scholarly journals Pixel-Aware Deep Function-Mixture Network for Spectral Super-Resolution

2020 ◽  
Vol 34 (07) ◽  
pp. 12821-12828 ◽  
Author(s):  
Lei Zhang ◽  
Zhiqiang Lang ◽  
Peng Wang ◽  
Wei Wei ◽  
Shengcai Liao ◽  
...  
Keyword(s):  
Receptive Field ◽  
Hyperspectral Image ◽  
Receptive Fields ◽  
Super Resolution ◽  
Mapping Function ◽  
Basis Functions ◽  
Field Size ◽  
Receptive Field Size ◽  
New Class ◽  
Rgb Image

Spectral super-resolution (SSR) aims at generating a hyperspectral image (HSI) from a given RGB image. Recently, a promising direction is to learn a complicated mapping function from the RGB image to the HSI counterpart using a deep convolutional neural network. This essentially involves mapping the RGB context within a size-specific receptive field centered at each pixel to its spectrum in the HSI. The focus thereon is to appropriately determine the receptive field size and establish the mapping function from RGB context to the corresponding spectrum. Due to their differences in category or spatial position, pixels in HSIs often require different-sized receptive fields and distinct mapping functions. However, few efforts have been invested to explicitly exploit this prior.To address this problem, we propose a pixel-aware deep function-mixture network for SSR, which is composed of a new class of modules, termed function-mixture (FM) blocks. Each FM block is equipped with some basis functions, i.e., parallel subnets of different-sized receptive fields. Besides, it incorporates an extra subnet as a mixing function to generate pixel-wise weights, and then linearly mixes the outputs of all basis functions with those generated weights. This enables us to pixel-wisely determine the receptive field size and the mapping function. Moreover, we stack several such FM blocks to further increase the flexibility of the network in learning the pixel-wise mapping. To encourage feature reuse, intermediate features generated by the FM blocks are fused in late stage, which proves to be effective for boosting the SSR performance. Experimental results on three benchmark HSI datasets demonstrate the superiority of the proposed method.

Receptive Field Size and Response Latency Are Correlated Within the Cat Visual Thalamus

2005 ◽  
Vol 93 (6) ◽  
pp. 3537-3547 ◽  
Author(s):  
Chong Weng ◽  
Chun-I Yeh ◽  
Carl R. Stoelzel ◽  
Jose-Manuel Alonso
Keyword(s):  
Receptive Field ◽  
Spatial Frequency ◽  
Response Latency ◽  
Time Course ◽  
Frequency Tuning ◽  
Receptive Fields ◽  
Cortical Cell ◽  
Field Size ◽  
Receptive Field Size ◽  
Spatial Frequency Tuning

Each point in visual space is encoded at the level of the thalamus by a group of neighboring cells with overlapping receptive fields. Here we show that the receptive fields of these cells differ in size and response latency but not at random. We have found that in the cat lateral geniculate nucleus (LGN) the receptive field size and response latency of neighboring neurons are significantly correlated: the larger the receptive field, the faster the response to visual stimuli. This correlation is widespread in LGN. It is found in groups of cells belonging to the same type (e.g., Y cells), and of different types (i.e., X and Y), within a specific layer or across different layers. These results indicate that the inputs from the multiple geniculate afferents that converge onto a cortical cell (approximately 30) are likely to arrive in a sequence determined by the receptive field size of the geniculate afferents. Recent studies have shown that the peak of the spatial frequency tuning of a cortical cell shifts toward higher frequencies as the response progresses in time. Our results are consistent with the idea that these shifts in spatial frequency tuning arise from differences in the response time course of the thalamic inputs.

scholarly journals Reversible deactivation of higher-order posterior parietal areas. I. Alterations of receptive field characteristics in early stages of neocortical processing

2014 ◽  
Vol 112 (10) ◽  
pp. 2529-2544 ◽  
Author(s):  
Dylan F. Cooke ◽  
Adam B. Goldring ◽  
Mary K. L. Baldwin ◽  
Gregg H. Recanzone ◽  
Arnold Chen ◽  
...  
Keyword(s):  
Receptive Field ◽  
Premotor Cortex ◽  
Receptive Fields ◽  
Macaque Monkey ◽  
Field Size ◽  
Somatosensory Processing ◽  
Receptive Field Size ◽  
Area 5 ◽  
Posterior Parietal ◽  
Area 7B

