Trustworthy clinical AI solutions: A unified review of uncertainty quantification in Deep Learning models for medical image analysis

The full acceptance of Deep Learning (DL) models in the clinical field is rather low with respect to the quantity of high-performing solutions reported in the literature. End users are particularly reluctant to rely on the opaque predictions of DL models. Uncertainty quantification methods have been proposed in the literature as a potential solution, to reduce the black-box effect of DL models and increase the interpretability and the acceptability of the result by the final user. In this review, we propose an overview of the existing methods to quantify uncertainty associated with DL predictions. We focus on applications to medical image analysis, which present specific challenges due to the high dimensionality of images and their variable quality, as well as constraints associated with real-world clinical routine. Moreover, we discuss the concept of structural uncertainty, a corpus of methods to facilitate the alignment of segmentation uncertainty estimates with clinical attention. We then discuss the evaluation protocols to validate the relevance of uncertainty estimates. Finally, we highlight the open challenges for uncertainty quantification in the medical field.

Graph-based methods coupled with specific distributional distances for adversarial attack detection

Artificial neural networks are prone to being fooled by carefully perturbed inputs which cause an egregious misclassification. These adversarial attacks have been the focus of extensive research. Likewise, there has been an abundance of research in ways to detect and defend against them. We introduce a novel approach of detection and interpretation of adversarial attacks from a graph perspective. For an input image, we compute an associated sparse graph using the layer-wise relevance propagation algorithm (Bach et al., 2015). Specifically, we only keep edges of the neural network with the highest relevance values. Three quantities are then computed from the graph which are then compared against those computed from the training set. The result of the comparison is a classification of the image as benign or adversarial. To make the comparison, two classification methods are introduced: (1) an explicit formula based on Wasserstein distance applied to the degree of node and (2) a logistic regression. Both classification methods produce strong results which lead us to believe that a graph-based interpretation of adversarial attacks is valuable.

ImUnity: A generalizable VAE-GAN solution for multicenter MR image harmonization

ImUnity is an original 2.5D deep-learning model designed for efficient and flexible MR image harmonization. A VAE-GAN network, coupled with a confusion module and an optional biological preservation module, uses multiple 2D slices taken from different anatomical locations in each subject of the training database, as well as image contrast transformations for its training. It eventually generates 'corrected' MR images that can be used for various multi-center population studies. Using 3 open source databases (ABIDE, OASIS and SRPBS), which contain MR images from multiple acquisition scanner types or vendors and a large range of subjects ages, we show that ImUnity: (1) outperforms state-of-the-art methods in terms of quality of images generated using traveling subjects; (2) removes sites or scanner biases while improving patients classification; (3) harmonizes data coming from new sites or scanners without the need for an additional fine-tuning and (4) allows the selection of multiple MR reconstructed images according to the desired applications. Tested here on T1-weighted images, ImUnity could be used to harmonize other types of medical images.

Nodal-statistics-based equivalence relation for graph collections

Node role explainability in complex networks is very difficult yet is crucial in different application domains such as social science, neurosciences, or computer science. Many efforts have been made on the quantification of hubs revealing particular nodes in a network using a given structural property. Yet, in several applications, when multiple instances of networks are available and several structural properties appear to be relevant, the identification of node roles remains largely unexplored. Inspired by the node automorphically equivalence relation, we define an equivalence relation on graph nodes associated with any collection of nodal statistics (i.e., any functions on the node set). This allows us to define new graph global measures: the power coefficient and the orthogonality score to evaluate the parsimony and heterogeneity of a given nodal statistics collection. In addition, we introduce a new method based on structural patterns to compare graphs that have the same vertices set. This method assigns a value to a node to determine its role distinctiveness in a graph family. Extensive numerical results of our method are conducted on both generative graph models and real data concerning human brain functional connectivity. The differences in nodal statistics are shown to be dependent on the underlying graph structure. Comparisons between generative models and real networks combining two different nodal statistics reveal the complexity of human brain functional connectivity with differences at both global and nodal levels. Using a group of 200 healthy controls connectivity networks, our method computes high correspondence scores among the whole population to detect homotopy and finally quantify differences between comatose patients and healthy controls.

Automated Quantification of Brain Lesion Volume From Post-trauma MR Diffusion-Weighted Images

Objectives

Determining the volume of brain lesions after trauma is challenging. Manual delineation is observer-dependent and time-consuming and cannot therefore be used in routine practice. The study aimed to evaluate the feasibility of an automated atlas-based quantification procedure (AQP) based on the detection of abnormal mean diffusivity (MD) values computed from diffusion-weighted MR images.

Methods

The performance of AQP was measured against manual delineation consensus by independent raters in two series of experiments based on: (i) realistic trauma phantoms (n = 5) where low and high MD values were assigned to healthy brain images according to the intensity, form and location of lesion observed in real TBI cases; (ii) severe TBI patients (n = 12 patients) who underwent MR imaging within 10 days after injury.

