme issue 'Advanced computation in cardiovascular physiology: new challenges and opportunities'.

Global and local brain connectivity changes associated with sudden unilateral sensorineural hearing loss.

Recent studies suggest that even moderate sudden sensorineural hearing loss (SSNHL) causes reduction of gray matter volume in the primary auditory cortex, diminishing its ability to react to sound stimulation, as well as reorganization of functional brain networks. We employed resting-state functional MRI (rs-fMRI), in conjunction with graph-theoretical analysis and a newly developed functional "disruption index," to study whole-brain as well as local functional changes in patients with unilateral SSNHL. We also assessed the potential of graph-theoretical measures as biomarkers of disease, in terms of their relationship to clinically relevant audiological parameters. Eight patients with moderate or severe unilateral SSNHL and 15 healthy controls were included in this prospective pilot study. All patients underwent rs-fMRI to study potential changes in brain connectivity. From rs-fMRI data, global and local graph-theoretical measures, disruption index, and audiological examinations were estimated. Mann-Whitney U tests were used to study the differences between SSNHL patients and healthy controls. Associations between brain metrics and clinical variables were studied using multiple linear regressions, and the presence or absence of brain network hubs was assessed using Fisher's exact test. No statistically significant differences between SSNHL patients and healthy controls were found in global or local network measures. However, when analyzing brain networks through the disruption index, we found a brain-wide functional network reorganization (p < 0.001 as compared with controls), whose extent was associated with clinical impairment (p < 0.05). We also observed several functional hubs in SSNHL patients that were not present in healthy controls and vice versa. Our results demonstrate a brain involvement in SSNHL patients, not detectable using conventional graph-theoretical analysis, which may yield subtle disease clues and possibly aid in monitoring disease progression in clinical trials.

Functional brain network reorganization in HIV infection.

BACKGROUND AND PURPOSE: To investigate the reorganization of the central nervous system provided by resting state-functional MRI (rs-fMRI), graph-theoretical analysis, and a newly developed functional brain network disruption index in patients with human immunodeficiency virus (HIV) infection. METHODS: Forty HIV-positive patients without neurological impairment and 20 age- and sex-matched healthy controls underwent rs-fMRI at 3T; blood sampling was obtained the same day to evaluate biochemical variables (absolute, relative, and nadir CD4 T-lymphocytes value and plasmatic HIV-RNA). From fMRI data, disruption indices, as well as global and local graph theoretical measures, were estimated and examined for group differences (HIV vs. controls) as well as for associations with biochemical variables (HIV only). Finally, all data (global and local graph-theoretical measures, disruption indices, and biochemical variables) were tested for putative differences across three patient groups based on the duration of combined antiretroviral therapy (cART). RESULTS: Brain function of HIV patients appeared to be deeply reorganized as compared to normal controls. The disruption index showed significant negative association with relative CD4 values, and a positive significant association between plasmatic HIV-RNA and local graph-theoretical metrics in the left lingual gyrus and the right lobule IV and V of right cerebellar hemisphere was also observed. Finally, a differential distribution of HIV clinical biomarkers and several brain metrics was observed across cART duration groups. CONCLUSION: Our study demonstrates that rs-fMRI combined with advanced graph theoretical analysis and disruption indices is able to detect early and subtle functional changes of brain networks in HIV patients.

Deep computational pathology in breast cancer.

