医学图像的零树小波编码

背景：医学数字图像必须是高质量的、高分辨率,所以数据量很大,如此巨大的数据量不利于图像存档与传输系统的运行和数字化医院、远程医疗的实现。因此,图像压缩成为图像存档与传输系统要解决的重要问题。 目的：分析零树小波变编码算法原理并编程实现对医学数字图像的压缩,使之能够满足医学图像的传输和诊断要求。 方法：应用嵌入式零树小波编码算法,探讨小波基和小波变换层数的选择,编程实现对医学数字图像的压缩。 结果与结论：选择双正交小波基对医学图像进行4层小波变换实现压缩,获得了较高的峰值信噪比,取得了较好的压缩效果。 关键词：小波变换;图像压缩;嵌入式零树小波;医学数字图像;数字化医学     

基于稀疏性与同步性的视皮质假体图像处理策略

背景：视觉系统是人类获知外部世界的重要感觉系统,因各种疾病或其他损伤导致视觉丧失的盲人数量与日俱增,视皮质假体的研究有望为盲人提供一条复明的途径。视皮质假体虽然经过多年的研究,在某些方面取得了一定的进展和成果,但仍存在很多问题和挑战。 目的：针对置入视皮质的刺激电极数量有限这一问题,提出一种基于稀疏性和同步性的视皮质假体图像处理策略。 方法：该策略借鉴视觉皮质的生理特性,首先采用基于稀疏性的拓扑独立成分分析方法提取自然图像的基函数及神经元的响应,然后采用基于同步性的脉冲耦合神经网络方法对响应系数进行筛选,获取最佳神经元响应,最后利用最佳神经元响应刺激相应的神经元达到恢复自然图像的目的。 结果与结论：实验结果表明,该策略理论上可以利用较少的神经元表征自然图像中的重要信息,应用于视皮质假体可有效减少置入刺激电极的数量,达到更好的信息传递效果。     

比较二维小波和二维小波包技术在不同压缩模式下的降噪处理

背景：小波和小波包技术是进行时频信号分析的重要方法。医学图像数字化采集后断层多，数据信息量大，易受噪声影响。采用二维小波技术和小波包技术可以实现肝癌图像的完美压缩和降噪。 目的：比较二维小波和二维小波包技术在不同压缩模式下压缩肝癌图像的优劣以及小波降噪的技巧。 方法：选用同一幅动脉期肝癌图像，进行4 层分解，利用二维小波和二维小波包技术分别进行Balance sparsity-norm、Remove near0和Bal.sparsity-norm(sqrt)三种模式的压缩处理，再利用小波函数对含噪声信号的图像进行降噪处理。 结果与结论：对于同一种压缩模式，二维小波包技术压缩肝癌图像优于二维小波技术，3种压缩模式中Bal.sparsity-norm(sqrt)模式和Remove near0 mode模式压缩比例更小，图像清晰度更好；小波降噪能很好地消除噪声信号。提示利用二维小波技术和小波包技术都可以实现肝癌图像的完美压缩和降噪。     

形觉剥夺性弱视成年大鼠视皮质c-fos表达的研究

目的研究双眼形觉剥夺性弱视成年大鼠光刺激诱导视皮质神经元c-fos表达情况。方法缝合14日龄SD大鼠双侧眼睑建立双眼形觉剥夺性弱视模型9只．正常对照组5只。饲养至100日龄两组成年大鼠均暴露于外界自然光线中0．5h后取材．对两组视皮质组织进行Nissl染色和c-fos蛋白免疫组化染色，并对染色结果进行计算机图像分析。结果弱视组大鼠视皮质表达c—fos蛋白阳性神经元较正常对照组多，两组差异有显著性（P〈0．05）。结论双眼形觉剥夺性弱视成年大鼠视皮质经光刺激诱导c-fos表达增加，提示已过视觉发育敏感期成年大鼠的视皮质仍存留一定程度的视觉可塑性。     

