Transfer Of Spatial Information Research Articles

Laplacian-guided hierarchical transformer: A network for medical image segmentation

Background and Objective:Accurate medical image segmentation is crucial for diagnosis and treatment planning, particularly in tumor localization and organ measurement. Despite the success of Transformer models in various domains, they still struggle to capture high-frequency features, limiting their performance in medical image segmentation, especially in edge texture extraction. To overcome this limitation and improve segmentation accuracy, this study proposes a novel model architecture aimed at enhancing the Transformer’s ability to capture and integrate both high-frequency and low-frequency features. Methods:Our model combines the extraction of high-frequency features using a Laplacian pyramid with the capture of low-frequency features through a Local-Global Feature Aggregation Module. A Feature Interaction Fusion module is employed to integrate these features, focusing on target areas. Additionally, a new bridging module facilitates the transfer of spatial information between the encoder and decoder via layer-wise attention mechanisms. The model’s performance was evaluated using the Synapse dataset with statistical measures such as the Dice Similarity Coefficient and Hausdorff Distance. The code is available at https://github.com/chenyuxiao123/LGHF. Results:The proposed model demonstrated state-of-the-art performance in 2D medical image segmentation, achieving a Dice Similarity Coefficient of 84.10% and a Hausdorff Distance of 12.78. The evaluation metrics indicate significant improvements compared to existing methods. Conclusion:This novel model architecture, with its enhanced capability to capture and integrate both high-frequency and low-frequency features, shows significant potential for advancing medical image segmentation. The results on the Synapse dataset demonstrate its effectiveness and suggest its application could improve diagnosis and treatment planning in clinical settings.

ReconU-Net: a direct PET image reconstruction using U-Net architecture with back projection-induced skip connection

Objective. This study aims to introduce a novel back projection-induced U-Net-shaped architecture, called ReconU-Net, based on the original U-Net architecture for deep learning-based direct positron emission tomography (PET) image reconstruction. Additionally, our objective is to visualize the behavior of direct PET image reconstruction by comparing the proposed ReconU-Net architecture with the original U-Net architecture and existing DeepPET encoder–decoder architecture without skip connections. Approach. The proposed ReconU-Net architecture uniquely integrates the physical model of the back projection operation into the skip connection. This distinctive feature facilitates the effective transfer of intrinsic spatial information from the input sinogram to the reconstructed image via an embedded physical model. The proposed ReconU-Net was trained using Monte Carlo simulation data from the Brainweb phantom and tested on both simulated and real Hoffman brain phantom data. Main results. The proposed ReconU-Net method provided better reconstructed image in terms of the peak signal-to-noise ratio and contrast recovery coefficient than the original U-Net and DeepPET methods. Further analysis shows that the proposed ReconU-Net architecture has the ability to transfer features of multiple resolutions, especially non-abstract high-resolution information, through skip connections. Unlike the U-Net and DeepPET methods, the proposed ReconU-Net successfully reconstructed the real Hoffman brain phantom, despite limited training on simulated data. Significance. The proposed ReconU-Net can improve the fidelity of direct PET image reconstruction, even with small training datasets, by leveraging the synergistic relationship between data-driven modeling and the physics model of the imaging process.

Mapmixture: An R package and web app for spatial visualisation of admixture and population structure.

The mapmixture R package and interactive web app are tools to aid visualisation of admixture and population structure in geographic space. The purpose of mapmixture is to enable and encourage molecular ecologists, and in particular population geneticists and phylogeneticists, to plot their admixture, ancestry or assignment results on a map when location information is available. mapmixture accepts data in the format typically generated by admixture analyses and visualises proportions to each genetic cluster per site as pie charts on a projected (optional) map. Combining this site-based map presentation approach with the routine individual-based presentation of admixture (structure) barplots will enhance interpretation of genetic-geographic patterns. Additionally, in the context of science communication, this enables clearer transfer of spatial genetic information to readers or listeners, and especially to audiences that do not have a background in genetics but who are able to use the genetic information as evidence in conservation management. The latest version of mapmixture is available on GitHub (https://github.com/tom-jenkins/mapmixture), which details installation instructions and examples of how to use the package and interactive web app.