Results

In realistic TBI phantoms, no statistical differences in Dice similarity coefficient, precision and brain lesion volumes were found between AQP, the rater consensus and the ground truth lesion delineations. Similar findings were obtained when comparing AQP and manual annotations for TBI patients. The intra-class correlation coefficient between AQP and manual delineation was 0.70 in realistic phantoms and 0.92 in TBI patients. The volume of brain lesions detected in TBI patients was 59 ml (19–84 ml) (median; 25–75th centiles).

Conclusions

Our results support the feasibility of using an automated quantification procedure to determine, with similar accuracy to manual delineation, the volume of low and high MD brain lesions after trauma, and thus allow the determination of the type and volume of edematous brain lesions. This approach had comparable performance with manual delineation by a panel of experts. It will be tested in a large cohort of patients enrolled in the multicenter OxyTC trial (NCT02754063).

Effective connectivity in subcortical visual structures in de novo Patients with Parkinson's Disease

BACKGROUND

Parkinson's disease (PD) manifests with the appearance of non-motor symptoms before motor symptoms onset. Among these, dysfunctioning visual structures have recently been reported to occur at early disease stages.

OBJECTIVE

This study addresses effective connectivity in the visual network of PD patients.

METHODS

Using functional MRI and dynamic causal modeling analysis, we evaluated the connectivity between the superior colliculus, the lateral geniculate nucleus and the primary visual area V1 in de novo untreated PD patients (n = 22). A subset of the PD patients (n = 8) was longitudinally assessed two times at two months and at six months after starting dopaminergic treatment. Results were compared to those of age-matched healthy controls (n = 22).

RESULTS

Our results indicate that the superior colliculus drives cerebral activity for luminance contrast processing both in healthy controls and untreated PD patients. The same effective connectivity was observed with neuromodulatory differences in terms of neuronal dynamic interactions. Our main findings were that the modulation induced by luminance contrast changes of the superior colliculus connectivity (self-connectivity and connectivity to the lateral geniculate nucleus) was inhibited in PD patients (effect of contrast: p = 0.79 and p = 0.77 respectively). The introduction of dopaminergic medication in a subset (n = 8) of the PD patients failed to restore the effective connectivity modulation observed in the healthy controls.

INTERPRETATION

The deficits in luminance contrast processing in PD was associated with a deficiency in connectivity adjustment from the superior colliculus to the lateral geniculate nucleus and to V1. No differences in cerebral blood flow were observed between controls and PD patients suggesting that the deficiency was at the neuronal level. Administration of a dopaminergic treatment over six months was not able to normalize the observed alterations in inter-regional coupling. These findings highlight the presence of early dysfunctions in primary visual areas, which might be used as early markers of the disease.

Neural correlates of intra-saccadic motion perception

Retinal motion of the visual scene is not consciously perceived during ocular saccades in normal everyday conditions. It has been suggested that extra-retinal signals actively suppress intra-saccadic motion perception to preserve stable perception of the visual world. However, using stimuli optimized to preferentially activate the M-pathway, Castet and Masson (2000) demonstrated that motion can be perceived during a saccade. Based on this psychophysical paradigm, we used electroencephalography and eye-tracking recordings to investigate the neural correlates related to the conscious perception of intra-saccadic motion. We demonstrated the effective involvement during saccades of the cortical areas V1-V2 and MT-V5, which convey motion information along the M-pathway. We also showed that individual motion perception was related to retinal temporal frequency.

A Multicenter Preclinical MRI Study: Definition of Rat Brain Relaxometry Reference Maps

Similarly to human population imaging, there are several well-founded motivations for animal population imaging, the most notable being the improvement of the validity of statistical results by pooling a sufficient number of animal data provided by different imaging centers. In this paper, we demonstrate the feasibility of such a multicenter animal study, sharing raw data from forty rats and processing pipelines between four imaging centers. As specific use case, we focused on T1 and T2 mapping of the healthy rat brain at 7T. We quantitatively report about the variability observed across two MR data providers and evaluate the influence of image processing steps on the final maps, using three fitting algorithms from three centers. Finally, to derive relaxation times from different brain areas, two multi-atlas segmentation pipelines from different centers were performed on two different platforms. Differences between the two data providers were 2.21% for T1 and 9.52% for T2. Differences between processing pipelines were 1.04% for T1 and 3.33% for T2. These maps, obtained in healthy conditions, may be used in the future as reference when exploring alterations in animal models of pathology.

