Development and characterization of deep learning techniques for neuroimaging data

Deep learning methods are extremely promising machine learning tools to analyze neuroimaging data. However, their potential use in clinical settings is limited because of the existing challenges of applying these methods to neuroimaging data. In this study, first a data leakage type caused by slice-level data split that is introduced during training and validation of a 2D CNN is surveyed and a quantitative assessment of the model’s performance overestimation is presented. Second, an interpretable, leakage-fee deep learning software written in a python language with a wide range of options has been developed to conduct both classification and regression analysis. The software was applied to the study of mild cognitive impairment (MCI) in patients with small vessel disease (SVD) using multi-parametric MRI data where the cognitive performance of 58 patients measured by five neuropsychological tests is predicted using a multi-input CNN model taking brain image and demographic data. Each of the cognitive test scores was predicted using different MRI-derived features. As MCI due to SVD has been hypothesized to be the effect of white matter damage, DTI-derived features MD and FA produced the best prediction outcome of the TMT-A score which is consistent with the existing literature. In a second study, an interpretable deep learning system aimed at 1) classifying Alzheimer disease and healthy subjects 2) examining the neural correlates of the disease that causes a cognitive decline in AD patients using CNN visualization tools and 3) highlighting the potential of interpretability techniques to capture a biased deep learning model is developed. Structural magnetic resonance imaging (MRI) data of 200 subjects was used by the proposed CNN model which was trained using a transfer learning-based approach producing a balanced accuracy of 71.6%. Brain regions in the frontal and parietal lobe showing the cerebral cortex atrophy were highlighted by the visualization tools.

The cerebral sulci and gyri

The aim of this study was to describe in detail the microanatomy of the cerebral sulci and gyri, clarifying the nomenclature for microneurosurgical purposes. An extensive review of the literature regarding the historical, evolutionary, embryological, and anatomical aspects pertinent to human cerebral sulci and gyri was conducted, with a special focus on microneuroanatomy issues in the field of neurosurgery. An intimate knowledge of the cerebral sulci and gyri is needed to understand neuroimaging studies, as well as to plan and execute current microneurosurgical procedures. (DOI: 10.3171/2009.11.FOCUS09245)

Antinociceptive effect of stimulating the occipital or retrosplenial cortex in rats

A role for the occipital or retrosplenial cortex in nociceptive processing has not been demonstrated yet, but connections from these cortices to brain structures involved in descending pain-inhibitory mechanisms were already demonstrated. This study demonstrated that the electrical stimulation of the occipital or retrosplenial cortex produces antinociception in the rat tail-flick and formalin tests. Bilateral lesions of the dorsolateral funiculus abolished the effect of cortical stimulation in the tail-flick test. Injection of glutamate into the same targets was also antinociceptive in the tail-flick test. No rats stimulated in the occipital or retrosplenial cortex showed any change in motor performance on the Rota-rod test, or had epileptiform changes in the EEG recording during or up to 3 hours after stimulation. The antinociception induced by occipital cortex stimulation persisted after neural block of the retrosplenial cortex. The effect of retrosplenial cortex stimulation also persisted after neural block of the occipital cortex. We conclude that stimulation of the occipital or retrosplenial cortex in rats leads to antinociception activating distinct descending pain-inhibitory mechanisms, and this is unlikely to result from a reduced motor performance or a postictal phenomenon. Perspective: This study presents evidence that stimulation of the retrosplenial or occipital cortex produces antinociception in rat models of acute pain. These findings enhance our understanding of the role of the cerebral cortex in control of pain. (C) 2010 by the American Pain Society