The role of spatial information in an approximate cross-modal number matching task

The approximate number system (ANS) is thought to be an innate cognitive system that allows humans to perceive numbers (>4) in a fuzzy manner. One assumption of the ANS is that numerosity is represented amodally due to a mechanism, which filters out nonnumerical information from stimulus material. However, some studies show that nonnumerical information (e.g., spatial parameters) influence the numerosity percept as well. Here, we investigated whether there is a cross-modal transfer of spatial information between the haptic and visual modality in an approximate cross-modal number matching task. We presented different arrays of dowels (haptic stimuli) to 50 undergraduates and asked them to compare haptically perceived numerosity to two visually presented dot arrays. Participants chose which visually presented array matched the numerosity of the haptic stimulus. The distractor varied in number and displayed a random pattern, whereas the matching (target) dot array was either spatially identical or spatially randomized (to the haptic stimulus). We hypothesized that if a “numerosity” percept is based solely on number, neither spatially identical nor spatial congruence between the haptic and the visual target arrays would affect the accuracy in the task. However, results show significant processing advantages for targets with spatially identical patterns and, furthermore, that spatial congruency between haptic source and visual target facilitates performance. Our results show that spatial information was extracted from the haptic stimuli and influenced participants’ responses, which challenges the assumption that numerosity is represented in a truly abstract manner by filtering out any other stimulus features.

Efficient transfer of spatial intensity and phase information of arbitrary modes via four-wave mixing in an atomic vapor

We propose a scheme to transfer the spatial intensity as well as phase information encoded initially in the spatial profile of a weak probe field onto a newly generated Stokes field in a nonlinear process of four-wave mixing (FWM). The FWM process is explored in a gaseous medium consisting of atoms modeled as a three-level system in the $\mathrm{\ensuremath{\Lambda}}$ configuration. We found that different orders of Hermite-Gaussian and Laguerre-Gaussian modes carried by a probe beam are effectively transferred to a Stokes beam. Interestingly, the transferred intensity of the Stokes beam is a clone of the probe beam, whereas the phase profile is conjugate to the probe. The phase distribution of the transmitted modes at the exit of the medium is explored by superimposing them on a copropagating plane wave. Moreover, the parametric amplification due to the nonlinear process of FWM prevents any information loss due to linear absorption and permits us to work at high optical depths. We found a structural similarity between transferred images of about $99%$, which provides clear evidence for the successful transfer of information in the FWM process.

Global and Local Feature Reconstruction for Medical Image Segmentation.

Learning how to capture long-range dependencies and restore spatial information of down-sampled feature maps are the basis of the encoder-decoder structure networks in medical image segmentation. U-Net based methods use feature fusion to alleviate these two problems, but the global feature extraction ability and spatial information recovery ability of U-Net are still insufficient. In this paper, we propose a Global Feature Reconstruction (GFR) module to efficiently capture global context features and a Local Feature Reconstruction (LFR) module to dynamically up-sample features, respectively. For the GFR module, we first extract the global features with category representation from the feature map, then use the different level global features to reconstruct features at each location. The GFR module establishes a connection for each pair of feature elements in the entire space from a global perspective and transfers semantic information from the deep layers to the shallow layers. For the LFR module, we use low-level feature maps to guide the up-sampling process of high-level feature maps. Specifically, we use local neighborhoods to reconstruct features to achieve the transfer of spatial information. Based on the encoder-decoder architecture, we propose a Global and Local Feature Reconstruction Network (GLFRNet), in which the GFR modules are applied as skip connections and the LFR modules constitute the decoder path. The proposed GLFRNet is applied to four different medical image segmentation tasks and achieves state-of-the-art performance.