Small Animal Shanoir (SAS) A Cloud-Based Solution for Managing Preclinical MR Brain Imaging Studies

Clinical multicenter imaging studies are frequent and rely on a wide range of existing tools for sharing data and processing pipelines. This is not the case for preclinical (small animal) studies. Animal population imaging is still in infancy, especially because a complete standardization and control of initial conditions in animal models across labs is still difficult and few studies aim at standardization of acquisition and post-processing techniques. Clearly, there is a need of appropriate tools for the management and sharing of data, post-processing and analysis methods dedicated to small animal imaging. Solutions developed for Human imaging studies cannot be directly applied to this specific domain. In this paper, we present the Small Animal Shanoir (SAS) solution for supporting animal population imaging using tools compatible with open data. The integration of automated workflow tools ensures accessibility and reproducibility of research outputs. By sharing data and imaging processing tools, hosted by SAS, we promote data preparation and tools for reproducibility and reuse, and participation in multicenter or replication "open science" studies contributing to the improvement of quality science in preclinical domain. SAS is a first step for promoting open science for small animal imaging and a contribution to the valorization of data and pipelines of reference.

Multivariate pattern analysis of fMRI data for imaginary and real colours in grapheme-colour synaesthesia

Grapheme-colour synaesthesia is a subjective phenomenon related to perception and imagination, in which some people involuntarily but systematically associate specific, idiosyncratic colours to achromatic letters or digits. Its investigation is relevant to unravel the neural correlates of colour perception in isolation from low-level neural processing of spectral components, as well as the neural correlates of imagination by being able to reliably trigger imaginary colour experiences. However, functional MRI studies using univariate analyses failed to provide univocal evidence of the activation of the "colour network" by synaesthesia. Applying multivariate (multivoxel) pattern analysis (MVPA) on 20 synaesthetes and 20 control participants, we tested whether the neural processing of real colours (concentric rings) and synaesthetic colours (black graphemes) shared patterns of activations. Region of interest analyses in retinotopically and anatomically defined visual areas revealed neither evidence of shared circuits for real and synaesthetic colour processing, nor processing difference between synaesthetes and controls. We also found no correlation with individual experiences, characterised by measuring the strength of synaesthetic associations. The whole brain searchlight analysis led to similar results. We conclude that revealing the neural coding of the synaesthetic experience of colours is a hard task which requires the improvement of our current methodology: for example involving more individuals and achieving higher MR signal to noise ratio and spatial resolution. So far, we have not found any evidence of the involvement of the cortical colour network in the subjective experience of synaesthetic colours.

No Structural Differences Are Revealed by VBM in 'De Novo' Parkinsonian Patients

The identification of brain morphological alterations in newly diagnosed PD patients (i.e. 'de novo') could potentially serve as a biomarker and accelerate diagnosis. However, presently no consensus exists in the literature possibly due to several factors: small size cohorts, differences in segmentation techniques or bad control of false positive rates. In this study, we use the CAT12 pipeline, to seek for morphological brain differences in gray and white matter of 66 controls and 144 de novo PD patients from the PPMI database. Moreover, we search for subcortical structure differences using the VolBrain pipeline. We found no structural brain differences in this de novo Parkinsonian population, neither in tissues using a whole brain analysis nor in any of nine subcortical structures analyzed separately. We conclude that some results published in the literature may appear as false positives and we contest their reproductibility.

Objective Evaluation of Multiple Sclerosis Lesion Segmentation using a Data Management and Processing Infrastructure

We present a study of multiple sclerosis segmentation algorithms conducted at the international MICCAI 2016 challenge. This challenge was operated using a new open-science computing infrastructure. This allowed for the automatic and independent evaluation of a large range of algorithms in a fair and completely automatic manner. This computing infrastructure was used to evaluate thirteen methods of MS lesions segmentation, exploring a broad range of state-of-theart algorithms, against a high-quality database of 53 MS cases coming from four centers following a common definition of the acquisition protocol. Each case was annotated manually by an unprecedented number of seven different experts. Results of the challenge highlighted that automatic algorithms, including the recent machine learning methods (random forests, deep learning, …), are still trailing human expertise on both detection and delineation criteria. In addition, we demonstrate that computing a statistically robust consensus of the algorithms performs closer to human expertise on one score (segmentation) although still trailing on detection scores.

Neural circuits for long-range color filling-in

Surface color appearance depends on both local surface chromaticity and global context. How are these inter-dependencies supported by cortical networks? Combining functional imaging and psychophysics, we examined if color from long-range filling-in engages distinct pathways from responses caused by a field of uniform chromaticity. We find that color from filling-in is best classified and best correlated with appearance by two dorsal areas, V3A and V3B/KO. In contrast, a field of uniform chromaticity is best classified by ventral areas hV4 and LO. Dynamic causal modeling revealed feedback modulation from area V3A to areas V1 and LO for filling-in, contrasting with feedback from LO modulating areas V1 and V3A for a matched uniform chromaticity. These results indicate a dorsal stream role in color filling-in via feedback modulation of area V1 coupled with a cross-stream modulation of ventral areas suggesting that local and contextual influences on color appearance engage distinct neural networks.

