基于皮质慢电位特征分析的神经皮质运动区功能定位

背景：皮质慢电位及其偏移变化在所有个体中都存在,基于皮质慢电位及其变化规律检测的神经皮质(运动区)功能定位方法可以有效避免漏检,关于此方面的详细研究鲜有报道。 目的：探讨皮质脑电中皮质慢电位用于术中神经皮质(运动区)功能定位的可行性和特点。 方法：采集华盛顿西雅图海港医院3例患者位于大脑神经皮质(运动)功能区手指皮质区域的皮质脑电信号数据,同时采集相应手指弯曲运动数据,作自身对照。利用小波变换对皮质脑电信号进行分解和重构,提取运动事件相关皮质慢电位在运动事件发生前后的能量比(事件相关电位指标)为特征量,并构造特定阈值进行分类,结果与相应手指弯曲运动数据比较,进行检测正确率分析。将试验采集数据分成训练和测试组,分别用于特征提取和分类算法的设计和性能检测。 结果与结论：以皮质慢电位信号的事件相关电位指标为特征量,以1.6为阈值进行分类,其分类定位检出正确率达到84%。提示通过皮质(运动区)慢电位的特征提取和分类可以更有效地进行术中运动功能区皮质定位,具有检测分辨率高、避免漏检的优点。     

U2F-GAN: Weakly Supervised Super-pixel Segmentation in Thyroid Ultrasound Images

Precise nodule segmentation in thyroid ultrasound images is important for clinical quantitative analysis and diagnosis. Fully supervised deep learning method can effectively extract representative features from nodules and background. Despite the great success, deep learning–based segmentation methods still face a critical hindrance: the difficulty in acquiring sufficient training data due to high annotation costs. To this end, we propose a weakly supervised framework called uncertainty to fine generative adversarial network (U2F-GAN) for nodule segmentation in thyroid ultrasound images that exploits only a handful of rough bounding box annotations to successfully generate reliable labels from these weak supervisions. Based on feature-matching GAN, the proposed method alternates between generating masks and learning a segmentation network in an adversarial manner. Super-pixel processing mechanism is adopted to reflect low-level image structure features for learning and inferring semantic segmentation, which largely improve the efficiency of training process. In addition, we introduce a similarity comparison module and a distributed loss function with constraints to effectively remove noise in localization annotations and enhance the generalization capability of the network, thus strengthen the overall segmentation performance. Compared to existing weakly supervised approaches, our proposed U2F-GAN yields a significant performance boost. The segmentation results are also comparable to fully supervised methods, but the annotation burden is much lower. Also, the training speed of the network model is much faster than other methods with weak supervisions, which enables the network to be updated in time, thus is beneficial to high-throughput medical image setting.

     

A novel tracing algorithm for high throughput imaging Screening of neuron-based assays

High throughput neuron image processing is an important method for drug screening and quantitative neurobiological studies. The method usually includes detection of neurite structures, feature extraction, quantification, and statistical analysis. In this paper, we present a new algorithm for fast and automatic extraction of neurite structures in microscopy neuron images. The algorithm is based on novel methods for soma segmentation, seed point detection, recursive center-line detection, and 2D curve smoothing. The algorithm is fully automatic without any human interaction, and robust enough for usage on images with poor quality, such as those with low contrast or low signal-to-noise ratio. It is able to completely and accurately extract neurite segments in neuron images with highly complicated neurite structures. Robustness comes from the use of 2D smoothening techniques and the idea of center-line extraction by estimating the surrounding edges. Efficiency is achieved by processing only pixels that are close enough to the line structures, and by carefully chosen stopping conditions. These make the proposed approach suitable for demanding image processing tasks in high throughput screening of neuron-based assays. Detailed results on experimental validation of the proposed method and on its comparative performance with other proposed schemes are included.     

Hippocampal dysfunction and behavioral deficit in the water maze in mice: an unresolved issue?

Dysfunction of the hippocampal formation manifests as impaired relational learning and memory in humans and animals. One of the most frequently applied relational learning paradigms in animals is the Morris water maze (MWM), in which the subject is required to learn complex spatial relationships of visual cues. MWM has been employed as a diagnostic tool to investigate effects of drugs and mutations. However, the validity of this test and its ability to properly detect hippocampal dysfunction have been questioned. In order to corroborate the role of hippocampus in spatial learning, we employed ibotenic acid lesioning and ablated the hippocampus bilaterally or unilaterally in mice, as ascertained by magnetic resonance imaging. We found a significant impairment in response to hippocampal disruption that was more pronounced in mice with bilateral lesion than with unilateral lesion. However, the results also indicated that even the mice with bilateral lesion could improve their performance, which confirms the notion that the MWM has an important non-hippocampal component. It is thus possible that experimental alteration of brain function does not manifest as modified performance in MWM, even when hippocampal function is modified (false-negative finding), or manifest as altered performance without varying hippocampal function (false-positive finding), possibilities that have important implications for studies using genetic and pharmacological manipulation of the brain.     