Deep Learning (DL) algorithms are a set of techniques that exploit large and/or complex real-world datasets for cross-domain and cross-discipline prediction and classification tasks. DL architectures excel in computer vision tasks, and in particular image processing and interpretation. This has prompted a wave of disruptingly innovative applications in medical imaging, where DL strategies have the potential to vastly outperform human experts. This is particularly relevant in the context of histopathology, where whole slide imaging (WSI) of stained tissue in conjuction with DL algorithms for their interpretation, selection and cancer staging are beginning to play an ever increasing role in supporting human operators in visual assessments. This has the potential to reduce everyday workload as well as to increase precision and reproducibility across observers, centers, staining techniques and even pathologies. In this paper we introduce the most common DL architectures used in image analysis, with a focus on histopathological image analysis in general and in breast histology in particular. We briefly review how, state-of-art DL architectures compare to human performance on across a number of critical tasks such as mitotic count, tubules analysis and nuclear pleomorphism analysis. Also, the development of DL algorithms specialized to pathology images have been enormously fueled by a number of world-wide challenges based on large, multicentric image databases which are now publicly available. In turn, this has allowed most recent efforts to shift more and more towards semi-supervised learning methods, which provide greater flexibility and applicability. We also review all major repositories of manually labelled pathology images in breast cancer and provide an in-depth discussion of the challenges specific to training DL architectures to interpret WSI data, as well as a review of the state-of-the-art methods for interpretation of images generated from immunohistochemical analysis of breast lesions. We finally discuss the future challenges and opportunities which the adoption of DL paradigms is most likely to pose in the field of pathology for breast cancer detection, diagnosis, staging and prognosis. This review is intended as a comprehensive stepping stone into the field of modern computational pathology for a transdisciplinary readership across technical and medical disciplines.

Radiomics in breast cancer classification and prediction.

Breast Cancer (BC) is the common form of cancer in women. Its diagnosis and screening are usually performed through different imaging modalities such as mammography, magnetic resonance imaging and ultrasound. However, mammography and ultrasound-imaging techniques have limited sensitivity and specificity both in identifying lesions and in differentiating malign from benign lesions, especially in presence of dense breast parenchyma. Due to the higher resolution of magnetic resonance images, MRI represents the method with the higher specificity and sensitivity among all the available tools, in both lesions' identification and diagnosis. However, especially for diagnosis, even MRI has limitations that are only partially solved if combined with mammography. Unfortunately, due to the limits of all these imaging tools, in order to have a certain diagnosis, patients often receive painful and costly bioptics procedures. In this context, several computational approaches have been developed to increase sensitivity, while maintaining the same specificity, in BC diagnosis and screening. Amongst these, radiomics has been increasingly gaining ground in oncology to improve cancer diagnosis, prognosis and treatment. Radiomics derives multiple quantitative features from single or multiple medical imaging modalities, highlighting image traits which are not visible to the naked eye and hence significantly augmenting the discriminatory and predictive potential of medical imaging. This review article aims to summarize the state of the art in radiomics-based BC research. The dominating evidence extracted from the literature points towards a high potential of radiomics in disentangling malignant from benign breast lesions, classifying BC types and grades and also in predicting treatment response and recurrence risk. In the era of personalized medicine, radiomics has the potential to improve diagnosis, prognosis, prediction, monitoring, image-based intervention, and assessment of therapeutic response in BC.

Disruption of brain network organization in patients with human immunodeficiency virus (HIV) infection.

In 2019, approximately 38 million people were living with human immunodeficiency virus (HIV). Combined antiretroviral therapy (cART) has determined a change in the course of HIV infection, transforming it into a chronic condition which results in cumulative exposure to antiretroviral drugs, inflammatory effects and aging. Relatedly, at least one quarter of HIV-infected patients suffer from cognitive, motor and behavioral disorder, globally known as HIV-associated neurocognitive disorders (HAND). In this context, objective, neuroimaging-based biomarkers are therefore highly desirable in order to detect, quantify and monitor HAND in all disease stages. In this study, we employed functional MRI in conjunction with graph-theoretical analysis as well as a newly developed functional brain network disruption index to assess a putative functional reorganization in HIV positive patients. We found that brain function of HIV patients is deeply reorganized as compared to normal controls. Interestingly, the regions in which we found reorganized hubs are integrated into neuronal networks involved in working memory, motor and executive functions often altered in patients with HAND. Overall, our study demonstrates that rs-fMRI combined with advanced graph theoretical analysis and disruption indices is able to detect early, subtle functional changes of brain networks in HIV patients before structural changes become evident.

Global and local reorganization of brain network connectivity in sudden sensorineural hearing loss.

Recent reports suggested that even moderate sudden sensorineural hearing loss (SSNHL) can be partly responsible for a loss of gray matter volume in the primary auditory cortex, hence reducing the capacity of the auditory cortical areas to react to sound stimulation. There is also evidence for a plastic reorganization of brain functional networks visible as enhanced local functional connectivity. The aim of this study was to use rs-fMRI, in conjunction with graph- theoretical analysis and a newly developed functional "disruption index" to study whole-brain as well as local functional changes in patients with acute and unilateral sensorineural hearing loss. No statistically significant differences in global or local network measures we found between SSNHL patients and healthy controls. However, when analyzing local metrics through the disruption index k, we found negative values for k which were statistically different from zero both in single subject analysis. Additionally, we found several associations between graph-theoretical metrics and clinical parameters.