Magnetic resonance imaging does not reveal structural alterations in the brain of grapheme-color synesthetes

Several publications have reported structural changes in the brain of synesthetes compared to controls, either local differences or differences in connectivity. In the present study, we pursued this quest for structural brain differences that might support the subjective experience of synesthesia. In particular, for the first time in this field, we investigated brain folding in comparing 45 sulcal shapes in each hemisphere of control and grapheme-color synesthete populations. To overcome flaws relative to data interpretation based only on p-values, common in the synesthesia literature, we report confidence intervals of effect sizes. Moreover, our statistical maps are displayed without introducing the classical, but misleading, p-value level threshold. We adopt such a methodological procedure to facilitate appropriate data interpretation and promote the "New Statistics" approach. Based on structural or diffusion magnetic resonance imaging data, we did not find any strong cerebral anomaly, in sulci, tissue volume, tissue density or fiber organization that could support synesthetic color experience. Finally, by sharing our complete datasets, we strongly support the multi-center construction of a sufficient large dataset repository for detecting, if any, subtle brain differences that may help understanding how a subjective experience, such as synesthesia, is mentally constructed.

The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)

In this paper we report the set-up and results of the Multimodal Brain Tumor Image Segmentation Benchmark (BRATS) organized in conjunction with the MICCAI 2012 and 2013 conferences. Twenty state-of-the-art tumor segmentation algorithms were applied to a set of 65 multi-contrast MR scans of low- and high-grade glioma patients-manually annotated by up to four raters-and to 65 comparable scans generated using tumor image simulation software. Quantitative evaluations revealed considerable disagreement between the human raters in segmenting various tumor sub-regions (Dice scores in the range 74%-85%), illustrating the difficulty of this task. We found that different algorithms worked best for different sub-regions (reaching performance comparable to human inter-rater variability), but that no single algorithm ranked in the top for all sub-regions simultaneously. Fusing several good algorithms using a hierarchical majority vote yielded segmentations that consistently ranked above all individual algorithms, indicating remaining opportunities for further methodological improvements. The BRATS image data and manual annotations continue to be publicly available through an online evaluation system as an ongoing benchmarking resource.

A multilayer ontology of instruments for neurological, behavioral and cognitive assessments

Advances in neuroscience are underpinned by large, multicenter studies and a mass of heterogeneous datasets. When investigating the relationships between brain anatomy and brain functions under normal and pathological conditions, measurements obtained from a broad range of brain imaging techniques are correlated with the information on each subject's neurologic states, cognitive assessments and behavioral scores derived from questionnaires and tests. The development of ontologies in neuroscience appears to be a valuable way of gathering and handling properly these heterogeneous data - particularly through the use of federated architectures. We recently proposed a multilayer ontology for sharing brain images and regions of interest in neuroimaging. Here, we report on an extension of this ontology to the representation of instruments used to assess brain and cognitive functions and behavior in humans. This extension consists of a 'core' ontology that accounts for the properties shared by all instruments supplemented by 'domain' ontologies that conceptualize standard instruments. We also specify how this core ontology has been refined to build domain ontologies dedicated to widely used instruments and how various scores used in the neurosciences are represented. Lastly, we discuss our design choices, the ontology's limitations and planned extensions aimed at querying and reasoning across distributed data sources.

Age-Related Differences in Spatial Frequency Processing during Scene Categorization

Visual analysis of real-life scenes starts with the parallel extraction of different visual elementary features at different spatial frequencies. The global shape of the scene is mainly contained in low spatial frequencies (LSF), and the edges and borders of objects are mainly contained in high spatial frequencies (HSF). The present fMRI study investigates the effect of age on the spatial frequency processing in scenes. Young and elderly participants performed a categorization task (indoor vs. outdoor) on LSF and HSF scenes. Behavioral results revealed performance degradation for elderly participants only when categorizing HSF scenes. At the cortical level, young participants exhibited retinotopic organization of spatial frequency processing, characterized by medial activation in the anterior part of the occipital lobe for LSF scenes (compared to HSF), and the lateral activation in the posterior part of the occipital lobe for HSF scenes (compared to LSF). Elderly participants showed activation only in the anterior part of the occipital lobe for LSF scenes (compared to HSF), but not significant activation for HSF (compared to LSF). Furthermore, a ROI analysis revealed that the parahippocampal place area, a scene-selective region, was less activated for HSF than LSF for elderly participants only. Comparison between groups revealed greater activation of the right inferior occipital gyrus in young participants than in elderly participants for HSF. Activation of temporo-parietal regions was greater in elderly participants irrespective of spatial frequencies. The present findings indicate a specific low-contrasted HSF deficit for normal elderly people, in association with an occipito-temporal cortex dysfunction, and a functional reorganization of the categorization of filtered scenes.