Somatosensory processing in the anesthetized macaque monkey was examined by reversibly deactivating posterior parietal areas 5L and 7b and motor/premotor cortex (M1/PM) with microfluidic thermal regulators developed by our laboratories. We examined changes in receptive field size and configuration for neurons in areas 1 and 2 that occurred during and after cooling deactivation. Together the deactivated fields and areas 1 and 2 form part of a network for reaching and grasping in human and nonhuman primates. Cooling area 7b had a dramatic effect on receptive field size for neurons in areas 1 and 2, while cooling area 5 had moderate effects and cooling M1/PM had little effect. Specifically, cooling discrete locations in 7b resulted in expansions of the receptive fields for neurons in areas 1 and 2 that were greater in magnitude and occurred in a higher proportion of sites than similar changes evoked by cooling the other fields. At some sites, the neural receptive field returned to the precooling configuration within 5–22 min of rewarming, but at other sites changes in receptive fields persisted. These results indicate that there are profound top-down influences on sensory processing of early cortical areas in the somatosensory cortex.

Synthesis of Multiwhisker-Receptive Fields in Subcortical Stations of the Vibrissa System

2004 ◽  
Vol 91 (4) ◽  
pp. 1510-1515 ◽  
Author(s):  
Elena Timofeeva ◽  
Philippe Lavallée ◽  
Dominique Arsenault ◽  
Martin Deschênes
Keyword(s):  
Receptive Field ◽  
Brain Stem ◽  
Receptive Fields ◽  
Field Size ◽  
Electrolytic Lesion ◽  
Evoked Responses ◽  
Similar Reduction ◽  
Receptive Field Size ◽  
Brain Stem Lesion ◽  
Before And After

This study addresses the origins of multiwhisker-receptive fields of neurons in the thalamic ventral posterior medial (VPM) nucleus of the rat. We sought to determine whether multiwhisker-receptive field synthesis occurs in VPM through convergent projections from the principalis (PrV) and interpolaris (SpVi) nuclei, or in PrV by intersubnuclear projections from the spinal trigeminal complex. We tested these hypotheses by recording whisker-evoked responses in PrV and VPM before and after electrolytic lesion of the SpVi in lightly anesthetized rats. Before the lesion PrV cells responded, on average, to 3.2 ± 1.2 whiskers but responsiveness was reduced to 1.07 ± 0.31 whisker after the lesion. A similar reduction of receptive field size was observed in VPM, where neurons responded, on average, to 2.94 ± 0.95 whiskers before the lesion and to 1.05 ± 0.22 whisker after the lesion. Thus one can conclude that intersubnuclear projections mediate surround whisker-receptive fields in PrV, and therefore in VPM. However, it has previously been shown that parasagittal brain stem transection, which severed ascending projections from SpVi, but left intersubnuclear connections intact, rendered VPM cells monowhisker responsive. We wondered whether midline brain stem lesion modified receptive field properties in SpVi. In normal rats SpVi cells responded, on average, to 7.52 ± 4.25 whiskers, but responsiveness was dramatically reduced to 1.47 ± 1.07 whisker after the lesion. Together these results indicate that the synthesis of surround receptive fields in subcortical stations relies almost exclusively on intersubnuclear projections from the spinal trigeminal complex to the PrV.

Control of Size and Excitability of Mechanosensory Receptive Fields in Dorsal Column Nuclei by Homolateral Dorsal Horn Neurons

1998 ◽  
Vol 80 (1) ◽  
pp. 120-129 ◽  
Author(s):  
Robert W. Dykes ◽  
A. D. Craig
Keyword(s):  
Spinal Cord ◽  
Receptive Field ◽  
Dorsal Horn ◽  
Cobalt Chloride ◽  
Receptive Fields ◽  
Dorsal Column ◽  
Field Size ◽  
Dorsal Column Nuclei ◽  
Receptive Field Size ◽  
Dorsal Horn Neurons

