Visual Discrimination Model Research Articles

Integrated network analyses identify MYB4R1 neofunctionalization in the UV-B adaptation of Tartary buckwheat

A hallmark of adaptive evolution is innovation in gene function, which is associated with the development of distinct roles for genes during plant evolution; however, assessing functional innovation over long periods of time is not trivial. Tartary buckwheat (Fagopyrum tataricum) originated in the Himalayan region and has been exposed to intense UV-B radiation for a long time, making it an ideal species for studying novel UV-B response mechanisms in plants. Here, we developed a workflow to obtain a co-functional network of UV-B responses using data from more than 10,000 samples in more than 80 projects with multi-species and multi-omics data. Dissecting the entire network revealed that flavonoid biosynthesis was most significantly related to the UV-B response. Importantly, we found that the regulatory factor MYB4R1, which resides at the core of the network, has undergone neofunctionalization. In vitro and in vivo experiments demonstrated that MYB4R1 regulates flavonoid and anthocyanin accumulation in response to UV-B in buckwheat by binding to L-box motifs in the FtCHS, FtFLS, and FtUFGT promoters. We used deep learning to develop a visual discrimination model of buckwheat flavonoid content based on natural populations exposed to global UV-B radiation. Our study highlights the critical role of gene neofunctionalization in UV-B adaptation.

Compressing pathology whole-slide images using a human and model observer evaluation

Introduction: We aim to determine to what degree whole-slide images (WSI) can be compressed without impacting the ability of the pathologist to distinguish benign from malignant tissues. An underlying goal is to demonstrate the utility of a visual discrimination model (VDM) for predicting observer performance. Materials and Methods: A total of 100 regions of interest (ROIs) from a breast biopsy whole-slide images at five levels of JPEG 2000 compression (8:1, 16:1, 32:1, 64:1, and 128:1) plus the uncompressed version were shown to six pathologists to determine benign versus malignant status. Results: There was a significant decrease in performance as a function of compression ratio (F = 14.58, P < 0.0001). The visibility of compression artifacts in the test images was predicted using a VDM. Just-noticeable difference (JND) metrics were computed for each image, including the mean, median, ≥90th percentiles, and maximum values. For comparison, PSNR (peak signal-to-noise ratio) and Structural Similarity (SSIM) were also computed. Image distortion metrics were computed as a function of compression ratio and averaged across test images. All of the JND metrics were found to be highly correlated and differed primarily in magnitude. Both PSNR and SSIM decreased with bit rate, correctly reflecting a loss of image fidelity with increasing compression. Observer performance as measured by the Receiver Operating Characteristic area under the curve (ROC Az) was nearly constant up to a compression ratio of 32:1, then decreased significantly for 64:1 and 128:1 compression levels. The initial decline in Az occurred around a mean JND of 3, Minkowski JND of 4, and 99th percentile JND of 6.5. Conclusion: Whole-slide images may be compressible to relatively high levels before impacting WSI interpretation performance. The VDM metrics correlated well with artifact conspicuity and human performance.

A random-walk model of accuracy and reaction time applied to three experiments on pigeon visual discrimination.

A random-walk model of visual discrimination is described and applied to reaction time (RT) distributions from three discrete-trial experiments with pigeons. Experiment 1 was a two-choice hue discrimination task with multiple hues. Choice percentages changed with hue discriminability; RTs were shortest for the least and most discriminable stimuli. Experiments 2 and 3 used go/no-go hue discriminations. Blocks of sessions differed in reward probability associated with a variable red stimulus in Experiment 2 and with a constant green stimulus in Experiment 3. Changes in hue had a large effect on response percentage and a small effect on RT; changes in reward shifted RT distributions on the time axis. The "random-walk, pigeon" model applied to these data is closely related to Ratcliff's diffusion model (Ratcliff, 1978; Ratcliff & Rouder, 1998). Simulations showed that stimulus discriminability affected the speed with which evidence accumulated toward a response threshold, in line with comparable effects in human subjects. Reward probability affected bias, modeled as the amount of evidence needed to reach one threshold rather than the other. The effects of reward probability are novel, and their isolation from stimulus effects within the decision process can guide development of a broader model of discrimination.

Functional Dissociations within the Ventral Object Processing Pathway: Cognitive Modules or a Hierarchical Continuum?

