Search

In this paper we introduce a biologically inspired Convolutional Neural Network (CNN) architecture called LGN-CNN that has a first convolutional layer composed of a single filter that mimics the role of the Lateral Geniculate Nucleus (LGN). The first layer of the neural network shows a rotational symmetric pattern justified by the structure of the net itself that turns up to be an approximation of a Laplacian of Gaussian (LoG). The latter function is in turn a good approximation of the receptive field profiles (RFPs) of the cells in the LGN. The analogy with the visual system is established, emerging directly from the architecture of the neural network. A proof of rotation invariance of the first layer is given on a fixed LGN-CNN architecture and the computational results are shown. Thus, contrast invariance capability of the LGN-CNN is investigated and a comparison between the Retinex effects of the first layer of LGN-CNN and the Retinex effects of a LoG is provided on different images. A statistical study is done on the filters of the second convolutional layer with respect to biological data. In conclusion, the model we have introduced approximates well the RFPs of both LGN and V1 attaining similar behavior as regards long range connections of LGN cells that show Retinex effects. (C) 2021 Elsevier Ltd. All rights reserved.

Linear and nonlinear systems analysis are tools that can be used to study communication systems like the visual system. The first step of systems analysis often is to test whether or not the system is linear. Retinal pathways are surprisingly linear, and some neurons in the visual cortex also emulate linear sensory transducers. We conclude that the retinal linearity depends on specialized ribbon synapses while cortical linearity is the result of balanced excitatory and inhibitory synaptic interactions.

An attempt was made to clarify whether the site of postchiasmal lesions affects subjective perception of homonymous visual field defects during stimulation with flickering random dot patterns (white noise-field). Out of 56 patients with homonymous hemianopia, 38 (68%) perceived scotomata in this situation, but 18 (32%) discerned none at all. Neuroradiologic superposition of cerebral lesions detected by computed tomography (CT) or magnetic resonance imaging (MRI) showed that nearly all patients who perceived their scotomata had lesions involving the primary visual cortex or the perigeniculate region, whereas those who received no scotoma had lesions centered within the optic radiation. Functional MRI of six normal subjects during stimulation with flickering random dot patterns indicated predominant activation of the primary visual cortex. Since noise-field defects were most frequently perceived by patients whose lesion involved the primary visual cortex, it appears that the sensitivity of noise-field campimetry depends on the site of damage in the visual pathway. The explanation for this may be that damage to long-range horizontal connections impairs filling-in processes.

A recent computer vision technique for object classification in still images is the biologically-inspired Expert Object Recognition (EOR). This thesis adapts and extends the EOR approach for use with segmented video data. Properties of this data, such as segmentation masks and the visibility of an object over multiple frames, are exploited to decrease human supervision and increase accuracy. Several types of runtime learning are facilitated: class-level learning in which object types that are not included in the training set are given artifcial classes; viewpoint-level learning in which novel views of training objects are associated with existing classes; and instance-level learning of images that are somewhat similar to training images. The architecture of EOR, consisting of feature extraction, clustering, and cluster-specific principal component analysis, is retained. However, the K-means clustering algorithm used in EOR is replaced in this system by an augmented version of Fuzy K-means. This algorithm is incrementally run over the lifetime of the system, and automatically determines an appropriate number of partitions based on the data in memory and on a system parameter. In addition, the edge and line-based feature extraction of EOR is replaced with a global application of the principal component analysis, which increases accuracy when used with segmented video data. Classification output for the system consists of a multi-class hypothesis for each tracked object, from which a single-class "hard" hypothesis may be determined. The system, named VEOR (video expert object recognition), is designed for and tested with noisy, automatically segmented real-world data, consisting of both videos and still images of vehicle (car, pickup truck, and van) profiles

Projections from different visual areas of the cerebral cortex to the ipsilateral claustrum have been investigated in the cat by intracortical injections of [3H]proline. As a consequence of axonal flow an accumulation of radioactive material within a dorso-caudal region of the ipsilateral claustrum was found. The proline-transporting terminal arborizations of projections from areas 17, 18, 19 and the Clare-Bishop area overlap in the same claustral region. No point-to-point relation was observed between any one visual area and its claustral projection. Differences in the terminal distribution pattern were detectable according to the visual areas of origin.

Direct projections from visual areas 17, 18, 19, and lateral suprasylvian visual area (LS) to the striatum were searched for in 12 adult cats using the autoradiographic technique to detect neuronal pathways. Striatal labels were found only after injections in areas 19 and LS. Projections homolateral to the injection sites were observed from both areas to the head and body of the caudate nucleus and to the putamen. Contralateral projections were found from both areas 19 and LS: however, area 19 did not project to the contralateral putamen. The extent of contralateral projections was smaller and they were confined within the same regions as the homolateral ones. Silver grains were often arranged in cluster-like patches, which were more evident ipsilaterally, in the head of the caudate nucleus and after injections in area LS. The present data support the view of a not strictly topographical segregation of striatal projections from the cat visual cortex.