单眼视觉剥夺大鼠视皮质基因表达差异的研究

本实验利用差异显示反转录ＰＣＲ（ＤＤＲＴ－ＰＣＲ）技术分析单眼视觉剥夺大鼠左、右侧视皮质基因差异表达情况，并通过反Northern杂交去除假阳性筛选差异表达基因片段，最终获得３个差异表达基因片段。通过克隆、测序和同源性比较发现，其中一个片段５'端的１６６个碱基与人ＫＩＡＡ０５４１蛋白的mRNA编码区的１６６bp区段的８３％的同源性。本研究表明单眼视沉剥夺导致视皮质基因表达的改变。     

小波先验在OSL重建算法中的应用***★

背景：MAP(最大后验)统计重建方法可以在重建过程中引入合适的先验知识达到去除噪声的目的。 目的：根据小波系数的统计特性及能量平衡的原理对高频信息做相应的处理，并将多尺度的小波先验应用到OSL重建算法中以去除噪声。 方法：实验从“变换域”的思想出发，在小波域上根据小波系数的统计特性及能量平衡原理对不同尺度的高频信息做相应的处理，并利用处理后的小波系数进行小波重建。 结果与结论：基于小波先验的OSL算法比ML-EM算法重建的图像与测试模型的误差变小、相关性变大、噪声变少，重建图像变得比较平滑，视觉效果较清楚。     

图片命名处理过程中的脑电变化及其空间溯源定位

目的探讨高密度脑电图及其溯源定位技术研究视觉语言处理加工过程,构建视觉语言处理的时空间网络,阐述大脑对图片信息的动态处理。方法本研究入组30例正常志愿受试者,均完成靶刺激和背景刺激记录。剔除数据记录不佳者,实际数据分析23例,分为靶刺激组和背景刺激组,每组均为23例。使用无创256导联高密度脑电图监测图片命名任务,分析事件相关电位和γ波小波,利用任务脑电信息与背景刺激有显著相关时间段的脑电数据进行溯源定位分析。结果事件相关电位分析:图片命名任务处理过程早期在100 ms和170 ms处出现峰值电位,295ms和550～640 ms时段脑电数据目标刺激与背景刺激时段比较,差异有统计学意义(P0.05)。脑电γ波小波分析显示波段35～45 Hz、时段350～400 ms处差异有统计学意义(P0.05)。时间段溯源定位研究提示:在100 ms、170 ms、295ms、350～400 ms和550～640 ms时段,分别发现双侧顶下小叶参与视觉信息的整合过程、双侧前颞叶参与概念的提取过程,左侧前颞叶和左侧顶下小叶协同对词意表征进行处理、左侧缘上回与岛盖部进行了语音编码、Broca区与运动皮质协同处理发音过程。结论高密度脑电图及溯源定位技术可以从语言处理过程的时间维度和定位空间维度构建视觉-语言通路在不同脑区之间的动态联系过程。     

全膝关节假体三维有限元模型的建立

背景：由于全膝关节的结构形态具有运动复杂、受力复杂等特性，造成了数据采集，模型建立困难，影响了三维实体模型的准确性。 目的：建立全膝关节假体三维有限元模型。 方法：通过Microscribe G2三维定位扫描仪取得假体数据、Geomagic软件进行曲面拟合、导入大型有限元分析软件Abaqus6.7.2建立全膝关节假体三维有限元模型，对如何提高三维有限元模型的精确性，扩大模型的开放性以及提高建模效率，增加关节外科医师在建模过程中的参与性进行探讨。 结果与结论：通过Microscribe G2三维定位扫描仪取得假体数据、Geomagic软件进行曲面拟合、导入大型有限元分析软件Abaqus6.7.2建立了全膝关节假体三维有限元模型。与以往建模方法比较，该模块设计使模型更加精准，使用更灵活，简化了有限元前期处理过程，明显降低了建模难度，提高了建模效率，增加了模型的扩展形，并获得了更高的精度。     