We examined the organization and function of the ventral object processing pathway. The prevailing theoretical approach in this field holds that the ventral object processing stream has a modular organization, in which visual perception is carried out in posterior regions and visual memory is carried out, independently, in the anterior temporal lobe. In contrast, recent work has argued against this modular framework, favoring instead a continuous, hierarchical account of cognitive processing in these regions. We join the latter group and illustrate our view with simulations from a computational model that extends the perceptual-mnemonic feature-conjunction model of visual discrimination proposed by Bussey and Saksida [Bussey, T. J., & Saksida, L. M. The organization of visual object representations: A connectionist model of effects of lesions in perirhinal cortex. European Journal of Neuroscience, 15, 355-364, 2002]. We use the extended model to revisit early data from Iwai and Mishkin [Iwai, E., & Mishkin, M. Two visual foci in the temporal lobe of monkeys. In N. Yoshii & N. Buchwald (Eds.), Neurophysiological basis of learning and behavior (pp. 1-11). Japan: Osaka University Press, 1968]; this seminal study was interpreted as evidence for the modularity of visual perception and visual memory. The model accounts for a double dissociation in monkeys' visual discrimination performance following lesions to different regions of the ventral visual stream. This double dissociation is frequently cited as evidence for separate systems for perception and memory. However, the model provides a parsimonious, mechanistic, single-system account of the double dissociation data. We propose that the effects of lesions in ventral visual stream on visual discrimination are due to compromised representations within a hierarchical representational continuum rather than impairment in a specific type of learning, memory, or perception. We argue that consideration of the nature of stimulus representations and their processing in cortex is a more fruitful approach than attempting to map cognition onto functional modules.

Using a visual discrimination model for the detection of compression artifacts in virtual pathology images.

A major issue in telepathology is the extremely large and growing size of digitized "virtual" slides, which can require several gigabytes of storage and cause significant delays in data transmission for remote image interpretation and interactive visualization by pathologists. Compression can reduce this massive amount of virtual slide data, but reversible (lossless) methods limit data reduction to less than 50%, while lossy compression can degrade image quality and diagnostic accuracy. "Visually lossless" compression offers the potential for using higher compression levels without noticeable artifacts, but requires a rate-control strategy that adapts to image content and loss visibility. We investigated the utility of a visual discrimination model (VDM) and other distortion metrics for predicting JPEG 2000 bit rates corresponding to visually lossless compression of virtual slides for breast biopsy specimens. Threshold bit rates were determined experimentally with human observers for a variety of tissue regions cropped from virtual slides. For test images compressed to their visually lossless thresholds, just-noticeable difference (JND) metrics computed by the VDM were nearly constant at the 95th percentile level or higher, and were significantly less variable than peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) metrics. Our results suggest that VDM metrics could be used to guide the compression of virtual slides to achieve visually lossless compression while providing 5-12 times the data reduction of reversible methods.

An alternate method for using a visual discrimination model (VDM) to optimize softcopy display image quality

Researchers have developed visual discrimination models (VDMs) that can predict a human observer's ability to detect a target object superposed on an image. These models incorporate sophisticated knowledge of the properties of the human visual system. In the predictive approach, termed conventional VDM usage, two input images with and without a target are analyzed by an algorithm that calculates a just-noticeable-difference (JND) index, which is a taken as a measure of the detectability of the target. A new method of using the VDM is described, termed channelized VDM, which involves finding the linear combination of the VDM-generated channels (which are not used in conventional VDM analysis) that has optimal classification ability between normal and abnormal images. The classification ability can be measured using receiver operating characteristic (ROC) or two alternative forced choice (2AFC) experiments, and in special cases they can also be predicted by signal detection theory (SDT) based model-observer methods. In this study simulated background and nodule containing regions were used to validate the new method. It was found that the channelized VDM predictions were in excellent qualitative agreement with human-observer validated SDT predictions. Either VDM method (conventional or channelized) has potential applicability to soft-copy display optimization. An advantage of any VDM-based approach is that complex effects, such as visual masking, are automatically accounted for, which effects are usually not included in SDT-based methods.

Multifractal structure in nonrepresentational art

Multifractal analysis techniques are applied to patterns in several abstract expressionist artworks, painted by various artists. The analysis is carried out on two distinct types of structures: the physical patterns formed by a specific color ("blobs") and patterns formed by the luminance gradient between adjacent colors ("edges"). It is found that the multifractal analysis method applied to "blobs" cannot distinguish between artists of the same movement, yielding a multifractal spectrum of dimensions between about 1.5 and 1.8. The method can distinguish between different types of images, however, as demonstrated by studying a radically different type of art. The data suggest that the "edge" method can distinguish between artists in the same movement and is proposed to represent a toy model of visual discrimination. A "fractal reconstruction" analysis technique is also applied to the images in order to determine whether or not a specific signature can be extracted which might serve as a type of fingerprint for the movement.