Quantitative evaluation of fMRI retinotopic maps, from V1 to V4, for cognitive experiments

FMRI retinotopic mapping is a non-invasive technique for the delineation of low-level visual areas in individual subjects. It generally relies upon the analysis of functional responses to periodic visual stimuli that encode eccentricity or polar angle in the visual field. This technique is used in vision research when the precise assignation of brain activation to retinotopic areas is an issue. It involves processing steps computed with different algorithms and embedded in various software suites. Manual intervention may be needed for some steps. Although the diversity of the available processing suites and manual interventions may potentially introduce some differences in the final delineation of visual areas, no documented comparison between maps obtained with different procedures has been reported in the literature. To explore the effect of the processing steps on the quality of the maps obtained, we used two tools, BALC, which relies on a fully automated procedure, and BrainVoyager, where areas are delineated “by hand” on the brain surface. To focus on the mapping procedures specifically, we used the same SPM pipeline for pretreatment and the same tissue segmentation tool. We document the consistency and differences of the fMRI retinotopic maps obtained from “routine retinotopy” experiments on 10 subjects. The maps obtained by skilled users are never fully identical. However, the agreement between the maps, around 80% for low-level areas, is probably sufficient for most applications. Our results also indicate that assigning cognitive activations, following a specific experiment (here, color perception), to individual retinotopic maps is not free of errors. We provide measurements of this error, that may help for the cautious interpretation of cognitive activation projection onto fMRI retinotopic maps. On average, the magnitude of the error is about 20%, with much larger differences in a few subjects. More variability may even be expected with less trained users or using different acquisition parameters and preprocessing chains.

A critical review of the neuroimaging literature on synesthesia

Synesthesia refers to additional sensations experienced by some people for specific stimulations, such as the systematic arbitrary association of colors to letters for the most studied type. Here, we review all the studies (based mostly on functional and structural magnetic resonance imaging) that have searched for the neural correlates of this subjective experience, as well as structural differences related to synesthesia. Most differences claimed for synesthetes are unsupported, due mainly to low statistical power, statistical errors, and methodological limitations. Our critical review therefore casts some doubts on whether any neural correlate of the synesthetic experience has been established yet. Rather than being a neurological condition (i.e., a structural or functional brain anomaly), synesthesia could be reconsidered as a special kind of childhood memory, whose signature in the brain may be out of reach with present brain imaging techniques.

Contributions of contour frequency, amplitude, and luminance to the watercolor effect estimated by conjoint measurement

The watercolor effect is a long-range, assimilative, filling-in phenomenon induced by a pair of distant, wavy contours of different chromaticities. Here, we measured joint influences of the contour frequency and amplitude and the luminance of the interior contour on the strength of the effect. Contour pairs, each enclosing a circular region, were presented with two of the dimensions varying independently across trials (luminance/frequency, luminance/amplitude, frequency/amplitude) in a conjoint measurement paradigm (Luce & Tukey, 1964). In each trial, observers judged which of the stimuli evoked the strongest fill-in color. Control stimuli were identical except that the contours were intertwined and generated little filling-in. Perceptual scales were estimated by a maximum likelihood method (Ho, Landy, & Maloney, 2008). An additive model accounted for the joint contributions of any pair of dimensions. As shown previously using difference scaling (Devinck & Knoblauch, 2012), the strength increases with luminance of the interior contour. The strength of the phenomenon was nearly independent of the amplitude of modulation of the contour but increased with its frequency up to an asymptotic level. On average, the strength of the effect was similar along a given dimension regardless of the other dimension with which it was paired, demonstrating consistency of the underlying estimated perceptual scales.

Spatial selectivity of the watercolor effect

The spatial selectivity of the watercolor effect (WCE) was assessed by measuring its strength as a function of the luminance contrast of its inducing contours for different spatial configurations, using a maximum likelihood scaling procedure. The approach has previously been demonstrated to provide an efficient method for investigating the WCE as well as other perceptual dimensions. We show that the strength is narrowly tuned to the width of the contour, that it is optimal when its pair of inducing contours are of equal width, and that the strength can be increased by varying the overall size of the stimulus when the width of the inducing contour is not optimal. The results support a neural substrate that has characteristics not unlike double-opponent, color-luminance cells observed in cortical area V1.

Feasibility and reliability of an automated controller of inspired oxygen concentration during mechanical ventilation

INTRODUCTION

Hypoxemia and high fractions of inspired oxygen (FiO2) are concerns in critically ill patients. An automated FiO2 controller based on continuous oxygen saturation (SpO2) measurement was tested. Two different SpO2-FiO2 feedback open loops, designed to react differently based on the level of hypoxemia, were compared. The results of the FiO2 controller were also compared with a historical control group.

METHODS

The system measures SpO2, compares with a target range (92% to 96%), and proposes in real time FiO2 settings to maintain SpO2 within target. In 20 patients under mechanical ventilation, two different FiO2-SpO2 open loops were applied by a dedicated research nurse during 3 hours, each in random order. The times spent in and outside the target SpO2 values were measured. The results of the automatic controller were then compared with a retrospective control group of 30 ICU patients. SpO2-FiO2 values of the control group were collected over three different periods of 6 hours.