视皮质假体中多路信号的无线传输

背景：信号无线传输系统担任着视皮质假体体内外信息交互的任务，为了使假体能够对刺激区域进行多靶位的有效刺激，需要从体外传输进来多个不同的刺激信号。 目的：提出一种多路信号无线传输系统的设计方案。 方法：利用频分复用技术，能够在单一通道上完成多路信号的传输，整个系统设计分体外发射电路和体内接收电路两部分，且接收电路采用无源设计，减少了供能给植入体带来的危害。在给出具体设计电路的同时，对影响系统传输效率的因素做了进一步分析，通过Multisim仿真实验验证其可行性。 结果与结论：该多路信号无线传输系统可以扩展应用到各类植入式假体的信号传输部分，并且采用体内无源设计，避免了为电路供能给人体带来的伤害，通过初步的仿真实验证明该系统具有可行性，具有广泛的应用前景。     

基于小波变换和高斯差分的冷冻电镜生物大分子图像自动分割

冷冻电镜生物大分子图像分割是进行冷冻电镜生物大分子颗粒识别的基础。本文分析了冷冻电镜生物大分子图像的主要特点,提出了基于小波变换和高斯差分（DoG）的冷冻电镜生物大分子图像自动分割方法。该方法利用小波变换得到原图像的低分辨率图像,抑制了噪声,提高了图像的对比度;同时采用DoG算子解决了图像亮度不均匀的问题,并对DoG图像采用基于灰度梯度信息融合的分割方法。实验结果表明该算法能有效的减少噪音对边缘提取的影响,分割效果良好,是一种全新的冷冻电镜生物大分子图像自动分割算法。     

Changes of spatial and temporal frequency tuning properties of neurons in the middle temporal area of aged rhesus monkeys

Aged humans exhibit severe deficits in visual motion perception and contrast sensitivity under various levels of spatial and temporal modulation. Previous studies indicated that many of these deficits are probably mediated by the neural degradation of the central visual system. To clarify the neuronal response mechanisms underlying the visual degradation during aging, we examined the spatial and temporal frequency tuning properties of neurons from anesthetised and paralysed aged monkeys at the middle temporal area (area MT), which is downstream of the primary visual cortex in the visual processing pathway and thought to be critical for motion perception. We found that the preferred spatial and temporal frequencies, spatial resolution and high temporal frequency cutoff of area MT neurons were reduced in aged monkeys, and were accompanied by the broadened tuning width of spatial frequency, elevated spontaneous activity, and decreased signal‐to‐noise ratio. These results showed that, for neurons in area MT, aging significantly changed both the spatial and temporal frequency response tuning properties. Such evidence provides new insight into the changes occurring at the electrophysiological level that may be related to the aging‐related visual deficits, especially in processing spatial and temporal information.     

Attentional modulation in visual cortex is modified during perceptual learning

Practicing a visual task commonly results in improved performance. Often the improvement does not transfer well to a new retinal location, suggesting that it is mediated by changes occurring in early visual cortex, and indeed neuroimaging and neurophysiological studies both demonstrate that perceptual learning is associated with altered activity in visual cortex. Theoretical treatments tend to invoke neuroplasticity that refines early sensory processing. An alternative possibility is that performance is improved because of an altered attentional strategy and that the changes in early visual areas reflect locally altered top-down attentional modulation. To test this idea, we have used functional MRI to examine changes in attentional modulation in visual cortex while participants learn an orientation discrimination task. By examining activity in visual cortex during the preparatory period when the participant has been cued to attend to an upcoming stimulus, we isolated the top-down modulatory signal received by the visual cortex. We show that this signal changes as learning progresses, possibly reflecting gradual automation of the task. By manipulating task difficulty, we show that the change mirrors performance, occurring most quickly for easier stimuli. The effects were seen only at the retinal locus of the stimulus, ruling out a generalized change in alertness. The results suggest that spatial attention changes during perceptual learning and that this may account for some of the concomitant changes seen in visual cortex.     