Human visual system modeling for selecting the optimal display for digital radiography

Observer performance studies and receiver operating characteristic (ROC) analysis are useful for determining experimentally the effects of imaging parameters on the diagnostic performance of radiologists. Comprehensive optimization of image acquisition, processing, and display systems for digital radiography based solely on measurements of reader performance is impractical, however, due to the large number of interdependent variables to be tested. Observer models that account for the effects of those variables on the visual perception of images offer potential for expediting the evaluation and optimization of new technologies. We have conducted several studies to evaluate the applicability of a visual discrimination model (VDM), JNDmetrix, for predicting trends in observer performance in detection and discrimination tasks as a function of various image display and processing factors: display function and peak luminance, CRT phosphor noise, MTF compensation, viewing angle (LCD vs. CRT), image compression, and grayscale window enhancements. Our results suggest that the VDM could be useful in the assessment and optimization of new image display and processing technologies for digital radiography.

Improved compressibility of multislice CT datasets using 3D JPEG2000 compression

This study evaluated the compressibility of multislice CT (MSCT) datasets and its dependence on (1) slice thickness and (2) the use of three-dimensional (3D) vs. 2D JPEG2000 compression methods. Five thoracic CT datasets were obtained using a 16-detector MSCT scanner with a collimation of 0.75 mm (120 kVp, 90 mAs) and reconstructed at five slice thicknesses from 0.75 to 10.0 mm. These datasets were irreversibly compressed using a standard 2D JPEG2000 encoder and a developmental 3D JPEG2000 algorithm based on Part 2 of the JPEG2000 standard. Compression ratios ranged from 4:1 to 64:1. Image distortion was computed utilizing peak signal-to-noise ratio (PSNR) and the Sarnoff JNDmetrix visual discrimination model. For 2D compression, the thinnest sections were substantially less compressible than thicker sections for the same level of image quality, particularly at higher compression ratios. Applying 3D compression yielded consistently higher image quality in most cases compared to 2D compression at the same ratios. The advantage of 3D compression increased for thinner slices and higher compression ratios. These results indicate that 3D JPEG2000 (Part 2) compression offers substantial advantages over the current 2D JPEG2000 standard, yielding better quantitative image quality at similar compression ratios or comparable image quality at higher compression ratios.

Using a human visual system model to optimize soft-copy mammography display: influence of MTF compensation

Rationale and Objectives. The investigators developed an efficient method for optimizing cathode ray tube (CRT) monitor performance for digital mammography, based on the correlation between the performance of human observers and the performance of a mathematical computer model of the human visual system. The investigators examined observer performance on soft-copy display of mammographic images that were either unprocessed or processed to compensate for modulation transfer function (MTF) deficiencies in the CRT display. The results were used to validate the human visual system model. Materials and Methods. Six radiologists viewed a series of 250 mammographic images with microcalcification clusters with different contrast levels on a CRT monitor. The images were viewed twice: once without image processing and once with processing designed to compensate for MTF deficiencies in the CRT monitor. The images were analyzed with the JNDmetrix Visual Discrimination Model, which is based on the principles of just-noticeable difference measurement and frequency-channel vision modeling. Receiver operating characteristic (ROC) curves were generated for the human observers and compared statistically with the model observers’ performance. Results. Both human and model performance was better overall with the MTF-compensated images, especially for microcalcifications in the midlevel contrast range. There was a very high correlation between human and model observers. Conclusion. The use of image-processing methods to compensate for limitations in the MTF of CRT monitors can improve the detection performance of radiologists searching for microcalcifications in mammographic images, and a model based on characteristics of the human visual system can be used to predict human observer results accurately.