First spike latency of ON/OFF neurons in the optic tectum of pigeons.

Some shallow and middle optic tectum (OT) neurons have stable, asymmetric full-screen ON and OFF stimulus response properties, which makes them candidates for delay encoding. In this paper, we investigated the delay encoding mechanism for the neuronal clusters in the OT region of pigeons and determined the mechanism of delay coding in the OT region. By analyzing the responses of the neuron cluster under full-screen switch-on and switch-off stimulation, we found that the delay coding was widespread in the OT region where the ON/OFF stimulation time difference was 4-6 ms. Information theory analysis under grating stimulation and experiments based on single-neuron character reconstruction of neurons showed that OT neuron clusters use the first spike latency (FSL) for the rapid transfer of spatial structure information. Furthermore, 4 models were used to predict the first spike latency of these OT neurons. The best simulation results were obtained using an architecture where the ON and OFF paths of multiple retinal ganglion cells (RGCs) were integrated and summed, respectively.

Transfer of Spatial Contact Information Among Limbs and the Notion of Peripersonal Space in Insects.

Internal representation of far-range space in insects is well established, as it is necessary for navigation behavior. Although it is likely that insects also have an internal representation of near-range space, the behavioral evidence for the latter is much less evident. Here, we estimate the size and shape of the spatial equivalent of a near-range representation that is constituted by somatosensory sampling events. To do so, we use a large set of experimental whole-body motion capture data on unrestrained walking, climbing and searching behavior in stick insects of the species Carausius morosus to delineate ‘action volumes’ and ‘contact volumes’ for both antennae and all six legs. As these volumes are derived from recorded sampling events, they comprise a volume equivalent to a representation of coinciding somatosensory and motor activity. Accordingly, we define this volume as the peripersonal space of an insect. It is of immediate behavioral relevance, because it comprises all potential external object locations within the action range of the body. In a next step, we introduce the notion of an affordance space as that part of peripersonal space within which contact-induced spatial estimates lie within the action ranges of more than one limb. Because the action volumes of limbs overlap in this affordance space, spatial information from one limb can be used to control the movement of another limb. Thus, it gives rise to an affordance as known for contact-induced reaching movements and spatial coordination of footfall patterns in stick insects. Finally, we probe the computational properties of the experimentally derived affordance space for pairs of neighboring legs. This is done by use of artificial neural networks that map the posture of one leg into a target posture of another leg with identical foot position.

Enhanced phase contrast transfer using ptychography combined with a pre-specimen phase plate in a scanning transmission electron microscope

The ability to image light elements in both crystalline and noncrystalline materials at near atomic resolution with an enhanced contrast is highly advantageous to understand the structure and properties of a wide range of beam sensitive materials including biological specimens and molecular hetero-structures. This requires the imaging system to have an efficient phase contrast transfer at both low and high spatial frequencies. In this work we introduce a new phase contrast imaging method in a scanning transmission electron microscope (STEM) using a pre-specimen phase plate in the probe forming aperture, combined with a fast pixelated detector to record diffraction patterns at every probe position, and phase reconstruction using ptychography. The phase plate significantly enhances the contrast transfer of low spatial frequency information, and ptychography maximizes the extraction of the phase information at all spatial frequencies. In addition, the STEM probe with the presence of the phase plate retains its atomic resolution, allowing simultaneous incoherent Z-contrast imaging to be obtained along with the ptychographic phase image. An experimental image of Au nanoparticles on a carbon support shows high contrast for both materials. Multislice image simulations of a DNA molecule shows the capability of imaging soft matter at low dose conditions, which implies potential applications of low dose imaging of a wide range of beam sensitive materials.

From Capture to Inhibition: How does Irrelevant Information Influence Visual Search? Evidence from a Spatial Cuing Paradigm.