Immunology of multiple sclerosis

Multiple sclerosis (MS) is an autoimmune disease of the central nervous system (CNS) leading to demyelination, axonal damage, and progressive neurologic disability. The development of MS is influenced by environmental factors, particularly the Epstein-Barr virus (EBV), and genetic factors, which include specific HLA types, particularly DRB1*1501-DQA1*0102-DQB1*0602, and a predisposition to autoimmunity in general. MS patients have increased circulating T-cell and antibody reactivity to myelin proteins and gangliosides. It is proposed that the role of EBV is to infect autoreactive B cells that then seed the CNS and promote the survival of autoreactive T cells there. It is also proposed that the clinical attacks of relapsing-remitting MS are orchestrated by myelin-reactive T cells entering the white matter of the CNS from the blood, and that the progressive disability in primary and secondary progressive MS is caused by the action of autoantibodies produced in the CNS by ­meningeal lymphoid follicles with germinal centers.

The expression of contextual fear conditioning involves activation of an NMDA receptor-nitric oxide pathway in the medial prefrontal cortex

The ventral portion of medial prefrontal cortex (vMPFC) is involved in contextual fear-conditioning expression in rats. In the present study, we investigated the role of local N-methyl-D-aspartic acid (NMDA) glutamate receptors and nitric oxide (NO) in vMPFC on the behavioral (freezing) and cardiovascular (increase of arterial pressure and heart rate) responses of rats exposed to a context fear conditioning. The results showed that both freezing and cardiovascular responses to contextual fear conditioning were reduced by bilateral administration of NMDA receptor antagonist LY235959 (4 nmol/200 nL) into the vMPFC before reexposition to conditioned chamber. Bilateral inhibition of neuronal NO synthase (nNOS) by local vMPFC administration of the N omega-propyl-L-arginine (N-propyl, 0.04 nmol/200 nL) or the NO scavenger carboxy-PTI0 (1 nmol/200 A) caused similar results, inhibiting the fear responses. We also investigated the effects of inhibiting glutamate- and NO-mediated neurotransmission in the vMPFC at the time of aversive context exposure on reexposure to the same context. It was observed that the 1st exposure results in a significant attenuation of the fear responses on reexposure in vehicle-treated animals, which was not modified by the drugs. The present results suggest that a vMPFC NMDA-NO pathway may play an important role on expression of contextual fear conditioning.

Role of the superior colliculus in the expression of acute and kindled audiogenic seizures in Wistar audiogenic rats

P>Purpose: The role of the superior colliculus (SC) in seizure expression is controversial and appears to be dependent upon the epilepsy model. This study shows the effect of disconnection between SC deep layers and adjacent tissues in the expression of acute and kindling seizures. Methods: Subcollicular transections, ablation of SC superficial and deep layers, and ablation of only the cerebral cortex were evaluated in the Wistar audiogenic rat (WAR) strain during acute and kindled audiogenic seizures. The audiogenic seizure kindling protocol started 4 days after surgeries, with two acoustic stimuli per day for 10 days. Acute audiogenic seizures were evaluated by a categorized seizure severity midbrain index (cSI) and kindled seizures by a severity limbic index (LI). Results: All subcollicular transections reaching the deep layers of the SC abolished audiogenic seizures or significantly decreased cSI. In the unlesioned kindled group, a reciprocal relationship between limbic and brainstem pattern of seizures was seen. The increased number of stimuli provoked an audiogenic kindling phenomenon. Ablation of the entire SC (ablation group) or of the cerebral cortex only (ctx-operated group) hampered the acquisition of limbic behaviors. There was no difference in cSI and LI between the ctx-operated and ablation groups, but there was a difference between ctx-operated and the unlesioned kindled group. There was also no difference in cSI between SC deep layer transection and ablation groups. Results of histologic analyses were similar for acute and kindled audiogenic seizure groups. Conclusions: SC deep layers are involved in the expression of acute and kindled audiogenic seizure, and the cerebral cortex is essential for audiogenic kindling development.

