Directional features for automatic tumor classification of mammogram images

One way for breast cancer diagnosis is provided by taking radiographic (X-ray) images (termed mammograms) for suspect patients, images further used by physicians to identify potential abnormal areas thorough visual inspection. When digital mammograms are available, computer-aided based diagnostic may help the physician in having a more accurate decision. This implies automatic abnormal areas detection using segmentation, followed by tumor classification. This work aims at describing an approach to deal with the classification of digital mammograms. Patches around tumors are manually extracted to segment the abnormal areas from the remaining of the image, considered as background. The mammogram images are filtered using Gabor wavelets and directional features are extracted at different orientation and frequencies. Principal Component Analysis is employed to reduce the dimension of filtered and unfiltered high-dimensional data. Proximal Support Vector Machines are used to final classify the data. Superior mammogram image classification performance is attained when Gabor features are extracted instead of using original mammogram images. The robustness of Gabor features for digital mammogram images distorted by Poisson noise with different intensity levels is also addressed.     

Analysis of Visual Cognitive Impairments in Schizophrenia at the Early Stages of the Disease and Their Correction by Interactive Virtual Environment

The paper describes the results of studies aimed at evaluating the effect of interactive virtual environments on the visual system, including the magno-and parvo-systems. Analysis was conducted in patients diagnosed with paranoid schizophrenia diagnosed from one to five years ago. Comparative analysis of visual evoked potentials during the perception of images that differed in their semantic (animate/inanimate) and physical characteristics (filtration images at high/low spatial frequencies) was used for the assessment of the impact of virtual environments. The images of objects were filtered via digital filtration for selective effect on the magno-and parvo-channels of the visual system. To evaluate the function of visual perception, the measurement of contrast sensitivity using Gabor elements was used. At the early stages of schizophrenia, the patients exhibited a decrease in the amplitudes of the components of cognitive visual evoked potentials to stimuli filtered at high spatial frequencies and reduced contrast sensitivity at high spatial frequencies. The effect of virtual environments on the visual system resulted in a significant increase in the amplitude of the cognitive components of visual evoked potentials in the paradigm of presentation of images filtered at the high spatial frequencies, which allows the conclusion about a stimulating effect of the virtual environment on the parvo-system functioning. The activation of the magno-system occurred to a lesser extent. The present study represents the findings obtained by the studies of cognitive impairments in schizophrenia and the methods of their correction conducted at the Laboratory of Physiology of Vision of the Pavlov Institute of Physiology of the Russian Academy of Sciences and at the Laboratory of Neurobiology of Action Programming of the Bechtereva Institute of the Human Brain of the Russian Academy of Sciences.     

An ASIC-chip for stereoscopic depth analysis in video-real-time based on visual cortical cell behavior

In a stereoscopic system both eyes or cameras have a slightly different view. As a consequence small variations between the projected images exist ("disparities") which are spatially evaluated in order to retrieve depth information. We will show that two related algorithmic versions can be designed which recover disparity. Both approaches are based on the comparison of filter outputs from filtering the left and the right image. The difference of the phase components between left and right filter responses encodes the disparity. One approach uses regular Gabor filters and computes the spatial phase differences in a conventional way as described already in 1988 by Sanger. Novel to this approach, however, is that we formulate it in a way which is fully compatible with neural operations in the visual cortex. The second approach uses the apparently paradoxical similarity between the analysis of visual disparities and the determination of the azimuth of a sound source. Animals determine the direction of the sound from the temporal delay between the left and right ear signals. Similarly, in our second approach we transpose the spatially defined problem of disparity analysis into the temporal domain and utilize two resonators implemented in the form of causal (electronic) filters to determine the disparity as local temporal phase differences between the left and right filter responses. This approach permits video real-time analysis of stereo image sequences (see movies at http://www.neurop.ruhr-uni-bochum.de/Real- Time-Stereo) and a FPGA-based PC-board has been developed which performs stereo-analysis at full PAL resolution in video real-time. An ASIC chip will be available in March 2000.     