Study of Haralick’s and GLCM texture analysis on 3D medical images

Purpose of the study: Medical field has highly evolved with advancements in the technologies which prove to be beneficial for radiologists and patients for better diagnosis. The era of medical science provides best healthcare solutions with the help of medical images. Till now, 2D MRIs played a prominent role in early detection of disease but with latest technologies taking over the charge, 3D MRIs are highly effective and great in demand nowadays. With the aid of advanced techniques such as edge detection, segmentation and texture analysis on these images, the disease detection may become much easier.

Materials and Methods: Texture of any image is recognized by distribution of gray levels in the neighborhood. The Texture Analysis plays an important role in study of medical images. It identifies the prominent features of an image and highlights the same using different feature extraction technique. In this paper, 3D MRI of human brain is considered and texture analysis based on Haralick's and GLCM texture features is performed. Haralick's feature explains the image intensities of each pixel and their relationship with neighborhood pixels. The entire data set consists of 40 brain tumor patients, out of which a sample has been depicted.

Results: The analysis of different features such as Contrast, Correlation, Energy, Homogeneity and Entropy is carried out. Conclusion: Further, the study highlights about the highly useful features for early detection of brain tumor disease.     

Fast Image Recognition Based on Independent Component Analysis and Extreme Learning Machine

Nowadays, image recognition has become a highly active research topic in cognitive computation community, due to its many potential applications. Generally, the image recognition task involves two subtasks: image representation and image classification. Most feature extraction approaches for image representation developed so far regard independent component analysis (ICA) as one of the essential means. However, ICA has been hampered by its extremely expensive computational cost in real-time implementation. To address this problem, a fast cognitive computational scheme for image recognition is presented in this paper, which combines ICA and the extreme learning machine (ELM) algorithm. It tries to solve the image recognition problem at a much faster speed by using ELM not only in image classification but also in feature extraction for image representation. As an example, our proposed approach is applied to the face image recognition with detailed analysis. Firstly, common feature hypothesis is introduced to extract the common visual features from universal images by the traditional ICA model in the offline recognition process, and then ELM is used to simulate ICA for the purpose of facial feature extraction in the online recognition process. Lastly, the resulting independent feature representation of the face images extracted by ELM rather than ICA will be fed into the ELM classifier, which is composed of numerous single hidden layer feed-forward networks. Experimental results on Yale face database and MNIST digit database have shown the good performance of our proposed approach, which could be comparable to the state-of-the-art techniques at a much faster speed.     

Automated correction of fast motion artifacts for two-photon imaging of awake animals

Two-photon imaging of bulk-loaded calcium dyes can record action potentials (APs) simultaneously from dozens of spatially resolved neurons in vivo. Extending this technique to awake animals, however, has remained technically challenging due to artifacts caused by brain motion. Since in two-photon excitation microscopes image pixels are captured sequentially by scanning a focused pulsed laser across small areas of interest within the brain, fast displacements of the imaged area can distort the image nonuniformly. If left uncorrected, brain motion in awake animals will cause artifactual fluorescence changes, masking the small functional fluorescence increases associated with AP discharge. We therefore present a procedure for detection and correction of both fast and slow displacements in two-photon imaging of awake animals. Our algorithm, based on the Lucas-Kanade framework, operates directly on the motion-distorted imaging data, requiring neither external signals such as heartbeat nor a distortion-free template image. Motion correction accuracy was tested in silico over a wide range of simplified and realistic displacement trajectories and for multiple levels of fluorescence noise. Accuracy was confirmed in vivo by comparing solutions obtained from red and green fluorophores imaged simultaneously. Finally, the accuracy of AP detection from motion-displaced bulk-loaded calcium imaging is evaluated with and without motion correction, and we conclude that accurate motion correction as achieved by this procedure is both necessary and sufficient for single AP detection in awake animals.     