Using a Human Visual System Model to Optimize Soft-Copy Mammography Display: Influence of Display Phosphor

Rationale and Objectives The authors developed an efficient method for optimizing cathode ray tube performance for soft-copy digital mammography displays, based on correlation between the performance of human observers and the performance of a mathematical computer model of the human visual system. The authors measured radiologist performance on soft-copy display monitors with different phosphors and used these results to validate the human visual performance model. Materials and Methods Six radiologists viewed a series of 250 mammographic images with microcalcifications of different contrast levels. They viewed images on two soft-copy display monitors with phosphor luminescence—one with P45 and the other with P104. The same images were analyzed with the JNDmetrix Visual Discrimination Model, which is based on psychophysical just-noticeable difference measurement principles and on frequency-channel vision-modeling principles. Receiver operating characteristic curves were generated for the human and model observers' performances, and results were compared statistically. Results Both human and model performance (area under the receiver operating characteristic curve) was better overall with the P45 than with the P104 monitor, especially for microcalcifications in the midlevel contrast range. There was high correlation between the human and model observers. Conclusion The results indicate that the type of phosphor in a display monitor can influence observer performance significantly and that a model based on characteristics of the human visual system can be used to predict human observer performance accurately.

Neural dynamics of motion processing and speed discrimination.

A neural network model of visual motion perception and speed discrimination is presented. The model shows how a distributed population code of speed tuning, that realizes a size-speed correlation, can be derived from the simplest mechanisms whereby activations of multiple spatially short-range filters of different size are transformed into speed-turned cell responses. These mechanisms use transient cell responses to moving stimuli, output thresholds that covary with filter size, and competition. These mechanisms are proposed to occur in the V1-->MT cortical processing stream. The model reproduces empirically derived speed discrimination curves and simulates data showing how visual speed perception and discrimination can be affected by stimulus contrast, duration, dot density and spatial frequency. Model motion mechanisms are analogous to mechanisms that have been used to model 3-D form and figure-ground perception. The model forms the front end of a larger motion processing system that has been used to simulate how global motion capture occurs, and how spatial attention is drawn to moving forms. It provides a computational foundation for an emerging neural theory of 3-D form and motion perception.

Neural dynamics of motion grouping: from aperture ambiguity to object speed and direction

A neural network model of visual motion perception and speed discrimination is developed to simulate data concerning the conditions under which components of moving stimuli cohere or not into a global direction of motion, as in barberpole and plaid patterns (both type 1 and type 2). The model also simulates how the perceived speed of lines moving in a prescribed direction depends on their orientation, length, duration, and contrast. Motion direction and speed both emerge as part of an interactive motion grouping or segmentation process. The model proposes a solution to the global aperture problem by showing how information from feature tracking points, namely, locations from which unambiguous motion directions can be computed, can propagate to ambiguous motion direction points and capture the motion signals there. The model does this without computing intersections of constraints or parallel Fourier and non-Fourier pathways. Instead, the model uses orientationally unselective cell responses to activate directionally tuned transient cells. These transient cells, in turn, activate spatially short-range filters and competitive mechanisms over multiple spatial scales to generate speed-tuned and directionally tuned cells. Spatially long-range filters and top–down feedback from grouping cells are then used to track motion of featural points and to select and propagate correct motion directions to ambiguous motion points. Top–down grouping can also prime the system to attend a particular motion direction. The model hereby links low-level automatic motion processing with attention-based motion processing. Homologs of model mechanisms have been used in models of other brain systems to simulate data about visual grouping, figure–ground separation, and speech perception. Earlier versions of the model have simulated data about short-range and long-range apparent motion, second-order motion, and the effects of parvocellular and magnocellular lateral geniculate nucleus lesions on motion perception.

A Model of Corticostriatal Plasticity for Learning Oculomotor Associations and Sequences

We present models that learn context-dependent oculomotor behavior in (1) conditional visual discrimination and (2) sequence reproduction tasks, based on the following three principles: (1) Visual input and efferent copies of motor output produce patterns of activity in cortex. (2) Cortex influences the saccade system in part via corticostriatal projections. (3) A reinforcement learning mechanism modifies corticostriatal synapses to link patterns of cortical activity to the correct saccade responses during trial-and-error learning. Our conditional visual discrimination model learns to associate visual cues with the corresponding saccades to one of two left-right targets. A visual cue produces patterns of neuronal activity in inferotemporal cortex (IT) that projects to the oculomotor region of the striatum. Initially random saccadic "guesses," when directed to the correct target for the current cue, result in increased synaptic strength between the cue-related IT cells and the striatal cells that participate in the correct saccade, increasing the probability that this cue will later elicit the correct saccade. We show that the model generates "inhibitory gradients" on the striatum as the substrate for spatial generalization. Our sequence reproduction model learns, when presented with temporal sequences of spatial targets, to reproduce the corresponding sequence of saccades. At any point in the execution of a saccade sequence, the current pattern of activity in pre-frontal cortex (PFC), combined with visual input and the motor efferent copy of the previous saccade, produces a new pattern of activity in PFC to be associated with the next saccade. Like IT, PFC also projects to the oculomotor region of the striatum. Correct guesses for the subsequent saccade in the sequence results in strengthening of corticostriatal synapses between active PFC cells and striatal cells involved in the correct saccade. The sequence is thus reproduced as a concatenation of associations. We compare the results of this model with data previously obtained in the monkey and discuss the nature of cortical representations of spatiotemporal information.