Dykes, Robert W. and A. D. Craig. Control of size and excitability of mechanosensory receptive fields in dorsal column nuclei by homolateral dorsal horn neurons. J. Neurophysiol. 80: 120–129 1998. Both accidental and experimental lesions of the spinal cord suggest that neuronal processes occurring in the spinal cord modify the relay of information through the dorsal column-lemniscal pathway. How such interactions might occur has not been adequately explained. To address this issue, the receptive fields of mechanosensory neurons of the dorsal column nuclei were studied before and after manipulation of the spinal dorsal horn. After either a cervical or lumbar laminectomy and exposure of the dorsal column nuclei in anesthetized cats, the representation of the hindlimb or of the forelimb was defined by multiunit recordings in both the dorsal column nuclei and in the ipsilateral spinal cord. Next, a single cell was isolated in the dorsal column nuclei, and its receptive field carefully defined. Each cell could be activated by light mechanical stimuli from a well-defined cutaneous receptive field. Generally the adequate stimulus was movement of a few hairs or rapid skin indentation. Subsequently a pipette containing either lidocaine or cobalt chloride was lowered into the ipsilateral dorsal horn at the site in the somatosensory representation in the spinal cord corresponding to the receptive field of the neuron isolated in the dorsal column nuclei. Injection of several hundred nanoliters of either lidocaine or cobalt chloride into the dorsal horn produced an enlargement of the receptive field of the neuron being studied in the dorsal column nuclei. The experiment was repeated 16 times, and receptive field enlargements of 147–563% were observed in 15 cases. These data suggest that the dorsal horn exerts a tonic inhibitory control on the mechanosensory signals relayed through the dorsal column-lemniscal pathway. Because published data from other laboratories have shown that receptive field size is controlled by signals arising from the skin, we infer that the control of neuronal excitability, receptive field size and location for lemniscal neurons is determined by tonic afferent activity that is relayed through a synapse in the dorsal horn. This influence of dorsal horn neurons on the relay of mechanosensory information through the lemniscal pathways must modify our traditional views concerning the relative independence of these two systems.

scholarly journals Receptive field size, chemical and thermal responses, and fiber conduction velocity of rat chorda tympani geniculate ganglion neurons

2016 ◽  
Vol 115 (6) ◽  
pp. 3062-3072 ◽  
Author(s):  
Yusuke Yokota ◽  
Robert M. Bradley
Keyword(s):  
Electrical Stimulation ◽  
Receptive Field ◽  
Conduction Velocity ◽  
Receptive Fields ◽  
Field Size ◽  
Geniculate Ganglion ◽  
Chorda Tympani ◽  
Fungiform Papillae ◽  
Receptive Field Size ◽  
Chemical Stimuli

Afferent chorda tympani (CT) fibers innervating taste and somatosensory receptors in fungiform papillae have neuron cell bodies in the geniculate ganglion (GG). The GG/CT fibers branch in the tongue to innervate taste buds in several fungiform papillae. To investigate receptive field characteristics of GG/CT neurons, we recorded extracellular responses from GG cells to application of chemical and thermal stimuli. Receptive field size was mapped by electrical stimulation of individual fungiform papillae. Response latency to electrical stimulation was used to determine fiber conduction velocity. Responses of GG neurons to lingual application of stimuli representing four taste qualities, and water at 4°C, were used to classify neuron response properties. Neurons classified as SALT, responding only to NaCl and NH4Cl, had a mean receptive field size of six papillae. Neurons classified as OTHER responded to salts and other chemical stimuli and had smaller mean receptive fields of four papillae. Neurons that responded to salts and cold stimuli, classified as SALT/THERMAL, and neurons responding to salts, other chemical stimuli and cold, classified as OTHER/THERMAL, had mean receptive field sizes of six and five papillae, respectively. Neurons responding only to cold stimuli, categorized as THERMAL, had receptive fields of one to two papillae located at the tongue tip. Based on conduction velocity most of the neurons were classified as C fibers. Neurons with large receptive fields had higher conduction velocities than neurons with small receptive fields. These results demonstrate that GG neurons can be distinguished by receptive field size, response properties and afferent fiber conduction velocity derived from convergent input of multiple taste organs.

Effects of light stimuli on the release of dopamine from interplexiform cells in the white perch retina

1991 ◽  
Vol 7 (5) ◽  
pp. 451-458 ◽  
Author(s):  
Osamu Umino ◽  
Yunhee Lee ◽  
John E. Dowling
Keyword(s):  
Receptive Field ◽  
White Perch ◽  
Horizontal Cell ◽  
Receptive Fields ◽  
Plexiform Layer ◽  
Horizontal Cells ◽  
Field Size ◽  
Dopamine Antagonists ◽  
Flickering Light ◽  
Receptive Field Size