1. The projections from the superior temporal sulcus (STS) visual area to pontine nuclei were studied in the macaque monkey by the autoradiographic tracing method. Microinjections of a mixture of L-[5-3H] proline and L-[4,5-3H] leucine were carried out in the posterior band and the floor of the caudal half of the STS. Survival time was always 7 days. 2. The STS visual area projects to the dorsolateral part of ipsilateral pontine nuclei. Terminal projections are distributed in patches in the whole rostro-caudal extent of the pons. 3. These findings support the view that the STS visual area in the macaque monkey is homologous to the postero-medial lateral suprasylvian area in the cat.

Associated and commisural connections of visual areas 17 and 18 were studied in the light of the latest knowledge about the anatomo-functional organization of the cat's visual cortices. Injections of L-[5-3H] proline were placed, in different animals, in area 17 or 18 of the right hemisphere. Serial histological sections of the whole brain were processed by autoradiographic technique, after long (8-16 days) or short (30 hours) survival times. Cortical areas labelled by axonally transported radioactive material were then correlated with electrophysiological and cytoarchitectonical maps. Areas 17 and 18 were found to send associational projections to area 19, to the anterior and posterior subdivisions of the lateral suprasylvian visual area and to two regions lying on the crown of the suprasylvian gyrus (areas 21A and 20). In addition, area 18 sends projections to ipsilateral areas 17, 7 and 5. Contralaterally, projections from areas 17 and 18 reach the homonymous areas near the margin of the other. Homotopical callosal connections of area 17 affect the band of the contralateral 17 in which part of the ipsilateral visual hemi-field is represented. Heterotopical commisural projections from the areas studied affect areas 19 and LS in regions adjoining the central vertical meridian representations. These results suggest that, on an anatomical basis, the interactions between the cortical targets of the classical geniculo-striate and the estrageniculo-striate systems are more complicated than previously believed. Furthermore, the geniculo-striate system supplies some visual input to non-visual areas through area 18.

Injections of tritiated amino acids were made in the posterior bank and the fundus of the caudal third of the superior temporal sulcus (STs) of macaque monkeys. The injection sites lay mainly within the heavily myelinated region of STs, namely the middle temporal area. Labelled material was found in the surface of the caudal-most part of the prearcuate gyrus and in the anterior bank of the arcuate sulcus, that is in a restricted region of the part of the prefrontal cortex known as frontal eye field (FEF). The possibility that FEF may include several functional units receiving different visual inputs is considered.

Associational and commissural connections of area 19 and of the lateral suprasylvian visual area (LS) of the cat were studied by means of an axonal anterograde transport technique. Multiple injections of L-[5-3H] proline were placed, in different animals, in area 19 and in two different subdivisions of LS (PMLS, PLLS). Serial histological sections of the whole brains were processed by autoradiographic technique after long (8-12 days) and short (30 hrs) survival times. Cortical areas labelled by accumulation of radioactive material were correlated with electrophysiological and cytoarchitectonic maps of cortical areas. Results showed a number of projections of areas 19 and LS not previously described. Area 19 sends ipsilateral projections to area 5 and to the medial bank of the anterior suprasylvian sulcus. The postero-medial (PMLS) and postero-lateral (PLLS) subdivisions of LS project to ipsilateral areas 5, 6, 7, 20, 21, 35, to the suprasylvian fringe (SF), to the cingulate gyrus and to the insular cortex. Contralaterally, new projections are found from area 19 to areas 20 and 5, and from LS to areas 19, 20, 21, 5, 7, SF to the cingulate gyrus and to the insular cortex. Results lead to conclude that the intra-modal and cross-modal interactions between cortical areas processing visual input are more complicated than previously assumed. The crucial role of the cat's LS in the cortico-cortical distribution of the visual message in emphasized.