Prefrontal cortex hemispheric specialization for categorical and coordinate visual spatial memory

During visual spatial perception of multiple items, the left hemisphere has been shown to preferentially process categorical spatial relationships while the right hemisphere has been shown to preferentially process coordinate spatial relationships. We hypothesized that this hemispheric processing distinction would be reflected in the prefrontal cortex during categorical and coordinate visual spatial memory, and tested this hypothesis using functional magnetic resonance imaging (fMRI). During encoding, abstract shapes were presented in the left or right hemifield in addition to a dot at a variable distance from the shape (with some dots on the shape); participants were instructed to remember the position of each dot relative to the shape. During categorical memory retrieval, each shape was presented centrally and participants responded whether the previously corresponding dot was 'on' or 'off' of the shape. During coordinate memory retrieval, each shape was presented centrally and participants responded whether the previously corresponding dot was 'near' or 'far' from the shape (relative to a reference distance). Consistent with our hypothesis, a region in the left prefrontal cortex (BA10) was preferentially associated with categorical visual spatial memory and a region in the right prefrontal cortex (BA9/10) was preferentially associated with coordinate visual spatial memory. These results have direct implications for interpreting previous findings that the left prefrontal cortex is associated with source memory, as this cognitive process is categorical in nature, and the right prefrontal cortex is associated with item memory, as this process depends on the precise spatial relations among item features or components.     

An electrophysiological study on the interaction between emotional content and spatial frequency of visual stimuli

Previous studies suggest that the magnocellular pathway, a visual processing system that rapidly provides low spatial frequency information to fast-responding structures such as the amygdala, is more involved in the processing of emotional facial expressions than the parvocellular pathway (which conveys all spatial frequencies). The present experiment explored the spatio-temporal characteristics of the spatial frequency modulation of affect-related neural processing, as well as its generalizability to non-facial stimuli. To that aim, the event-related potentials (ERPs) elicited by low-pass filtered (i.e., high spatial frequencies are eliminated) and intact non-facial emotional images were recorded from 31 participants using a 60-electrode array. The earliest significant effect of spatial frequency was observed at 135 ms from stimulus onset: N135 component of the ERPs. In line with previous studies, the origin of N135 was localized at secondary visual areas for low-pass filtered stimuli and at primary areas for intact stimuli. Importantly, this component showed an interaction between spatial frequency and emotional content: within low-pass filtered pictures, negative stimuli elicited the highest N135 amplitudes. By contrast, within intact stimuli, neutral pictures were those eliciting the highest amplitudes. These results suggest that high spatial frequencies are not essential for the initial affect-related processing of visual stimuli, which would mainly rely on low spatial frequency visual information. According to present data, high spatial frequencies would come into play later on.     

Visual spatial and visual pattern working memory: neuropsychological evidence for a differential role of left and right dorsal visual brain

According to neurophysiological, neuroimaging, and behavioural evidence, visual working memory (WM) can be separated into a "what" and a "where" component, reflecting the duality of visual processing. Whereas a wealth of empirical data suggests a right-sided fronto-parietal network critical for the maintenance of spatial information, the cortical structures underlying maintenance of object information have remained controversial. Although visual object processing depends on ventral, inferior temporal areas, recent neuroimaging results suggest that maintenance of visual object information involves a left-sided or bilateral fronto-parietal network. The aim of the present study is to further clarify the role of the left and right parietal lobes for pattern and spatial visual WM. Seven patients with left-sided, seven with right-sided parietal brain injury, and two age-matched healthy control groups performed a delayed-matching-to-sample task using either pattern (shape) or spatial (location) information or both. In addition, eight patients with left-sided injury sparing parietal areas were tested to further examine the specific role of the left parietal cortex in pattern WM. Left parietal injury resulted in pattern WM impairment, only, while right parietal injury was associated with pattern and spatial WM deficits. Non-parietal injury was not associated with comparable deficits. These results suggest that visual spatial WM depends critically on right parietal areas; in contrast, pattern WM depends on both, left and right parietal areas.     