AbstractInterplexiform cells are centrifugal neurons in the retina carrying information from the inner to the outer plexiform layers. In teleost fish, interplexiform cells appear to release dopamine in the outer plexiform layer after prolonged darkness that modulates the receptive-field size and light responsiveness of horizontal cells (Mangel & Dowling, 1985; Yang et al., 1988a, b). It has been proposed that interplexiform cells may also release dopamine upon steady illumination because horizontal cells' receptive fields shrink in the light (Shigematsu & Yamada, 1988). Here, we report the shrinkage of the receptive fields of horizontal cells seen in the presence of background illumination is not blocked by dopamine antagonists, indicating that dopamine does not underlie the receptive-field size changes observed during steady illumination. Flickering light, however, does appear to stimulate the release of dopamine from the interplexiform cells, resulting in a marked reduction of horizontal cell receptive-field size. Taken together, experiments on horizontal cells indicate that dopamine is released from interplexiform cells in the teleost retina after prolonged darkness and during flickering light, but that dopamine release from interplexiform cells during steady retinal illumination is minimal.

Effect of spike blockade on the receptive-field size of amacrine and ganglion cells in the rabbit retina

1996 ◽  
Vol 75 (5) ◽  
pp. 1878-1893 ◽  
Author(s):  
S. A. Bloomfield
Keyword(s):  
Receptive Field ◽  
Sodium Channels ◽  
Ganglion Cells ◽  
Receptive Fields ◽  
Amacrine Cells ◽  
Field Size ◽  
Receptive Field Size ◽  
Voltage Gated ◽  
Voltage Gated Sodium Channels ◽  
Dendritic Arbors

1. Intracellular recordings were obtained from 21 amacrine cells and 12 ganglion cells in the isolated, superfused retina-eyecup of the rabbit. Cells were subsequently labeled with horseradish peroxidase (HRP) or N-(2-aminoethyl)-biotinamide hydrochloride (Neurobiotin) for morphologic identification. 2. Initial experiments performed on three amacrine cells and three ganglion cells showed that 1 microM tetrodotoxin (TTX) abolished all spiking. This included both large-amplitude and small-amplitude spikes recorded in many amacrine cells, indicating that they are mediated by voltage-gated sodium channels. 3. The center-receptive-field size of 18 amacrine cells and 9 ganglion cells was measured with the use of a 50-microns-wide/6.0-mm-long rectangular slit of light that was displaced along its minor axis (parallel to the visual streak) in steps as small as 3 microns. The retina was then bathed in 1 microM TTX, or individual cells were injected with 50 mM QX-314, a quatemary lidocaine derivative, to abolish all spiking, and the center-receptive field of each cell was then remeasured. 4. Although TTX blocked spiking in all ganglion cells (dendritic diameters ranging from 302 to 969 microns), it produced no significant change in the size of their center-receptive fields. This finding argues that passive, electrotonic spread of synaptic inputs to ganglion cell dendritic arbors is adequate for efficient propagation from terminal branches to the soma; active propagation via voltage-gated sodium channels plays no apparent role. 5. In contrast, TTX and QX-314 had variable effect on the receptive fields of amacrine cells, which was related to the size of their dendritic arbors. Whereas TTX had no significant effect on the receptive-field size of amacrine cells whose dendritic arbors were < 525 microns across, the center-receptive fields of larger amacrine cells were reduced, on average, by 40%; QX-314 produced a very similar average reduction of 39%. Moreover, for these larger cells, there was a direct relationship between the magnitude of the reduction in receptive-field size produced by TTX or QX-314 and the size of a cell's dendritic arbor. This relationship was true whether the change in receptive-field size was measured in absolute terms or as percent reduction from control values. 6. Interestingly, TTX and QX-314 also significantly reduced the amplitude of slow potentials recorded in amacrine cells by an average of 22 and 24%, respectively. However, the amplitude of slow potentials recorded in ganglion cells were relatively uneffected by TTX. 7. These findings are consistent with the idea that, for amacrine cells with dendritic arbors spanning > 525 microns, active propagation of synaptic signals mediated by voltage-gated sodium channels is necessary for efficient movement of information across a cell's dendritic arbor and thus plays a major role in shaping their receptive fields. Although the TTX effects may also reflect an indirect contribution from altered synaptic input derived from presynaptic spiking neurons, the strong similarity between the effects of TTX and QX-314 argues that any such contribution was minor. For smaller amacrine cells, passive, electrotonic spread of signals appears adequate for efficient propagation within their limited dendritic arbors.