人工智能影像技术在卒中诊疗中的应用

多模态影像是全面评估缺血性卒中发生发展的重要手段,其数据信息较多,且分析难度 较大,传统的计算机视觉方法依赖于手工提取特征,在复杂任务上性能有限且通用性不佳。人工智 能影像技术主要指利用人工智能方法处理计算机视觉任务,包括图像分类,病灶定位、检测、分割等, 可深入挖掘多维影像学信息,并综合其他临床资料,已在卒中的早期筛查、病灶识别、病情诊断和预 后预测等方面展开了广泛而深入的研究工作。人工智能影像技术在快速精准的影像学分析及标准化 诊疗辅助方面具有一定应用价值,但在临床价值验证和产品转化方面仍存在不足,且其在临床实践 中的应用发展亦面临诸多挑战。     

A feature extraction method for adaptive DBS using an improved EMD

Objective: Local field potential (LFP) of a patient with Parkinson's disease often shows abnormal oscillation phenomenon. Extracting and studying this phenomenon and designing adaptive deep brain stimulation (DBS) control library have great significance in the treatment of disease.

Materials and methods: This paper has designed a feature extraction method based on modified empirical mode decomposition (EMD) which extracts the abnormal oscillation signal in the time domain to increase the overall performance. The intrinsic mode function (IMF) component which contains abnormal oscillation is extracted by using EMD before an intrinsic characteristic of the oscillation signal is obtained. Abnormal oscillation signal is acquired using signal normalization, peak counting, and envelope method with a threshold which in turn keeps the integrity and accuracy as well as the efficiency.

Results: Comparative study of eight patients (six patients with DBS closed and drugs stopped; two patients with stimulation) has verified the feasibility of using modified EMD in extracting abnormal oscillation signal. The results showed that patients who take DBS suffer less abnormal oscillation than those who take no treatment. These results match the energy rise in the band of 3–30 Hz on local field potential spectrum of the patient with Parkinson's disease.

Conclusions: Unlike previous oscillation extraction algorithm, improved EMD feature extraction method directly isolates abnormal oscillation signal from LFP. Significant improvement has been made in feature extraction algorithm in adaptability, real–time performance, and accuracy.     

Giant axonal neuropathy: observations on a further patient.

A further child with giant axonal neuropathy (GAN), abnormally curly hair and consanguineous parents is described. Of the 19 patients with GAN so far reported in the literature, six, including the present patient, have resulted from consanguineous marriages. This makes autosomal recessive inheritance of GAN highly probable. Our patient also exhibited cerebellar ataxia and signs of pyramidal tract damage; magnetic resonance brain imaging demonstrated abnormalities within the cerebellar and cerebral white matter. Myelinated nerve fibre density in the sural nerve was reduced to 6790/mm2 at age 8 years and had fallen to 3812/mm2 16 months later, indicating that progressive axonal loss occurs in GAN.     

Improved Reversible Image Authentication Scheme

Image integrity authentication has aroused concerns because of the frequent modification on images. However, most of the image authentication schemes proposed so far employed the irreversible data hiding approach and the results of the published few reversible authentication methods are not satisfactory. To improve the detection accuracy as well as marked image quality, this paper proposes an improved reversible image authentication method based on Hilbert Curve mapping. In the proposed method, pixels are first mapped to a one-dimensional vector by using Hilbert Curve and divided into non-overlapping sets. Then, authentication codes can be embedded into each set by reversible data hiding approach. After comparing the extracted bits with the original authentication codes, the image set could be taken as modified one or unmodified one. Because image redundancy can be explored more fully and more flexibly by adopting Hilbert Curve mapping, more authentication codes can be embedded into the host image while leaving less distortion. Thus, both the detection accuracy and the marked image quality can be improved. The experimental results demonstrate the improvement compared with the latest development of reversible image authentication.     

Effects of different nerve autografts on greater auricular nerve deficit in rabbits