RESULTS

Time in the target range was higher than 95% with the controller. When the 20 patients were separated according to the median PaO2/FiO2 (160(133-176) mm Hg versus 239(201-285)), the loop with the highest slope was slightly better (P = 0.047) for the more-hypoxemic patients. Hyperoxemia and hypoxemia durations were significantly shorter with the controller compared with usual care: SpO2 target range was reached 90% versus 24%, 27% and 32% (P < .001) with the controller, compared with three historical control-group periods.

CONCLUSION

A specific FiO2 controller is able to maintain SpO2 reliably within a predefined target range. Two different feedback loops can be used, depending on the initial PaO2/FiO2; with both, the automatic controller showed excellent performance when compared with usual care.

Fast joint detection-estimation of evoked brain activity in event-related FMRI using a variational approach

In standard within-subject analyses of event-related functional magnetic resonance imaging (fMRI) data, two steps are usually performed separately: detection of brain activity and estimation of the hemodynamic response. Because these two steps are inherently linked, we adopt the so-called region-based joint detection-estimation (JDE) framework that addresses this joint issue using a multivariate inference for detection and estimation. JDE is built by making use of a regional bilinear generative model of the BOLD response and constraining the parameter estimation by physiological priors using temporal and spatial information in a Markovian model. In contrast to previous works that use Markov Chain Monte Carlo (MCMC) techniques to sample the resulting intractable posterior distribution, we recast the JDE into a missing data framework and derive a variational expectation-maximization (VEM) algorithm for its inference. A variational approximation is used to approximate the Markovian model in the unsupervised spatially adaptive JDE inference, which allows automatic fine-tuning of spatial regularization parameters. It provides a new algorithm that exhibits interesting properties in terms of estimation error and computational cost compared to the previously used MCMC-based approach. Experiments on artificial and real data show that VEM-JDE is robust to model misspecification and provides computational gain while maintaining good performance in terms of activation detection and hemodynamic shape recovery.

Retinotopic and lateralized processing of spatial frequencies in human visual cortex during scene categorization

Using large natural scenes filtered in spatial frequencies, we aimed to demonstrate that spatial frequency processing could not only be retinotopically mapped but could also be lateralized in both hemispheres. For this purpose, participants performed a categorization task using large black and white photographs of natural scenes (indoors vs. outdoors, with a visual angle of 24° × 18°) filtered in low spatial frequencies (LSF), high spatial frequencies (HSF), and nonfiltered scenes, in block-designed fMRI recording sessions. At the group level, the comparison between the spatial frequency content of scenes revealed first that, compared with HSF, LSF scene categorization elicited activation in the anterior half of the calcarine fissures linked to the peripheral visual field, whereas, compared with LSF, HSF scene categorization elicited activation in the posterior part of the occipital lobes, which are linked to the fovea, according to the retinotopic property of visual areas. At the individual level, functional activations projected on retinotopic maps revealed that LSF processing was mapped in the anterior part of V1, whereas HSF processing was mapped in the posterior and ventral part of V2, V3, and V4. Moreover, at the group level, direct interhemispheric comparisons performed on the same fMRI data highlighted a right-sided occipito-temporal predominance for LSF processing and a left-sided temporal cortex predominance for HSF processing, in accordance with hemispheric specialization theories. By using suitable method of analysis on the same data, our results enabled us to demonstrate for the first time that spatial frequencies processing is mapped retinotopically and lateralized in human occipital cortex.

Two new stable anatomical landmarks on the Central Sulcus: definition, automatic detection, and their relationship with primary motor functions of the hand

We present a method to automatically detect two new stable anatomical landmarks L(1) and L(2) on the Central Sulcus (CS). Those landmarks are shown to be representative of the Central Sulcus morphology and linked to the functional primary motor area of the hand. Detection is performed after introducing a new morphological characteristic, the sulcal profile. We show that when matching explicitly L(1) and L(2) across individuals the inter-subject matching of the central sulcus anatomy is improved, as well as the inter-subject matching of the primary motor area of the hand. This opens possibilities for morphological studies of the CS, more precise functional studies of primary motor function, and a better understanding of motor representations along the CS.

A BOLD signature of eyeblinks in the visual cortex

We are usually unaware of the brief but large illumination changes caused by blinks, presumably because of blink suppression mechanisms. In fMRI however, increase of the BOLD signal was reported in the visual cortex, e.g. during blocks of voluntary blinks (Bristow, Frith and Rees, 2005) or after spontaneous blinks recorded during the prolonged fixation of a static stimulus (Tse, Baumgartner and Greenlee, 2010). We tested whether such activation, possibly related to illumination changes, was also present during standard fMRI retinotopic and visual experiments and was large enough to contaminate the BOLD signal we are interested in. We monitored in a 3T scanner the eyeblinks of 14 subjects who observed three different types of visual stimuli, including periodic rotating wedges and contracting/expanding rings, event-related Mondrians and graphemes, while fixating. We performed event-related analyses on the set of detected spontaneous blinks. We observed large and widespread BOLD responses related to blinks in the visual cortex of every subject and whatever the visual stimulus. The magnitude of the modulation was comparable to visual stimulation. However, blink-related activations lay mostly in the anterior parts of retinotopic visual areas, coding the periphery of the visual field well beyond the extent of our stimuli. Blinks therefore represent an important source of BOLD variations in the visual cortex and a troublesome source of noise since any correlation, even weak, between the distribution of blinks and a tested protocol could trigger artifactual activities. However, the typical signature of blinks along the anterior calcarine and the parieto-occipital sulcus allows identifying, even in the absence of eyetracking, fMRI protocols possibly contaminated by a heterogeneous distribution of blinks.