Even though information is spatially and temporally irrelevant, it can influence the processing of subsequent information. The present study used a spatial cuing paradigm to investigate the origins of this persisting influence by means of event-related potentials (ERPs) of the EEG. An irrelevant color cue that was either contingent (color search) or non-contingent (shape search) on attentional sets was presented prior to a target array with different stimulus-onset asynchronies (SOA; 200, 400, 800 ms). Behavioral results indicated that color cues captured attention only when they shared target-defining properties. These same-location effects persisted over time but were pronounced when cue and target array were presented in close succession. N2 posterior contralateral (N2pc) showed that the color cue generally drew attention, but was strongest in the contingent condition. A subsequently emerging contralateral posterior positivity referred to the irrelevant cue (i.e., distractor positivity, Pd) was unaffected by the attentional set and therefore interpreted as an inhibitory process required to enable a re-direction of the attentional focus. Contralateral delay activity (CDA) was only observable in the contingent condition, indicating the transfer of spatial information into working memory and thus providing an explanation for the same-location effect for longer SOAs. Inhibition of this irrelevant information was reflected by a second contralateral positivity triggered through target presentation. The results suggest that distracting information is actively maintained when it resembles a sought-after object. However, two independent attentional processes are at work to compensate for attentional distraction: the timely inhibition of attentional capture and the active inhibition of mental representation of irrelevant information.

Somatotopy and temporal dynamics of sensorimotor interactions: evidence from double afferent inhibition.

Moving and interacting with the world requires that the sensory and motor systems share information, but while some information about tactile events is preserved during sensorimotor transfer the spatial specificity of this information is unknown. Afferent inhibition (AI) studies, in which corticospinal excitability (CSE) is inhibited when a single tactile stimulus is presented before a transcranial magnetic stimulation pulse over the motor cortex, offer contradictory results regarding the sensory-to-motor transfer of spatial information. Here, we combined the techniques of AI and tactile repetition suppression (the decreased neurophysiological response following double stimulation of the same vs. different fingers) to investigate whether topographic information is preserved in the sensory-to-motor transfer in humans. We developed a double AI paradigm to examine both spatial (same vs. different finger) and temporal (short vs. long delay) aspects of sensorimotor interactions. Two consecutive electrocutaneous stimuli (separated by either 30 or 125ms) were delivered to either the same or different fingers on the left hand (i.e. index finger stimulated twice or middle finger stimulated before index finger). Information about which fingers were stimulated was reflected in the size of the motor responses in a time-constrained manner: CSE was modulated differently by same and different finger stimulation only when the two stimuli were separated by the short delay (P=0.004). We demonstrate that the well-known response of the somatosensory cortices following repetitive stimulation is mirrored in the motor cortex and that CSE is modulated as a function of the temporal and spatial relationship between afferent stimuli.

Role of the thalamic nucleus reuniens in mediating interactions between the hippocampus and medial prefrontal cortex during spatial working memory

Despite decades of research, the neural mechanisms of spatial working memory remain poorly understood. Although the dorsal hippocampus is known to be critical for memory-guided behavior, experimental evidence suggests that spatial working memory depends not only on the hippocampus itself, but also on the circuit comprised of the hippocampus and the medial prefrontal cortex (mPFC). Disruption of hippocampal-mPFC interactions may result in failed transfer of spatial and contextual information processed by the hippocampus to the circuitry in mPFC responsible for decision making and goal-directed behavior. Oscillatory synchrony between the hippocampus and mPFC has been shown to increase in tasks with high spatial working memory demand. However, the mechanisms and circuitry supporting hippocampal-mPFC interactions during these tasks is unknown. The midline thalamic nucleus reuniens (RE) is reciprocally connected to both the hippocampus and the mPFC and has been shown to be critical for a variety of working memory tasks. Therefore, it is likely that hippocampal-mPFC oscillatory synchrony is modulated by RE activity. This article will review the anatomical connections between the hippocampus, mPFC and RE along with the behavioral studies that have investigated the effects of RE disruption on working memory task performance. The article will conclude with suggestions for future directions aimed at identifying the specific role of the RE in regulating functional interactions between the hippocampus and the PFC and investigating the degree to which these interactions contribute to spatial working memory.