Impaired object identification in idiopathic childhood occipital epilepsy

Purpose:: To investigate whether children with epilepsy primarily affecting the occipital cortex exhibit impairment of visual object identification and to what extent such a hypothesized dysfunction is related to an interfering functional, rather than structural, process. Method:: We studied nine children with idiopathic childhood occipital epilepsy (ICOE) and compared them to eight children with lesional posterior cortex epilepsy (PCEs) and to 60 age‐matched controls. We applied an “ascendent” paradigm of object identification, using a coarse‐to‐fine order procedure, which gradually integrated spatial frequency information from the most blurred image to the complete figure. In children with ICOE, we explored how epilepsy‐related variables might be related to object identification task. Key Findings:: Children with ICOE and those with PCEs needed more physical information than controls to identify visual stimuli. There was a decreasing accuracy from controls to children with ICOE and from children with ICOE to those with PCEs. Children with ICOE demonstrated slight selective impairment in visuospatial processing and those among them having experienced a higher number of seizures or in whom interictal electroencephalography (EEG) discharges had been present for a longer time, required a higher level of physical information to recognize objects. Significance:: The observation that children with ICOE performed worse than controls in object identification, although better than children with PCEs, might indicate that functional disruption caused by epileptiform EEG abnormalities and seizures, can interfere per se with perceptual processes, even in the absence of a lesion. This effect appears to be detected only by perceptual and cognitive screening.     

Visually induced signal-locked neuronal activity changes in precentral motor areas of the monkey: hierarchical progression of signal processing

In the precuing paradigm, two successive visual signals were presented to trained monkeys. The first one, the preparatory signal, provided complete, partial or no prior information about parameters, such as direction and extent of the forthcoming wrist movement. After a delay, the illumination of a second visual signal, the response signal, called for execution of the movement and indicated the target. Signal-locked neuronal activity changes, i.e. those which occured time-locked to the signal onset, were recorded in the premotor cortex and the primary motor cortex of the monkey and classified as selective or non-selective. Selective neurons were defined as those responding to particular information, for instance information about movement direction, provided by the signal, while non-selective neurons responded to all signals irrespective of any contained information. Clear latency differences according to both the selectivity of the neuronal response and the area in which the neuron was recorded could be discerned. The mean latency of non-selective activity changes was significantly shorter than that of selective activity changes. Furthermore, the mean latency of premotor cortical responses was significantly shorter than that of primary motor cortical responses. The data indicate the existence of distinct levels of signal processing from the very general to the highly specific.     

Facilitation of bottom-up feature detection following rTMS-interference of the right parietal cortex

In visual search tasks the optimal strategy should utilize relevant information ignoring irrelevant one. When the information at the feature and object levels are in conflict, un-necessary processing at higher level of object shape can interfere with detection of lower level orientation feature.We explored the effects of inhibitory trains of transcranial magnetic stimulation (rTMS) on the right and left parietal cortex in healthy subjects performing two visual search tasks. One task (Task A) was characterised by an object-to-feature interference. The other task (Task B) was without such interference. We found that rTMS of the right parietal cortex significantly reduced reaction times (RTs) in Task A, where object recognition interferes with detection of orientation. This significant RT reduction was present only for the first 10 trials. Interestingly, right parietal rTMS had no effect on Task B. Moreover, rTMS of the left parietal cortex did not modify subjects’ RTs in either task. Subjects’ accuracy was equally affected by rTMS in both tasks over time.We suggest that inhibition of the right parietal cortex by means of rTMS facilitates feature-based visual search by inhibiting the interfering feature binding and spatial attentional processes. This allows subjects to accomplish Task A faster.     

Altered figure-ground perception in monkeys with an extra-striate lesion

The visual system binds and segments the elements of an image into coherent objects and their surroundings. Recent findings demonstrate that primary visual cortex is involved in this process of figure-ground organization. In the primary visual cortex the late part of a neural response to a stimulus correlates with figure-ground segregation and perception. Such a late onset indicates an involvement of feedback projections from higher visual areas. To investigate the possible role of feedback in figure-ground perception we removed dorsal extra-striate areas of the monkey visual cortex. The findings show that figure-ground perception is reduced when the figure is presented in the lesioned hemifield and perception is normal when the figure appeared in the intact hemifield. In conclusion, our observations show the importance for recurrent processing in visual perception.