NeuroLOG: sharing neuroimaging data using an ontology-based federated approach

This paper describes the design of the NeuroLOG middleware data management layer, which provides a platform to share heterogeneous and distributed neuroimaging data using a federated approach. The semantics of shared information is captured through a multi-layer application ontology and a derived Federated Schema used to align the heterogeneous database schemata from different legacy repositories. The system also provides a facility to translate the relational data into a semantic representation that can be queried using a semantic search engine thus enabling the exploitation of knowledge embedded in the ontology. This work shows the relevance of the distributed approach for neurosciences data management. Although more complex than a centralized approach, it is also more realistic when considering the federation of large data sets, and open strong perspectives to implement multi-centric neurosciences studies.

The neural bases of grapheme-color synesthesia are not localized in real color-sensitive areas

The subjective experience of color by synesthetes when viewing achromatic letters and numbers supposedly relates to real color experience, as exemplified by the recruitment of the V4 color center observed in some brain imaging studies. Phenomenological reports and psychophysics tests indicate, however, that both experiences are different. Using functional magnetic resonance imaging, we tried to precise the degree of coactivation by real and synesthetic colors, by evaluating each color center individually, and applying adaptation protocols across real and synesthetic colors. We also looked for structural differences between synesthetes and nonsynesthetes. In 10 synesthetes, we found that color areas and retinotopic areas were not activated by synesthetic colors, whatever the strength of synesthetic associations measured objectively for each subject. Voxel-based morphometry revealed no white matter (WM) or gray matter difference in those regions when compared with 25 control subjects. But synesthetes had more WM in the retrosplenial cortex bilaterally. The joint coding of real and synesthetic colors, if it exists, must therefore be distributed rather than localized in the visual cortex. Alternatively, the key to synesthetic color experience might not lie in the color system.

Combination of nonlinear registration methods with high resolution fMRI for a fine exploration of human primary motor hand area

Functional investigation of human hand representation in the motor area M1 requires high resolution functional imaging, to finely separate activation in M1, and a perfect alignment of individual central sulci to improve functional areas overlap and significance of statistical parametric maps obtained from different hand movements. Based on anatomical measures, we show how recent global diffeomorphic registration techniques impact positively on the alignment of sulcal folds in M1 area. With functional measures, we evaluate their effect on the robust detection and localization of group brain activation for flexion/extension of right and left thumbs/fingers and wrists. The methodology we propose opens the way to a non invasive functional exploration of the human hand motor cortex at the group level under different normal, pathological or after rehabilitative conditions.

fMRI retinotopic mapping at 3 T: benefits gained from correcting the spatial distortions due to static field inhomogeneity

fMRI retinotopic mapping usually relies upon Fourier analysis of functional responses to periodic visual stimuli that encode eccentricity or polar angle in the visual field. Generally, phase estimations are assigned to a surface model of the cerebral cortex and borders between retinotopic areas are eventually determined following ad hoc phase analysis on the surface model. Assigning functional responses to a surface model of the cortex is particularly sensitive to geometric distortions of the 3D functional data due to static field inhomogeneity. Here, we assess and document the benefits gained from correcting the fMRI data for these effects, under standard experimental conditions (echo-planar imaging, 3.0-T field strength) and with well-chosen acquisition parameters (regarding slice orientation and phase-encoding direction). While it appears that, in the absence of correction, errors in the estimates of the borders between low-order visual areas do not then significantly exceed the variance of statistical origin, about half of the functional responses in a retinotopic experiment are misassigned to neighboring functional areas. Therefore, correction of the effects due to geometric distortions is important in any retinotopic mapping experiment and by extension in any fMRI experiment on the visual system.

Grid-wide neuroimaging data federation in the context of the NeuroLOG project

Grid technologies are appealing to deal with the challenges raised by computational neurosciences and support multi-centric brain studies. However, core grids middleware hardly cope with the complex neuroimaging data representation and multi-layer data federation needs. Moreover, legacy neuroscience environments need to be preserved and cannot be simply superseded by grid services. This paper describes the NeuroLOG platform design and implementation, shedding light on its Data Management Layer. It addresses the integration of brain image files, associated relational metadata and neuroscience semantic data in a heterogeneous distributed environment, integrating legacy data managers through a mediation layer.