Detection of a gabor patch superimposed on an illusory contour

Interactions between visual stimuli have been found to be specific to the spatial frequency, orientation and phase of the interacting stimuli. We asked if there are any interactions between luminance-defined Gabor patches and Kanizsa-type illusory contours. In psychophysical experiments we studied whether induction of a vertical illusory line affects detection thresholds for a Gabor patch superimposed on this line and whether these effects depend on the orientation, spatial frequency and phase of the Gabor elements. Employing a 2AFC method with a staircase procedure we measured contrast detection thresholds and varied the orientation, spatial frequency and phase of the test Gabor patch and the separation between the two pacmen in four experimental series. The results show that in a situation where the two inducers generate perception of an illusory line, the contrast detection of the Gabor patch is facilitated relative to a control condition where the rotated pacmen do not induce illusory contours. This facilitation was more pronounced for test Gabor signals that were collinear to the illusory line, but the observer's performance was not altered by changes in the spatial frequency or phase of the Gabor stimuli. With increasing spatial separation of the two pacmen (and, consequently, with a decreasing support ratio), the difference between performance in the test and control conditions diminished. From the data obtained we cannot infer that we have measured some neural interactions between Gabor patches and Kanizsa-type illusory contours, and nor can we draw a unique conclusion about what causes the facilitation of detection of the test Gabor patch in the experimental situation that allows induction of the illusory line. We discuss possible mechanisms of the facilitation, such as contextual influences or a reduction of uncertainty about spatial location of the test Gabor patch.     

A 'first stage' central performance drop in a Gabor luminance-modulation detection task

In an experiment, 20 participants had to detect a backward masked Gabor luminance-modulation target imposed on a field of uniform luminance at varying eccentricities along the horizontal meridian. Different spatial frequencies were used as target modulations. Results for a 7.0 c/deg target patch showed peak detection performance at the center of the visual field and a steady decrease toward the periphery. For 1.0 c/deg, 0.75 c/deg, and 0.5 c/deg target patches, in contrast, the peak was several degrees off retinal center and decreased steadily toward the center. Findings not only confirmed the familiar sensitivity loss toward peripheral areas for high spatial frequencies, but also indicated a sensitivity loss toward central areas for low spatial frequencies. It is concluded that they further support Gurnsey et al.'s (1996) 'mismatch hypothesis' extending its scope to also include 'first-stage' stimuli.     

Is correspondence search in human stereo vision a coarse-to-fine process?

One possible strategy for the solution of the correspondence problem of stereo matching is the coarse-to-fine mechanism: The matching process starts with a lowpass-filtered version of the stereogram where only a few, high-contrast image features can be extracted and the probability of false matches is therefore low. In subsequent stages, information from higher spatial frequencies is used gradually to improve the correspondence data obtained on the coarser scales. Coarse-to-fine strategies predict that information from coarse scale is used to disambiguate matching information on finer scales. We have tested this prediction by means of the wallpaper illusion using periodic intensity-profiles with different matching ambiguities on different spatial scale. Our psychophysical experiments show (i) that unambiguous information at coarse scale is not always used to disambiguate finer scale information, (ii) that unambiguous fine-scale information can be used to disambiguate coarse-scale information and (iii) that low spatial frequency is more efficient for disambiguation than higher frequency. We conclude that the human stereo vision system does not always proceed from coarse to fine. As an alternative scheme for scale-space integration, we discuss more symmetric schemes such as maximum likelihood combinations of data from different channels. Received: 29 March 1994 / Accepted in revised form: 20 September 1995     

What Image Properties Regulate Eye Growth?