Mild cognitive deficits associated to neocortical microgyria in mice with genetic deletion of cellular prion protein

The cellular prion protein (PrP(c)) has been implicated with the modulation of neuronal apoptosis, adhesion, neurite outgrowth and maintenance which are processes involved in the neocortical development. Malformations of cortical development (MCD) are frequently associated with neurological conditions including mental retardation, autism, and epilepsy. Here we investigated the behavioral performance of female adult PrP(c)-null mice (Prnp(%)) and their wild-type controls (Prnp(+/+)) presenting unilateral polymicrogyria, a MCD experimentally induced by neonatal freeze-lesion in the right hemisphere. injured mice from both genotypes presented similar locomotor activity but Prnp(%) mice showed a tendency to increase anxiety-related responses when compared to Prnp(+/+) animals. Additionally, injured Prnp(%) mice have a poorer performance in the social recognition task than sham-operated and Prnp(%) injured ones. Moreover the step-down inhibitory avoidance task was not affected by the procedure or the genotype of the animals. These data suggest that the genetic deletion of PrP(c) confers increased susceptibility to short-term social memory deficits induced by neonatal freezing model of polymicrogyria in mice. (C) 2008 Published by Elsevier B.V.

CHARACTERIZATION OF Kiss1 NEURONS USING TRANSGENIC MOUSE MODELS

Humans and mice with loss-of-function mutations of the genes encoding kisspeptins (Kiss1) or kisspeptin receptor (Kiss1r) are infertile due to hypogonadotropic hypogonadism. Within the hypothalamus, Kiss1 mRNA is expressed in the anteroventral periventricular nucleus (AVPV) and the arcuate nucleus (Arc). In order to better study the different populations of kisspeptin cells we generated Kiss1-Cre transgenic mice. We obtained one line with Cre activity specifically within Kiss1 neurons (line J2-4), as assessed by generating mice with Cre-dependent expression of green fluorescent protein or beta-galactosidase. Also, we demonstrated Kiss1 expression in the cerebral cortex and confirmed previous data showing Kiss1 mRNA in the medial nucleus of amygdala and anterodorsal preoptic nucleus. Kiss1 neurons were more concentrated towards the caudal levels of the Arc and higher leptin-responsivity was observed in the most caudal population of Arc Kiss1 neurons. No evidence for direct action of leptin in AVPV Kiss1 neurons was observed. Me lanocortin fibers innervated subsets of Kiss1 neurons of the preoptic area and Arc, and both populations expressed melanocortin receptors type 4 (MC4R). Specifically in the preoptic area, 18-28% of Kiss1 neurons expressed MC4R. In the Arc, 90% of Kiss1 neurons were glutamatergic, 50% of which also were GABAergic. In the AVPV, 20% of Kiss1 neurons were glutamatergic whereas 75% were GABAergic. The differences observed between the Kiss1 neurons in the preoptic area and the Arc likely represent neuronal evidence for their differential roles in metabolism and reproduction. (C) 2011 IBRO. Published by Elsevier Ltd. All rights reserved.

The human temporal cortex: Characterization of neurons expressing nitric oxide synthase, neuropeptides and calcium-binding proteins, and their glutamate receptor subunit profiles

Immunocytochemical techniques were used to examine the distribution of neurons immunoreactive (-ir) for nitric oxide synthase (nNOS), somatostatin (SOM), neuropeptide Y (NPY), parvalbumin (PV), calbindin (CB) and calretinin (CH), in the inferotemporal gyros (Brodmann's area 21) of the human neocortex. Neurons that colocalized either nNOS or SOM with PV, CB or CR were also identified by double-labeling techniques. Furthermore, glutamate receptor subunit profiles (GluR1, GluR2/3, GluR2/4, GluR5/6/7 and NMDAR1) were also determined for these cells. The number and distribution of cells containing nNOS, SOM, NPY, PV, CB or CR differed for each antigen. In addition, distinct subpopulations of neurons displayed different degrees of colocalization of these antigens depending on which antigens were compared. Moreover, cells that contained nNOS, SOM, NPY, PV, GB or CR expressed different receptor subunit profiles. These results show that specific subpopulations of neurochemically identified nonpyramidal cells may be activated via different receptor subtypes. As these different subpopulations of cells project to specific regions of pyramidal calls, facilitation of subsets of these cells via different receptor subunits may activate different inhibitory circuits. Thus, various distinct, but overlapping, inhibitory circuits may act in concert in the modulation of normal cortical function, plasticity and disease.