BACKGROUND: Multiple Sclerosis (MS) is an immune-mediated disease of the Central Nervous System with two major underlying etiopathogenic processes: inflammation and neurodegeneration. The latter determines the prognosis of this disease. MS is the main cause of non-traumatic disability in middle-aged populations. FINDINGS: The MS-VisualPath Cohort was set up to study the neurodegenerative component of MS using advanced imaging techniques by focusing on analysis of the visual pathway in a middle-aged MS population in Barcelona, Spain. We started the recruitment of patients in the early phase of MS in 2010 and it remains permanently open. All patients undergo a complete neurological and ophthalmological examination including measurements of physical and disability (Expanded Disability Status Scale; Multiple Sclerosis Functional Composite and neuropsychological tests), disease activity (relapses) and visual function testing (visual acuity, color vision and visual field). The MS-VisualPath protocol also assesses the presence of anxiety and depressive symptoms (Hospital Anxiety and Depression Scale), general quality of life (SF-36) and visual quality of life (25-Item National Eye Institute Visual Function Questionnaire with the 10-Item Neuro-Ophthalmic Supplement). In addition, the imaging protocol includes both retinal (Optical Coherence Tomography and Wide-Field Fundus Imaging) and brain imaging (Magnetic Resonance Imaging). Finally, multifocal Visual Evoked Potentials are used to perform neurophysiological assessment of the visual pathway. DISCUSSION: The analysis of the visual pathway with advance imaging and electrophysilogical tools in parallel with clinical information will provide significant and new knowledge regarding neurodegeneration in MS and provide new clinical and imaging biomarkers to help monitor disease progression in these patients.

平成27年度

Solutions of tritiated aminoacids were injected into the visual cortical region hidden in the depth of the superior temporal sulcus (STs) of the macaque monkey. Injection sites involved the middle temporal visual area, extending into the surrounding visual cortices. Projections were found homolaterally in both the striatum and claustrum. In the caudate nucleus labeled material affected mainly the body and spread both to the tail and the head of the nucleus. Label was also seen in the caudal third of the putamen and in the postero-ventral claustrum. Compared with the scarcity of afferents arising from occipital visual areas, the present data point to a heavy projection system from additional visual areas of the STs to the basal ganglia in the macaque monkey.

Cortical projections from the visual region and adjacent polysensory region of the superior temporal sulcus (STs) to the lateral geniculate body (LGb) were investigated in the macaque monkey using an autoradiographic tracing method. Solutions of tritiated aminoacids were injected into different parts of the caudal half of the STs of five animals. A survival time of 7 days was allowed. Labels were found in both subdivisions of the LGb: the dorsal lateral geniculate nucleus (DLGn) and the pregeniculate nucleus (PGn). In particular, part of the visual cortical region adjacent to the middle temporal area (MT) projects into the DLGn as well as the PGn, whereas the MT itself and the superior temporal polysensory region project into the PGn only. Afferents to the DLGn terminate in the magnocellular layers and in their adjoining interlaminar zones, completely sparing the parvocellular layers. Afferents to the PGn terminate in separate regions of this nucleus; the MT and adjacent visual cortices project into the internal layer of the PGn, whereas the polysensosy region of the STs projects into the external retinorecipient layer of the PGn. Possible functional implications of these projections are discussed.

1. The projections from cat cortical visuals areas to the contralateral superior colliculus (SC) were studied by the autoradiographical tracing method. Microinjections of L- [5-3H] proline were carried out in cortical visual areas 17, 18, 19 and the lateral suprasylvian visual area (LS) in different cats. Only one cortical area was injected in each animal. Survival times of 30 hours or 8 days were allowed. 2. Areas 17, 18, 19 and LS send projections to laminae, I, II and III of the frontal pole of the contralateral SC. Areas 19 and LS project also the contralateral pretectal nuclei, mainly to the posterior pretectal nucleus. 3. Cortical fibres reaching the contralateral SC pass through the brachium of the ipsilateral SC. They run along the caudal part of this structure, cross the midline and then run along the caudal part of the contralateral SC; finally, they turn anterolaterally and reach the rostal part of SC. This pathway is the tectal semidecussation (5). Cortical fibres from LS (in particular from PLLS; 15) reach the contralateral SC also via the commissure of SC. 4. Our data support the suggestion (3) that the frontal poles of SC, in the cat, may subserve straight-ahead attention and orientation to visual stimuli.

[ANGLÈS] The optic tract is a complex structure, not only because of its morphology, but also because of its location in the brain. This project aims to obtain the optic tract trajectories, by using diffusion MRI techniques, in an automated and reliable way, and extract diffusion characteristic parameters. [CASTELLÀ] El tracto óptico es una estructura compleja, no sólo por su morfología, sino también por su situación dentro del cerebro. Este proyecto pretende obtener la trayectoria de la vía visual con herramientas de difusión de manera automatizada y fiable, y extraer parámetros característicos asociados a MRI a lo largo de ésta. [CATALÀ] El tracte òptic és una estructura complexa, no només per la seva morfologia, sinó també per la seva situació dins del cervell. Aquest projecte pretén obtenir la trajectòria de la via visual amb eines de difusió de manera automatitzada i fiable, i extreure paràmetres característics associats a MRI al llarg d'aquesta.