The growth of the eye, unlike other parts of the body, is not ballistic. It is guided by visual feedback with the eventual aim being optimal focus of the retinal image or emmetropization . It has been shown in animal models that interference with the quality of the retinal image leads to a disruption to the normal growth pattern, resulting in the development of refractive errors and defocused retinal images . While it is clear that retinal images rich in pattern information are needed to control eye growth, it is unclear what particular aspect of image structure is relevant. Retinal images comprise a range of spatial frequencies at different absolute and relative contrasts and in different degrees of spatial alignment. Here we show, by using synthetic images, that it is not the local edge structure produced by relative spatial frequency alignments within an image but rather the spatial frequency composition per se that is used to regulate the growth of the eye. Furthermore, it is the absolute energy at high spatial frequencies regardless of the spectral slope that is most effective. Neither result would be expected from currently accepted ideas of how human observers judge the degree of image "blur" in a scene where both phase alignments and the relative energy distribution across spatial frequency (i.e., spectral slope) are important.     

Is slowness a learning principle of the visual cortex?

Slow feature analysis is an algorithm for extracting slowly varying features from a quickly varying signal. It has been shown in network simulations on one-dimensional stimuli that visual invariances to shift and other transformations can be learned in an unsupervised fashion based on slow feature analysis. More recently, we have shown that slow feature analysis applied to image sequences generated from natural images using a range of spatial transformations results in units that share many properties with complex and hypercomplex cells of the primary visual cortex. We find cells responsive to Gabor stimuli with phase invariance, sharpened or widened orientation or frequency tuning, secondary response lobes, end-stopping, and cells selective for direction of motion. These results indicate that slowness may be an important principle of self-organization in the visual cortex.     

Spatial and temporal dependencies of cross-orientation suppression in human vision

A well-known property of orientation-tuned neurons in the visual cortex is that they are suppressed by the superposition of an orthogonal mask. This phenomenon has been explained in terms of physiological constraints (synaptic depression), engineering solutions for components with poor dynamic range (contrast normalization) and fundamental coding strategies for natural images (redundancy reduction). A common but often tacit assumption is that the suppressive process is equally potent at different spatial and temporal scales of analysis. To determine whether it is so, we measured psychophysical cross-orientation masking (XOM) functions for flickering horizontal Gabor stimuli over wide ranges of spatio-temporal frequency and contrast. We found that orthogonal masks raised contrast detection thresholds substantially at low spatial frequencies and high temporal frequencies (high speeds), and that small and unexpected levels of facilitation were evident elsewhere. The data were well fit by a functional model of contrast gain control, where (i) the weight of suppression increased with the ratio of temporal to spatial frequency and (ii) the weight of facilitatory modulation was the same for all conditions, but outcompeted by suppression at higher contrasts. These results (i) provide new constraints for models of primary visual cortex, (ii) associate XOM and facilitation with the transient magno- and sustained parvostreams, respectively, and (iii) reconcile earlier conflicting psychophysical reports on XOM.     

Preferred spatial frequencies for human face processing are associated with optimal class discrimination in the machine

Psychophysical studies suggest that humans preferentially use a narrow band of low spatial frequencies for face recognition. Here we asked whether artificial face recognition systems have an improved recognition performance at the same spatial frequencies as humans. To this end, we estimated recognition performance over a large database of face images by computing three discriminability measures: Fisher Linear Discriminant Analysis, Non-Parametric Discriminant Analysis, and Mutual Information. In order to address frequency dependence, discriminabilities were measured as a function of (filtered) image size. All three measures revealed a maximum at the same image sizes, where the spatial frequency content corresponds to the psychophysical found frequencies. Our results therefore support the notion that the critical band of spatial frequencies for face recognition in humans and machines follows from inherent properties of face images, and that the use of these frequencies is associated with optimal face recognition performance.     