Reorganization of the structural connectome in primary open angle Glaucoma.

Primary open angle Glaucoma (POAG) is one of the most common causes of permanent blindness in the world. Recent studies have suggested the hypothesis that POAG is also a central nervous system disorder which may result in additional (i.e., extra-ocular) involvement. The aim of this study is to assess possible structural, whole-brain connectivity alterations in POAG patients. We evaluated 23 POAG patients and 15 healthy controls by combining multi-shell diffusion weighted imaging, multi-shell, multi-tissue probabilistic tractography, graph theoretical measures and a recently designed 'disruption index', which evaluates the global reorganization of brain networks. We also studied the associations between the whole-brain structural connectivity measures and indices of visual acuity including the field index (VFI) and two Optical Coherence Tomography (OCT) parameters, namely the Macula Ganglion Cell Layer (MaculaGCL) and Retinal Nerve Fiber Layer (RNFL) thicknesses. We found both global and local structural connectivity differences between POAG patients and controls, which extended well beyond the primary visual pathway and were localized in the left calcarine gyrus (clustering coefficient p = 0.036), left lateral occipital cortex (clustering coefficient p = 0.017, local efficiency p = 0.035), right lingual gyrus (clustering coefficient p = 0.009), and right paracentral lobule (clustering coefficient p = 0.009, local efficiency p = 0.018). Group-wise (clustering coefficient, p = 6.59∙10-7 and local efficiency p = 6.23·10-8) and subject-wise disruption indices (clustering coefficient, p = 0.018 and local efficiency, p = 0.01) also differed between POAG patients and controls. In addition, we found negative associations between RNFL thickness and local measures (clustering coefficient, local efficiency and strength) in the right amygdala (local efficiency p = 0.008, local strength p = 0.016), right inferior temporal gyrus (clustering coefficient p = 0.036, local efficiency p = 0.042), and right temporal pole (local strength p = 0.008). Overall, we show, in patients with POAG, a whole-brain structural reorganization that spans across a variety of brain regions involved in visual processing, motor control, and emotional/cognitive functions. We also identified a pattern of brain structural changes in relation to POAG clinical severity. Taken together, our findings support the hypothesis that the reduction in visual acuity from POAG can be driven by a combination of local (i.e., in the eye) and more extended (i.e., brain) effects.

Diffusional Kurtosis Imaging of White Matter Degeneration in Glaucoma.

Glaucoma is an optic neuropathy characterized by death of retinal ganglion cells and loss of their axons, progressively leading to blindness. Recently, glaucoma has been conceptualized as a more diffuse neurodegenerative disorder involving the optic nerve and also the entire brain. Consistently, previous studies have used a variety of magnetic resonance imaging (MRI) techniques and described widespread changes in the grey and white matter of patients. Diffusion kurtosis imaging (DKI) provides additional information as compared with diffusion tensor imaging (DTI), and consistently provides higher sensitivity to early microstructural white matter modification. In this study, we employ DKI to evaluate differences among healthy controls and a mixed population of primary open angle glaucoma patients ranging from stage I to V according to Hodapp-Parrish-Anderson visual field impairment classification. To this end, a cohort of patients affected by primary open angle glaucoma (n = 23) and a group of healthy volunteers (n = 15) were prospectively enrolled and underwent an ophthalmological evaluation followed by magnetic resonance imaging (MRI) using a 3T MR scanner. After estimating both DTI indices, whole-brain, voxel-wise statistical comparisons were performed in white matter using Tract-Based Spatial Statistics (TBSS). We found widespread differences in several white matter tracts in patients with glaucoma relative to controls in several metrics (mean kurtosis, kurtosis anisotropy, radial kurtosis, and fractional anisotropy) which involved localization well beyond the visual pathways, and involved cognitive, motor, face recognition, and orientation functions amongst others. Our findings lend further support to a causal brain involvement in glaucoma and offer alternative explanations for a number of multidomain impairments often observed in glaucoma patients.