Pyramidal neurones in macaque visual cortex: Interareal phenotypic variation of dendritic branching patterns

The basal dendritic arbors of over 500-layer III pyramidal neurones of the macaque cortex were compared by fractal analyses, which provides a measure of the space filling (or branching pattern) of dendritic arbors. Fractal values (D) of individual cells were compared between the cytochrome oxidase (CO)-rich blobs and CO-poor interblobs, of middle and upper layer III, and between sublaminae, in the primary visual area (Vi). These data were compared with those in the CO compartments in the second visual area (V2), and seven other extrastriate cortical areas. (V4, MT, LIP, 7a, TEO, TE and STP). There were significant differences in the fractal dimensions, and therefore the dendritic branching patterns, of cells in striate and extrastriate areas. Of the 55 possible pairwise comparisons of fractal dimension of neurones in different cortical areas (or CO compartments), 39 proved to be significantly different. The markedly different morphologies of pyramidal cells in the different cortical areas may be one of the features that determine the functional signatures of these cells by influencing the number of inputs received by, and propagation of potentials through, their dendritic arbors.

GABA(A) receptor sites in the developing human foetus

GABA(A) receptor sites were characterised in cerebral cortex tissue samples from deceased neurologically normal infants who had come to autopsy during the third trimester of pregnancy. Pharmacological parameters were obtained from homogenate binding studies which utilised the 'central-type' benzodiazepine ligands [H-3]diazepam and [H-3]flunitrazepam, and from the GABA activation of [H-3]diazepam binding. It was found that the two radioligands behaved differently during development. The affinity of [H-3]flunitrazepam for its binding site did not vary significantly between preparations, whereas the [H-3]diazepam K-D showed marked regional and developmental variations: infant tissues showed a distinctly lower affinity than adults for this ligand. The density of [H-3]flunitrazepam binding sites increased similar to35% during the third trimester to reach adult levels by term, whereas [H-3]diazepam binding capacity declined slightly but steadily throughout development. The GABA activation of [H-3]diazepam binding was less efficient early in the trimester, in that the affinity of the agonist was significantly lower, though it rose to adult levels by term. The strength of the enhancement response increased to adult levels over the same time-frame. The results strongly suggest that the subunit composition of cortical GABA(A) sites changes significantly during this important developmental stage. (C) 2002 Elsevier Science B.V. All rights reserved.

The pyramidal cell of the sensorimotor cortex of the macaque monkey: Phenotypic variation

Recent studies have revealed striking differences in pyramidal cell structure among cortical regions involved in the processing of different functional modalities. For example, cells involved in visual processing show systematic variation, increasing in morphological complexity with rostral progression from V1 through extrastriate areas. Differences have also been identified between pyramidal cells in somatosensory, motor and prefrontal cortex, but the extent to which the pyramidal cell phenotype may vary between these functionally related cortical regions remains unknown. In the present study we investigated the structure of layer III pyramidal cells in somatosensory and motor areas 3b, 4, 5, 6 and 7b of the macaque monkey. Cells were intracellularly injected in fixed, flat-mounted cortical slices and analysed for morphometric parameters. The size of the basal dendritic arbours, the number of their branches and their spine density were found to vary systematically between areas. Namely, we found a trend for increasing complexity in dendritic arbour structure through areas 3b, 5 and 7b. A similar trend occurred through areas 4 and 6. The differences in arbour structure may determine the number of inputs received by neurons and may thus be an important factor in determining function at the cellular and systems level.