A neural model of discrete and continuous modes of visual discrimination.

A simple neural model of visual pattern discrimination is presented, based on the assumption that visual discrimination is determined by the differences in activity of a certain set of filters. These filters are the input units of a three-layer network. The units of the second layer have probabilities of activation that depend on the differences in activity of the input units and in turn they determine the probabilities of activation of a layer of output units from which probabilities of correct discrimination can be predicted. This model has been used to simulate discrete and continuous modes of curved-line discrimination and the results show good agreement with experimental data found in the literature.

A model of visual texture discrimination using multiple weak operators and spatial averaging

A model of texture segmentation is presented based on the combination of the outputs of six weak texture operators. The combination procedure is a simple statistic in which the center of gravity of the set of suggested boundaries produced by the six operators is computed. The model is effective both as a procedure for texture segmentation and as a model of an important human visual process. The need for heuristic rules suggests that top-down processing may be a general requirement for any theory of human vision.

Configurational pattern discrimination responsible for dishabituation in common toads Bufo bufo (L.): behavioral tests of the predictions of a neural model.

Recently, a neural model of visual pattern discrimination for stimulus-specific habituation was developed, based on previous behavioral studies which demonstrated that toads exhibit a dishabituation hierarchy for different worm-like stimuli. The model suggests that visual objects are represented by temporal coding and predicts that the dishabituation hierarchy changes when the stimulus/background contrast direction is reversed or the stimulus size is varied. The behavioral experiments reported in this paper were designed to test these predictions. (1) For a pair of stimuli from the contrast reversal prediction, the experimental results validated the theory. (2) For a pair of stimuli from the size reduction prediction, the experimental results failed to validate the theory. Further experiments concerning size effects suggest that configural visual pattern discrimination in toads exhibits size invariance. (3) Inspired by the Groves-Thompson account of habituation, we found that dishabituation by a second stimulus has a separate process from habituation to a first stimulus. This paper serves as an example of a fruitful dialogue between experimentation and modeling, crucial for understanding brain functions.

Population coding of stimulus orientation by striate cortical cells

I have examined the performance of a population coding model of visual orientation discrimination, similar to the population coding models proposed for the coding of limb movements. The orientation of the stimulus is not represented by a single unit but by an ensemble of broadly tuned units in a distributed way. Each unit is represented by a vector whose magnitude and direction correspond to the response magnitude and preferred orientation of the unit, respectively. The orientation of the population vector, i.e. the vector sum of the ensemble of units, is the signalled orientation on a particular trial. The accuracy of this population vector orientation coding was determined as a function of a number of parameters by computer simulation. I have shown that even with broadly orientation tuned units possessing considerable response variance, the accuracy of the orientation of the population vector can be as good as behaviorally measured just noticeable differences in orientation. The accuracy of the population code is shown to depend upon the number of units, the average response strength, the orientation band-width, response variability and the response covariance. The results of these simulations were also compared to predictions derived from psychophysical studies of orientation discrimination.

Model of Visual Detection and Discrimination

A model of visual detection and discrimination is presented which incorporates recent concepts of multiple, spatially-tuned channels. Predictions about performance on detection and discrimination tasks are derived and supporting data presented. The predictions and data are inconsistent with threshold-type models.

Effect of scaling optic-nerve impulses on increment thresholds.

Renewal theory is applied to a particle-counter model of visual discrimination in order to determine the effects of neural-impulse scaling and dead time on the detection of increment thresholds. Let the ratio of absorbed photons to neural spikes (scaling factor) be denoted by r and the dead time by τ. We show that the particle counter is equivalent to one with dead time τ* = τ/r and scaling factor r* = 1. Further, if τ = 0, the particle counter does not exhibit the Weber–Fechner behavior for high background luminances as predicted by Barlow. These are asymptotic results, valid for large observation times. For more general observation times, the performance of a particle-counter mechanism with r = 2 and τ = 0 is evaluated for different types of starting procedures.