BACKGROUND: Autograft is commonly used to repair nerve deficit. Usually, the choice of donor nerves is based on their similarities in form and structures to the injured nerves. For the reason, the cutaneous antebrachii lateralis nerve is currently considered the most suited for digital nerve repair. OBJECTIVE: To compare early nerve regeneration after transplantation of three different autografts: the greater auricular nerve (GAN), the saphenous nerve (SN) and the lateral femoral cutaneous nerve (LFCN). DESIGN: Observational contrast study. SETTING: Department of Plastic Surgery and Burns, Tangdu Hospital, Fourth Military Medical University of Chinese PLA. MATERIALS: A total of 42 New Zealand rabbits, of both genders, 12–14 months old and weighing 2.0–2.5 kg, were used in this study. In addition, Moller-spetra 900 operating microscope (Germany), Olympus BX 51 microscope, DP 70 image collecting System (Japan), BL-420E+ Biologic function testing System (China), JEM-100 electron microscope (Japan), Reichet-JunG820 Cryostat (Swiss), and Libror-AEG-120 precision analytical Balance (Japan) were also used in this study. METHODS: The experiment was carried out in the Department of Plastic Surgery and Burns, Tangdu Hospital, Fourth Military Medical University of Chinese PLA from April to November 2005. After anaesthesia, the GAN were dissected bilaterally and a 1.2 cm deficit was made in each nerve. The animals were randomly divided into three groups, including GAN group, SN group and LFCN group with 14 in each group. ① Nerve pinch test: At 1, 2, and 4 weeks after operation, three animals in each group were tested. The nerve grafts, along with the proximal and distal GAN segments were exposed and pinched with microsurgical forceps in distal-proximal orientations. The distance between the proximal anastomosis site and the most distal point, where the pinch evoked an ear contraction response, was measured as distance of nerve regeneration. ② Computer image analysis: At 4 and 12 weeks, 2 μm sections were prepared, each stained with either HE or methylene blue to assess axon number and density, cross-section area, and myelin sheath thickness. ③ Electrophysidogical tests: At 12 weeks, the bilateral GAN along with the nerve grafts of 4 animals in each group were exposed. Points A, B and C were marked on each specimen: point A: at the proximal GAN segment, 7 cm from the proximal anastomosis; point B: 0.5 cm from the proximal anastomosis; point C: at the distal GAN segment, 0.5 cm from the distal anastomosis. The whole nerve including nerve graft and proximal and distal GAN segments, as a block, was harvested and immersed in Ren's solution for several minutes until its excitability was stabilized. The specimen was then placed on the electrodes of the shield box to examine the action potential and conduction velocity on segment AB and AC with BL-420E+biologic function testing system. AC/AB would be the recovery rate of action potential on segment AC. ④ Horseradish peroxidase (HRP) fascicle: At 12 weeks, at the site on the distal segment of GAN 1.0 cm from the distal anastomosis of nerve graft, the GAN was crushed by a pair of haemostatic forceps and HRP water solution was injected into the nerve. Two rabbits in GAN group, SN group and LFCN group, after having survived for 24 hours, 36 hours and 48 hours were selected. The C2 ganglion was exposed and the distance from C2 ganglion to HRP injection site was taken as the axoplasmic transport distance, from which the axoplasmic transport velocity and the mean density of the labeled C2 ganglion cells were calculated. MAIN OUTCOME MEASURES: ① The greatest distance of nerve regeneration; ② the axon number and density, cross-section area, and myelin sheath thickness; ③ the action potential and conduction velocity; ④ the axoplasmic transport velocity and the mean density of the labeled C2 ganglion cells. RESULTS: All 42 experimental rabbits were involved in the final analysis. ① The greatest distance of nerve regeneration: At 4 weeks after operation, the greatest distance of nerve regeneration was longer in the SN group than that in the GAN group and the LFCN group [(45.17±2.48), (41.83±2.32), (34.83±2.64) mm, P < 0.05], while the greatest distance of nerve regeneration was longer in the GAN group than that in the LFCN group (P < 0.05). ② The axon number and density, cross-section area, and myelin sheath thickness: The number of nerve fascicle was the greatest in the GAN group, and the cross-section area was the most; however, ratio between nerve fascicle and cross-section area, and the axon density were lower than those in other two groups (P < 0.05–0.01). In contrast, the axon density was the greatest in the SN group. At 4 weeks after operation, axon density was the most in the SN group, and then the GAN group and the LFCN group. There were significant differences among the three groups (P < 0.05–0.01). At 12 weeks after operation, density of myelinated fiber and axon section area were higher in the SN group than those in other two groups (P < 0.05–0.01). ③ The action potential and conduction velocity: At 12 weeks after operation, the maximal action potential, the recovery rate of action potential and the nerve conduction velocity were the highest in the SN group. HRP-labeled neurons early occurred in C2 ganglion, and the action potential and the recovery rate of action potential were increased (P > 0.05). At 12 weeks after operation, even though the maximal action potential, the recovery rate of action potential and the nerve conduction velocity on segment AB remained similar in different groups, on segment AC, the action potential, the recovery rate of action potential and nerve conduction velocity were greater in the SN group than those in other groups. ④ The axoplasmic transport velocity and the mean density of the labeled C2 ganglion cells: After HRP injection in the SN group, the positive labeled cells in C2 ganglion firstly appeared at 24 hours, and in other two groups, they did not appeared until 36 hours. The density of labeled cells was the greatest in the SN group and the lowest in the LFCN group. The axoplasmic transport velocity in the SN group was also significantly faster than in the GAN group and the LFCN group (P < 0.05–0.01). Otherwise, the axoplasmic transport velocity was faster in the SN group than that in the GAN group and the LFCN group. CONCLUSION: The donor nerve with greater axon number and density can achieve much better effects during early regeneration.     