Nutritional effects on neuron numbers

Undemutrition during early life is known to cause deficits and distortions of brain structure although it has remained uncertain whether or not this includes a diminution of the total numbers of neurons. Estimates of numerical density (e.g. number of cells per microscopic field, or number of cells per unit area of section, or number of cells per unit volume of tissue) are extremely difficult to interpret and do not provide estimates of total numbers of cells. However, advances in stereological techniques have made it possible to obtain unbiased estimates of total numbers of cells in well defined biological structures. These methods have been utilised in studies to determine the effects of varying periods of undernutrition during early life on the numbers of neurons in various regions of the rat brain. The regions examined so far have included the cerebellum, the dentate gyrus, the olfactory bulbs and the cerebral cortex. The only region to show, unequivocally, that a period of undernutrition during early life causes a deficit in the number of neurons was the dentate gyrus. These findings are discussed in the context of other morphological and functional deficits present in undernourished animals.

Alterations of neocortical pyramidal cell phenotype in the Ts65Dn mouse model of Down syndrome: Effects of environmental enrichment

Mental retardation in individuals with Down syndrome (DS) is thought to result from anomalous development and function of the brain; however, the underlying neuropathological processes have yet to be determined. Early implementation of special care programs result in limited, and temporary, cognitive improvements in DS individuals. In the present study, we investigated the possible neural correlates of these limited improvements. More specifically, we studied cortical pyramidal cells in the frontal cortex of Ts65Dn mice, a partial trisomy of murine chromosome 16 (MMU16) model characterized by cognitive deficits, hyperactivity, behavioral disruption and reduced attention levels similar to those observed in DS, and their control littermates. Animals were raised either in a standard or in an enriched environment. Environmental enrichment had a marked effect on pyramidal cell structure in control animals. Pyramidal cells in environmentally enriched control animals were significantly more branched and more spinous than non-enriched controls. However, environmental enrichment had little effect on pyramidal cell structure in Ts65Dn mice. As each dendritic spine receives at least one excitatory input, differences in the number of spines found in the dendritic arbors of pyramidal cells in the two groups reflect differences in the number of excitatory inputs they receive and, consequently, complexity in cortical circuitry. The present results suggest that behavioral deficits demonstrated in the Ts65Dn model could be attributed to abnormal circuit development.

Parvalbumin-, calbindin-, and calretinin-immunoreactive neurons in the prefrontal cortex of the owl monkey (Aotus trivirgatus): A standardized quantitative comparison with sensory and motor areas

Recent studies have revealed regional variation in the density and distribution of inhibitory neurons in different cortical areas, which are thought to reflect area-specific specializations in cortical circuitry. However, there are as yet few standardized quantitative data regarding how the inhibitory circuitry in prefrontal cortex (PFC), which is thought to be involved in executive functions such as cognition, emotion and decision making, compares to that in other cortical areas. Here we used immunohistochemical techniques to determine the density and distribution of parvalbumin (PV)-, calbindin (CB)-, and calretinin (CR)-immunoreactive (ir) neurons and axon terminals in the dorsolateral and orbital PFC of the owl monkey (Aotus trivirgatus), and compared them directly with data obtained using the same techniques in 11 different visual, somatosensory and motor areas. We found marked differences in the density of PV-ir, CB-ir, and CR-ir interneurons in several cortical areas. One hundred and twenty eight of all 234 possible between-area pairwise comparisons were significantly different. The density of specific subpopulations of these cells also varied among cortical areas, as did the density of axon terminals. Comparison of PFC with other cortical areas revealed that 40 of all 66 possible statistical comparisons of the density of PV-ir, CB-ir, and CR-ir cells were significantly different. We also found evidence for heterogeneity in the pattern of labeling of PV-ir, CB-ir, and CR-ir cells and axon terminals between the dorsolateral and orbital subdivisions of PFC. These data are likely to reflect basic differences in interneuron circuitry, which are likely to influence inhibitory function in the cortex. Copyright (C) 2003 S. Karger AG, Basel.