Distributed local MRF models for tissue and structure brain segmentation

Accurate tissue and structure segmentation of magnetic resonance (MR) brain scans is critical in several applications. In most approaches this task is handled through two sequential steps. We propose to carry out cooperatively both tissue and subcortical structure segmentation by distributing a set of local and cooperative Markov random field (MRF) models Tissue segmentation is performed by partitioning the volume into subvolumes where local MRFs are estimated in cooperation with their neighbors to ensure consistency. Local estimation fits precisely to the local intensity distribution and thus handles nonuniformity of intensity without any bias field modelization. Similarly, subcortical structure segmentation is performed via local MRF models that integrate localization constraints provided by a priori fuzzy description of brain anatomy. Subcortical structure segmentation is not reduced to a subsequent processing step but joined with tissue segmentation: the two procedures cooperate to gradually and conjointly improve model accuracy. We propose a framework to implement this distributed modeling integrating cooperation, coordination, and local model checking in an efficient way. Its evaluation was performed using both phantoms and real 3 T brain scans, showing good results and in particular robustness to nonuniformity and noise with a low computational cost. This original combination of local MRF models, including anatomical knowledge, appears as a powerful and promising approach for MR brain scan segmentation.

Towards an ontology for sharing medical images and regions of interest in neuroimaging

The goal of the NeuroBase project is to facilitate collaborative research in neuroimaging through a federated system based on semantic web technologies. The cornerstone and focus of this paper is the design of a common semantic model providing a unified view on all data and tools to be shared. For this purpose, we built a multi-layered and multi-components formal ontology. This paper presents two major contributions. The first is related to the general methodology we propose for building an application ontology based on consistent conceptualization choices provided by the DOLCE foundational ontology and core ontologies of domains that we reuse; the second concerns the domain ontology we designed for neuroimaging, which encompasses both the objective nature of image data and the subjective nature of image content, through annotations based on regions of interest made by agents (humans or computer programs). We report on realistic domain use-case queries referring to our application ontology.

Multimodal MRI segmentation of ischemic stroke lesions

The problem addressed in this paper is the automatic segmentation of stroke lesions on MR multi-sequences. Lesions enhance differently depending on the MR modality and there is an obvious gain in trying to account for various sources of information in a single procedure. To this aim, we propose a multimodal Markov random field model which includes all MR modalities simultaneously. The results of the multimodal method proposed are compared with those obtained with a mono-dimensional segmentation applied on each MRI sequence separately. We constructed an Atlas of blood supply territories to help clinicians in the determination of stroke subtypes and potential functional deficit.

A multicenter randomized trial of computer-driven protocolized weaning from mechanical ventilation

RATIONALE AND OBJECTIVES

Duration of weaning from mechanical ventilation may be reduced by the use of a systematic approach. We assessed whether a closed-loop knowledge-based algorithm introduced in a ventilator to act as a computer-driven weaning protocol can improve patient outcomes as compared with usual care.

METHODS AND MEASUREMENTS

We conducted a multicenter randomized controlled study with concealed allocation to compare usual care for weaning with computer-driven weaning. The computerized protocol included an automatic gradual reduction in pressure support, automatic performance of spontaneous breathing trials (SBT), and generation of an incentive message when an SBT was successfully passed. One hundred forty-four patients were enrolled before weaning initiation. They were randomly allocated to computer-driven weaning or to physician-controlled weaning according to local guidelines. Weaning duration until successful extubation and total duration of ventilation were the primary endpoints.

MAIN RESULTS

Weaning duration was reduced in the computer-driven group from a median of 5 to 3 d (p=0.01) and total duration of mechanical ventilation from 12 to 7.5 d (p=0.003). Reintubation rate did not differ (23 vs. 16%, p=0.40). Computer-driven weaning also decreased median intensive care unit (ICU) stay duration from 15.5 to 12 d (p=0.02) and caused no adverse events. The amount of sedation did not differ between groups. In the usual care group, compliance to recommended modes and to SBT was estimated, respectively, at 96 and 51%.

CONCLUSIONS

The specific computer-driven system used in this study can reduce mechanical ventilation duration and ICU length of stay, as compared with a physician-controlled weaning process.

Global integration of local color differences in transparency perception: An fMRI study

In normal viewing, the visual system effortlessly assigns approximately constant attributes of color and shape to perceived objects. A fundamental component of this process is the compensation for illuminant variations and intervening media to recover reflectance properties of natural surfaces. We exploited the phenomenon of transparency perception to explore the cortical regions implicated in such processes, using fMRI. By manipulating the coherence of local color differences around a region in an image, we interfered with their global perceptual integration and thereby modified whether the region appeared transparent or not. We found the major cortical activation due to global integration of local color differences to be in the anterior part of the parahippocampal gyrus. Regions differentially activated by chromatic versus achromatic geometric patterns showed no significant differential response related to the coherence/incoherence of local color differences. The results link the integration of local color differences in the extraction of a transparent layer with sites activated by object-related properties of an image.