Heart Rate Variability Parameters During Psychogenic Non-epileptic Seizures: Comparison Between Patients With Pure PNES and Comorbid Epilepsy.

Introduction: Psychogenic non-epileptic seizures (PNES) may resemble epileptic seizures. There are few data about ictal ANS activity alterations induced by PNES in patients with pure PNES (pPNES) compared to PNES with comorbid epilepsy (PNES/ES). We aimed to compare heart rate variability (HRV) parameters and hence autonomic regulation in PNES in epileptic and non-epileptic patients. Methods: We obtained HRV data from video-electroencephalography recordings in 22 patients presenting PNES (11 pPNES and 11 PNES/ES) in awake, and supine states. We calculated HRV parameters in both time and frequency domains including low frequency (LF) power, high frequency power (HF), LF/HF ratio, square root of the mean of the sum of the squares of differences between adjacent R wave intervals (RMSSD) and the standard deviation of all consecutive R wave intervals (SDNN). We also evaluated approximate entropy (ApEn), cardiosympathetic index (CSI), and cardiovagal index (CVI). Four conditions were considered: basal condition (BAS), before PNES (PRE), during PNES (ICT) and after PNES (POST). Results: HRV analysis showed significantly higher ICT LF and LF/HF ratio vs. each condition. We also found higher POST HF vs. PRE and BAS, lower RRI in ICT vs. each condition and PRE vs. BAS. POST RMSSD was significantly higher compared to all other states. ICT CSI was significantly higher compared to all other states, whereas CSI was significantly lower in POST vs. PRE and PRE CVI lower than ICT and higher in POST vs. BAS and PRE. Also, ICT ApEn was lower than in all other states. Higher LF in pPNES vs. PNES/ES was also evident when compared across groups. Significance: A few studies examined HRV alterations in PNES, reporting high sympathetic tone (although less evident than in epileptic seizures). Our data suggest a sympathetic overdrive before and during PNES followed by a post-PNES increase in vagal tone. A sympathovagal imbalance was more evident in pPNES as compared to PNES/ES.

Primary Open Angle Glaucoma Is Associated With Functional Brain Network Reorganization.

Background: Resting-state functional magnetic resonance imaging (rs-fMRI) is commonly employed to study changes in functional brain connectivity. The recent hypothesis of a brain involvement in primary open angle Glaucoma has sprung interest for neuroimaging studies in this classically ophthalmological pathology. Object: We explored a putative reorganization of functional brain networks in Glaucomatous patients, and evaluated the potential of functional network disruption indices as biomarkers of disease severity in terms of their relationship to clinical variables as well as select retinal layer thicknesses. Methods: Nineteen Glaucoma patients and 16 healthy control subjects (age: 50-76, mean 61.0 ± 8.2 years) underwent rs-fMRI examination at 3T. After preprocessing, rs-fMRI time series were parcellated into 116 regions using the Automated Anatomical Labeling atlas and adjacency matrices were computed based on partial correlations. Graph-theoretical measures of integration, segregation and centrality as well as group-wise and subject-wise disruption index estimates (which use regression of graph-theoretical metrics across subjects to quantify overall network changes) were then generated for all subjects. All subjects also underwent Optical Coherence Tomography (OCT) and visual field index (VFI) quantification. We then examined associations between brain network measures and VFI, as well as thickness of retinal nerve fiber layer (RNFL) and macular ganglion cell layer (MaculaGCL). Results: In Glaucoma, group-wise disruption indices were negative for all graph theoretical metrics. Also, we found statistically significant group-wise differences in subject-wise disruption indexes in all local metrics. Two brain regions serving as hubs in healthy controls were not present in the Glaucoma group. Instead, three hub regions were present in Glaucoma patients but not in controls. We found significant associations between all disruption indices and VFI, RNFL as well as MaculaGCL. The disruption index based on the clustering coefficient yielded the best discriminative power for differentiating Glaucoma patients from healthy controls [Area Under the ROC curve (AUC) 0.91, sensitivity, 100%; specificity, 78.95%]. Conclusions: Our findings support a possible relationship between functional brain changes and disease severity in Glaucoma, as well as alternative explanations for motor and cognitive symptoms in Glaucoma, possibly pointing toward an inclusion of this pathology in the heterogeneous group of disconnection syndromes.