Recording low and high spatial frequencies in exit wave reconstructions

Aberration corrected Transmission Electron Microscope (TEM) images can currently resolve information at significantly better than 0.1 nm. Aberration corrected imaging conditions seek to optimize the transfer of high-resolution information but in doing so they prevent the transfer of low spatial frequency information. To recover low spatial frequency information, aberration corrected images must be acquired at a large defocus which compromises high spatial frequency information transfer. In this paper we present two a posteriori solutions to this problem in which the information bandwidth in an exit wave reconstruction is increased. In the first we reconstruct the electron exit wavefunction from two focal series datasets, with different, uniform focal steps, experimentally demonstrating that the width of the transfer interval can be extended from 0.2 nm⁻¹ (∼5 nm) to better than 10 nm⁻¹ (0.1 nm). In the second we outline the use of a focal series recorded with a non-uniform focal step to recover a wider range of spatial frequencies without the need for a large number of images. Using simulated data we show that using this non-uniform focal step the spatial frequency interval for a five image data set may be increased to between 0.25 nm⁻¹ (4 nm) and 8.3 nm⁻¹ (0.12 nm) compared to between 0.74 nm⁻¹ (1.4 nm) and 8.3 nm⁻¹ (0.12 nm) for the standard focal series geometry.

Metallized DNA nanolithography for encoding and transferring spatial information for graphene patterning

The vision for graphene and other two-dimensional electronics is the direct production of nanoelectronic circuits and barrier materials from a single precursor sheet. DNA origami and single-stranded tiles are powerful methods to encode complex shapes within a DNA sequence, but their translation to patterning other nanomaterials has been limited. Here we develop a metallized DNA nanolithography that allows transfer of spatial information to pattern two-dimensional nanomaterials capable of plasma etching. Width, orientation and curvature can be programmed by specific sequence design and transferred, as we demonstrate for graphene. Spatial resolution is limited by distortion of the DNA template upon Au metallization and subsequent etching. The metallized DNA mask allows for plasmonic enhanced Raman spectroscopy of the underlying graphene, providing information on defects, doping and lattice symmetry. This DNA nanolithography enables wafer-scale patterning of two-dimensional electronic materials to create diverse circuit elements, including nanorings, three- and four-membered nanojunctions, and extended nanoribbons.

Interhemispheric transfer of spatial and semantic information: Electrophysiological evidence

The goal of this study was to cast light on the existence of functional callosal channels for the interhemispheric transfer (IHT) of spatial and semantic information. To do so, we recorded event-related potentials in healthy humans while performing a primed odd-even discrimination task. Targets were visually presented numbers preceded by single-letter primes signaling the probable presentation of an odd or an even number. Primes and targets could appear either in the same or in different visual fields, thus requiring an IHT in the latter case. The P1 and N2 components were influenced by IHT of spatial information only, whereas the later N400 was influenced by IHT of both spatial and semantic information. This was not the case for the P3b, which was modulated by semantic validity only. These results provide novel evidence of the existence of a temporally separated interhemispheric exchange of spatial and semantic information.

The concept of RNA-assisted protein folding: the role of tRNA.

We suggest that tRNA actively participates in the transfer of 3D information from mRNA to peptides - in addition to its well-known, "classical" role of translating the 3-letter RNA codes into the one letter protein code. The tRNA molecule displays a series of thermodynamically favored configurations during translation, a movement which places the codon and coded amino acids in proximity to each other and make physical contact between some amino acids and their codons possible. This specific codon-amino acid interaction of some selected amino acids is necessary for the transfer of spatial information from mRNA to coded proteins, and is known as RNA-assisted protein folding.

Progress in Understanding Spatial Coordinate Systems in the Primate Brain

A new study in this issue of Neuron shows that when monkeys reach to a visual target, neurons in the dorsal premotor cortex compare the location of the target, the hand, and the point of visual fixation. The neurons therefore encode space through a combination of eye-centered and hand-centered coordinates.