The effect of binocular stimulation on each component of transient and steady-state VEPs

We recorded the monocular and binocular VEPs to the alternation of sinusoidal gratings in order to evaluate the binocular interaction in each component of transient and steady-state VEPs in 13 normal subjects. Three spatial frequencies (1.3, 2.6 and 5.3 c/deg) with a 90% contrast were used as visual stimuli. The latencies and amplitudes of N70 and P100 of the transient VEPs were measured. The steady-state VEPs were Fourier analyzed, and both the phase and amplitude of the second (2F) and fourth (4F) harmonic responses were obtained. Binocular interaction was influenced by spatial frequency such that a binocular summation or even an inhibition occurred. For the transient VEPs, a binocular summation was more pronounced in the amplitude of N70 than in that of P100 at all spatial frequencies. There were no significant effects of binocular stimulation on latencies of N70 or P100. However, the latencies of N70 and P100 showed different spatial frequency characteristics. For the steady-state VEPs, the amplitude of 2F revealed a binocular summation that was more pronounced at 5.3 c/deg, whereas the 4F amplitude showed binocular inhibition at 2.6 and 5.3 c/deg. The 2F phase showed binocular inhibition at all spatial frequencies, whereas no such inhibition was observed in the 4F phase. These results suggest that individual components of transient and steady-state VEPs are physiologically distinct and may therefore be generated from different neuronal populations in striate cortex.     

Bending vibrations of the femur and the oscillatory behavior of a cemented femoral hip endoprosthesis

The paper presents a novel method for recording amplitude and phase of 6D-vibrations of a spatial pendulum over a wide frequency range (10 Hz up to 20 kHz). The six degrees of freedom of the pendulum mass were monitored by three electrodynamic stereo pickups. At rest, the tips of the needles and the pendulum's center of mass defined the reference system with respect to which the oscillations of the mass were recorded in terms of their amplitudes and phases. Its small dimensions, constant transfer characteristics, linearity, high dynamics, and virtual lack of reaction onto the moving system over the entire frequency range provided the advantages of the measuring system. This method was used to analyze the spatial 6D-vibrations of the head of a cemented femoral hip endoprosthesis when the femur was stimulated to bending vibrations. The head of the prosthesis carried out axial rotational vibrations at every frequency used to stimulate the femur. The amplitudes of the axial rotations of the cortical bone were small in comparison to the ones of the prosthesis head, indicating that axial rotational vibrations following femur bending vibrations mainly stressed the spongiosa and the cement layer. This was observed over the entire frequency range, including at the low frequencies relevant for gait. Over the low-frequency range, as well as at some of the higher resonance frequencies, stationary instantaneous helical axes characterized the vibrations. The measurements suggest the mechanism that the interface "implant-bone" may already be stressed by axial torsional loads when the femur is loaded by bending impacts that are known to occur during walking.     

Spatial pattern summation is phase-insensitive in the fovea but not in the periphery.

We measured the effect of phase coherence on the detectability of a chain of Gabor patches. The stimuli were elongated patterns consisting of 1-8 vertical Gabor patches, aligned vertically and either in phase or 180 deg out of phase from their immediate neighbors. In the fovea, the phase configuration had no effect on detection threshold, suggesting a full-wave rectification operation mediating foveal spatial summation. At 5 deg in the periphery, detection thresholds for the in-phase configuration were lower than those for the alternating-phase configuration, to an extent compatible with a half-wave rectification operation mediating peripheral spatial summation.     

Configuration effects on texture transparency

This study examined the factors producing the perception of transparency between overlaid regions composed of Gabor micro-patterns as functions of their spatial frequency, separation of overlaid regions, and types of orientation modulation. The results showed that the likelihood of perceiving transparency was high both when (1) the difference in Gabor spatial frequency between regions was large, and (2) the region boundary, which was formed by short-range orientation differences in the Gabor micro-patterns, clearly emerged. We conclude that texture transparency appears to result from an interaction between a boundary-detection mechanism defining the shape of each region and a surface-detection mechanism assigning the boundary.     

Does face image statistics predict a preferred spatial frequency for human face processing?