The light-induced reduction of horizontal cell receptive field size in the goldfish retina involves nitric oxide

2011 ◽  
Vol 28 (2) ◽  
pp. 137-144 ◽  
Author(s):  
BRYAN A. DANIELS ◽  
WILLIAM H. BALDRIDGE
Keyword(s):  
Nitric Oxide ◽  
Receptive Field ◽  
Horizontal Cell ◽  
Receptive Fields ◽  
Horizontal Cells ◽  
Field Size ◽  
Nitric Oxide Donor ◽  
Nitric Oxide Synthase Inhibitor ◽  
Receptive Field Size ◽  
Goldfish Retina

AbstractHorizontal cells of the vertebrate retina have large receptive fields as a result of extensive gap junction coupling. Increased ambient illumination reduces horizontal cell receptive field size. Using the isolated goldfish retina, we have assessed the contribution of nitric oxide to the light-dependent reduction of horizontal cell receptive field size. Horizontal cell receptive field size was assessed by comparing the responses to centered spot and annulus stimuli and from the responses to translated slit stimuli. A period of steady illumination decreased the receptive field size of horizontal cells, as did treatment with the nitric oxide donor (Z)-1-[N-(2-aminoethyl)-N-(2-ammonioethyl)amino]diazen-1-ium-1,2-diolate (100μM). Blocking the endogenous production of nitric oxide with the nitric oxide synthase inhibitor, NG-nitro-l-arginine methyl ester (1 mM), decreased the light-induced reduction of horizontal cell receptive field size. These findings suggest that nitric oxide is involved in light-induced reduction of horizontal cell receptive field size.

Functional role of GABA in cat primary somatosensory cortex: shaping receptive fields of cortical neurons

1984 ◽  
Vol 52 (6) ◽  
pp. 1066-1093 ◽  
Author(s):  
R. W. Dykes ◽  
P. Landry ◽  
R. Metherate ◽  
T. P. Hicks
Keyword(s):  
Receptive Field ◽  
Somatosensory Cortex ◽  
Cortical Neurons ◽  
Receptive Fields ◽  
Field Size ◽  
Primary Somatosensory Cortex ◽  
Gamma Aminobutyric Acid ◽  
Receptive Field Size ◽  
Thalamic Stimulation ◽  
Somatosensory Neurons

Extracellular recordings of 209 neurons were obtained with carbon fiber-containing multibarrel micropipettes. The cells were isolated in the primary somatosensory cortex of cats anesthetized with barbiturate and classified according to the nature of their response to natural stimuli, the nature of the surrounding multiunit responses to the same stimuli, the response to thalamic stimulation, and their depth in the cortex. To study factors controlling the excitability of somatosensory neurons, their receptive fields were examined in the presence of iontophoretically administered gamma-aminobutyric acid (GABA), glutamate, and bicuculline methiodide (BMI). Even when the neurons were depolarized to perithreshold levels with glutamate, or when local inhibitory influences mediated by GABA were antagonized by BMI, the apparent specificity for one class of afferent input was maintained. Neurons responding to stimulation of either cutaneous or deep receptors maintained their modality specificity, and neurons in cutaneous rapidly adapting regions never took on slowly adapting properties. When ejected at currents that did not elicit action potentials, glutamate lowered the threshold for activation by cutaneous stimuli but did not enlarge the receptive field. With larger ejecting currents, the neurons developed an on-going discharge, but even at these higher doses, glutamate did not produce an increase in the receptive-field size. Some neurons in regions of cortex exhibiting slowly adapting multiunit responses were relatively insensitive to glutamate. These cells required four to five times more glutamate to evoke discharges than did most neurons. Other cells, previously unresponsive to somatic stimuli, could be shown to possess distinct cutaneous receptive fields when either glutamate or BMI was ejected in their vicinity. Iontophoretically administered BMI altered the firing pattern of somatosensory neurons, causing them to discharge in bursts of 3-15 impulses. BMI enlarged the receptive-field size of neurons in regions displaying rapidly adapting multiunit background discharges but not in those regions with slowly adapting multiunit discharges. This differential effect of BMI, suggesting that GABA controls receptive-field size in rapidly adapting regions, also indicates that neurons in rapidly adapting regions differ pharmacologically from those in other submodality regions. In all cortical regions, BMI blocked the poststimulus inhibitory period that normally followed thalamic stimulation.(ABSTRACT TRUNCATED AT 400 WORDS)

Sign in / Sign up

Export Citation Format

Share Document