The Drosophila Embryonic Patterning Determinant Torsolike Is a Component of the Eggshell

The development of the head and tail regions of the Drosophila embryo is dependent upon the localized polar activation of Torso (Tor), a receptor tyrosine kinase that is uniformly distributed in the membrane of the developing embryo [1, 2]. Trunk (Trk), the proposed ligand for Tor, is secreted as an inactive precursor into the perivitelline fluid that lies between the embryonic membrane and the vitelline membrane (VM), the inner layer of the eggshell [3, 4]. The spatial regulation of Trk processing is thought to be mediated by the secreted product of the torsolike (tsl) gene, which is expressed during oogenesis by a specialized population of follicle cells present at the two ends of the oocyte [5, 6]. We show here that Tsl protein is specifically localized to the polar regions of the VM in laid eggs. We further demonstrate that although Tsl can associate with nonpolar regions of the VM, the activity of polar-localized Tsl is enhanced, suggesting the existence of another spatially restricted factor acting in this pathway. The incorporation of Tsl into the VM provides a mechanism for the transfer of spatial information from the follicle cells to the developing embryo. To our knowledge, Tsl represents the first example of an embryonic patterning determinant that is a component of the eggshell.

043.12 SPATIAL INFORMATION TRANSFER FROM VIRTUAL TO REAL VERSIONS OF THE KIEL LOCOMOTOR MAZE

The Kiel locomotor maze requires participants to choose five targets from among 20 locations marked by small red lights on the floor of a dimly lit circular environment having four wall-mounted extramaze cues and two intramaze cues at floor level. In the present study, acquisition of the real task was examined in 11-year-old children following prior accurate training in a virtual version, following misleading virtual training, or following no training. The virtual version was displayed on a desk-top computer monitor. Acquisition testing in the real maze was either locomotor or non-locomotor. Good transfer was achieved from virtual to real versions. Children’s exploration of the real maze prior to real maze acquisition training revealed a clear transfer of spatial information previously learned in the virtual version. Children taught the correct target configuration in the simulation made fewer errors and more rapid, confident responses to targets in the real maze than children given no training. However, acquisition was also better following misleading training than no training, suggesting that a non-specific components of performance also transferred. Male superiority was only seen following misleading training, which was interpreted in terms of male superiority in mental rotation. After acquisition, a single probe trial was performed, in which proximal cues and participants’ starting position were rotated, but this had equivalent effects on all groups’ performance. It is clear that transfer of spatial information occurs from the simulated Kiel maze to the real version. This has implications for its use in diagnosis and training.

Spatial information transfer from virtual to real versions of the Kiel locomotor maze

The Kiel locomotor maze requires participants to choose five targets from among 20 locations marked by small red lights on the floor of a dimly lit circular environment having four wall-mounted extramaze cues and two intramaze cues at floor level. In the present study, acquisition of the real task was examined in 11-year-old children following prior accurate training in a virtual version, following misleading virtual training, or following no training. The virtual version was displayed on a desk-top computer monitor. Acquisition testing in the real maze was either locomotor or non-locomotor. Good transfer was achieved from virtual to real versions. Children’s exploration of the real maze prior to real maze acquisition training revealed a clear transfer of spatial information previously learned in the virtual version. Children taught the correct target configuration in the simulation made fewer errors and more rapid, confident responses to targets in the real maze than children given no training. However, acquisition was also better following misleading training than no training, suggesting that a non-specific components of performance also transferred. Male superiority was only seen following misleading training, which was interpreted in terms of male superiority in mental rotation. After acquisition, a single probe trial was performed, in which proximal cues and participants’ starting position were rotated, but this had equivalent effects on all groups’ performance. It is clear that transfer of spatial information occurs from the simulated Kiel maze to the real version. This has implications for its use in diagnosis and training.