Psychophysical experiments suggested a relative importance of a narrow band of spatial frequencies for recognition of face identity in humans. There exists, however, no conclusive evidence of why it is that such frequencies are preferred. To address this question, I examined the amplitude spectra of a large number of face images and observed that face spectra generally fall off more steeply with spatial frequency compared with ordinary natural images. When external face features (such as hair) are suppressed, then whitening of the corresponding mean amplitude spectra revealed higher response amplitudes at those spatial frequencies which are deemed important for processing face identity. The results presented here therefore provide support for that face processing characteristics match corresponding stimulus properties.     

On the Contribution of Binocular Disparity to the Long-Term Memory for Natural Scenes

Binocular disparity is a fundamental dimension defining the input we receive from the visual world, along with luminance and chromaticity. In a memory task involving images of natural scenes we investigate whether binocular disparity enhances long-term visual memory. We found that forest images studied in the presence of disparity for relatively long times (7s) were remembered better as compared to 2D presentation. This enhancement was not evident for other categories of pictures, such as images containing cars and houses, which are mostly identified by the presence of distinctive artifacts rather than by their spatial layout. Evidence from a further experiment indicates that observers do not retain a trace of stereo presentation in long-term memory.     

A new method in computer-assisted imaging in neuroanatomy

A procedure is described yielding computed images of postmortem brains with high topographic accuracy. Structures of the brain are traced and registered by means of a digitizer capable of measuring coordinates three-dimensionally. The information corresponding to one brain model is stored on a flexible disk with a capacity of 256 Kbytes. According to the output desired, the resulting brain images are either completely or partially displayed on the computer screen as stereo pairs. The brain models possess a local fidelity of about 1 mm. The images are useful in simultaneously studying superficial and central parts of the brain, spatial relationships of the various structures and the projection of deep structures onto the surface of the brain. A RAM of about 100 Kbytes is necessary for a program enabling the user to perform stereo projections, three-dimensional transformations and other image manipulations. The special features of anatomical computer imaging as compared to computed tomography (CT) and nuclear magnetic resonance imaging (NMR) are outlined. A combination of these different techniques seems to improve clinical diagnosis.     

空间频率通道中噪声对体视匹配的影响

本工作希望了解噪声在各空间频率通道中对体视的影响.产生从5%至35%每隔5%的七个等级的随机点噪声图象,参照Wilson四通道模型的频率分别用两种不同滤波器对随机点立体图对(RDS)及噪声图象进行滤波,然后在各通道图对中加入不同通道噪声进行匹配观察.得到:1.高频通道的体视抗干扰能力高于低频通道.2.Gabor函数滤波后图对的体视抗干扰能力高于Butterworth滤波器滤波后的,但均低于未滤波的.3.当噪声大于15%时,低频噪声对低频图对及高频噪声对高、低频图对的体视匹配有抑制作用,而低频噪声对高频图对则影响很小.4.未滤波及各通道图对的最大噪声容限均不超过25%.     

Topographic Factor Analysis: A Bayesian Model for Inferring Brain Networks from Neural Data

The neural patterns recorded during a neuroscientific experiment reflect complex interactions between many brain regions, each comprising millions of neurons. However, the measurements themselves are typically abstracted from that underlying structure. For example, functional magnetic resonance imaging (fMRI) datasets comprise a time series of three-dimensional images, where each voxel in an image (roughly) reflects the activity of the brain structure(s)–located at the corresponding point in space–at the time the image was collected. FMRI data often exhibit strong spatial correlations, whereby nearby voxels behave similarly over time as the underlying brain structure modulates its activity. Here we develop topographic factor analysis (TFA), a technique that exploits spatial correlations in fMRI data to recover the underlying structure that the images reflect. Specifically, TFA casts each brain image as a weighted sum of spatial functions. The parameters of those spatial functions, which may be learned by applying TFA to an fMRI dataset, reveal the locations and sizes of the brain structures activated while the data were collected, as well as the interactions between those structures.