An Ad Hoc Random Initialization Deep Neural Network Architecture for Discriminating Malignant Breast Cancer Lesions in Mammographic Images.

Breast cancer is one of the most common cancers in women, with more than 1,300,000 cases and 450,000 deaths each year worldwide. In this context, recent studies showed that early breast cancer detection, along with suitable treatment, could significantly reduce breast cancer death rates in the long term. X-ray mammography is still the instrument of choice in breast cancer screening. In this context, the false-positive and false-negative rates commonly achieved by radiologists are extremely arduous to estimate and control although some authors have estimated figures of up to 20% of total diagnoses or more. The introduction of novel artificial intelligence (AI) technologies applied to the diagnosis and, possibly, prognosis of breast cancer could revolutionize the current status of the management of the breast cancer patient by assisting the radiologist in clinical image interpretation. Lately, a breakthrough in the AI field has been brought about by the introduction of deep learning techniques in general and of convolutional neural networks in particular. Such techniques require no a priori feature space definition from the operator and are able to achieve classification performances which can even surpass human experts. In this paper, we design and validate an ad hoc CNN architecture specialized in breast lesion classification from imaging data only. We explore a total of 260 model architectures in a train-validation-test split in order to propose a model selection criterion which can pose the emphasis on reducing false negatives while still retaining acceptable accuracy. We achieve an area under the receiver operatic characteristics curve of 0.785 (accuracy 71.19%) on the test set, demonstrating how an ad hoc random initialization architecture can and should be fine tuned to a specific problem, especially in biomedical applications.

A Parsimonious Granger Causality Formulation for Capturing Arbitrarily Long Multivariate Associations.

High-frequency neuroelectric signals like electroencephalography (EEG) or magnetoencephalography (MEG) provide a unique opportunity to infer causal relationships between local activity of brain areas. While causal inference is commonly performed through classical Granger causality (GC) based on multivariate autoregressive models, this method may encounter important limitations (e.g., data paucity) in the case of high dimensional data from densely connected systems like the brain. Additionally, physiological signals often present long-range dependencies which commonly require high autoregressive model orders/number of parameters. We present a generalization of autoregressive models for GC estimation based on Wiener-Volterra decompositions with Laguerre polynomials as basis functions. In this basis, the introduction of only one additional global parameter allows to capture arbitrary long dependencies without increasing model order, hence retaining model simplicity, linearity and ease of parameters estimation. We validate our method in synthetic data generated from families of complex, densely connected networks and demonstrate superior performance as compared to classical GC. Additionally, we apply our framework to studying the directed human brain connectome through MEG data from 89 subjects drawn from the Human Connectome Project (HCP) database, showing that it is able to reproduce current knowledge as well as to uncover previously unknown directed influences between cortical and limbic brain regions.

Variability and Reproducibility of Directed and Undirected Functional MRI Connectomes in the Human Brain.

A growing number of studies are focusing on methods to estimate and analyze the functional connectome of the human brain. Graph theoretical measures are commonly employed to interpret and synthesize complex network-related information. While resting state functional MRI (rsfMRI) is often employed in this context, it is known to exhibit poor reproducibility, a key factor which is commonly neglected in typical cohort studies using connectomics-related measures as biomarkers. We aimed to fill this gap by analyzing and comparing the inter- and intra-subject variability of connectivity matrices, as well as graph-theoretical measures, in a large (n = 1003) database of young healthy subjects which underwent four consecutive rsfMRI sessions. We analyzed both directed (Granger Causality and Transfer Entropy) and undirected (Pearson Correlation and Partial Correlation) time-series association measures and related global and local graph-theoretical measures. While matrix weights exhibit a higher reproducibility in undirected, as opposed to directed, methods, this difference disappears when looking at global graph metrics and, in turn, exhibits strong regional dependence in local graphs metrics. Our results warrant caution in the interpretation of connectivity studies, and serve as a benchmark for future investigations by providing quantitative estimates for the inter- and intra-subject variabilities in both directed and undirected connectomic measures.