基于图割与粗糙集的MRI脑部肿瘤图像检索方法

背景：基于内容的医学图像检索是一门涉及多领域的学科,由于各种医学图像的成像原理不同,产生的图像在颜色、纹理和形状等视觉特征方面存在差别,使得此方法的实现还存在许多需要解决的问题。 目的：针对基于内容的医学图像检索中存在特征提取困难、检索时间长的问题,提出一种基于图割与粗糙集结合的相似图像检索方法。 方法：为克服图割仅适用于较少象素的图像和倾向于小割集的缺陷,首先对图像进行聚类,然后构建图像的Gomory-Hu割树,按割值大小依次去掉值较小的边,提取出图像的特征子图并构建特征库。为实现快速检索,借助粗糙集对特征库中的特征进行约简,有效减少参与相似性比较的特征数量。并将此方法应用到MRI脑部肿瘤图像的检索。 结果与结论：实验结果表明该方法能快速有效地检索出MRI脑部图像库中的肿瘤图像,检索的平均查准率为78.4%,平均查全率为62.9%。     

A Novel Multiple Feature-Based Engine Knock Detection System using Sparse Bayesian Extreme Learning Machine

Automotive engine knock is an abnormal combustion phenomenon that affects engine performance and lifetime expectancy, but it is difficult to detect. Collecting engine vibration signals from an engine cylinder block is an effective way to detect engine knock. This paper proposes an intelligent engine knock detection system based on engine vibration signals. First, filtered signals are obtained by utilizing variational mode decomposition (VMD), which decomposes the original time domain signals into a series of intrinsic mode functions (IMFs). Moreover, the values of the balancing parameter and the number of IMF modes are optimized using genetic algorithm (GA). IMFs with sample entropy higher than the mean are then selected as sensitive subcomponents for signal reconstruction and subsequently removed. A multiple feature learning approach that considers time domain statistical analysis (TDSA), multi-fractal detrended fluctuation analysis (MFDFA) and alpha stable distribution (ASD) simultaneously, is utilized to extract features from the denoised signals. Finally, the extracted features are trained by sparse Bayesian extreme learning machine (SBELM) to overcome the sensitivity of hyperparameters in conventional machine learning algorithms. A test rig is designed to collect the raw engine data. Compared with other technology combinations, the optimal scheme from signal processing to feature classification is obtained, and the classification accuracy of the proposed integrated engine knock detection method can achieve 98.27%. We successfully propose and test a universal intelligence solution for the detection task.

     

基于LabVIEW运动想象的脑机接口系统***◆

背景：基于事件相关电位的脑-机接口,可广泛应用于残障患者的康复,显示出其重要性和未来实现的可行性。 目的：提出基于LabVIEW环境下的运动想象脑-机接口系统的实现方案。 方法：研究的关键部分是视觉刺激器的设计和脑电特征信号的特征提取两部分。测试者通过观察视觉刺激器上的左右手连续播放图像刺激产生脑电信号,采用带通滤波提高信噪比,用滑动窗截取脑电数据并且对截取的数据从能量的角度分析得到运动想象特征,同时可以在线提取特征,为实现实时系统打下了基础。 结果与结论：该方案能有效地提取出运动想象特征,并且通过离线模式识别进行了有效的分类,分类效果达到了82%。     

Mel-Frequency Cepstral Coefficient Features Based on Standard Deviation and Principal Component Analysis for Language Identification Systems

Spoken language identification (LID) is the process of determining and classifying natural language from a given content and dataset. Data must be processed to extract useful features to perform LID. The mel-frequency cepstral coefficient (MFCC) is one of the most popular feature extraction techniques in LID. The MFCC features are generated to serve as inputs for the classification stage. In this study, reduction in the MFCC feature dimension is investigated because large data size affects the computational time and resources (i.e., memory space) and slows the identification speed. The implementation of data reduction techniques to retain the most important feature parameters is also evaluated in this study. The investigation of data reduction is based on standard deviation (STD) calculation and principal component analysis (PCA). The features based on MFCC and the reduced dimensions based on STD and PCA results are then used as inputs to an optimized extreme learning machine (ELM) classifier called the optimized genetic algorithm-ELM (OGA-ELM). Several sets of data samples with one dimension of principal components (i.e., 119) are utilized for the evaluation. The results are generated using two different datasets. The first dataset is derived from eight separate languages, whereas the second dataset is a part of the National Institute of Standards and Technology Language Recognition Evaluation 2009 dataset. To evaluate the performance of the proposed method, this study utilizes several assessment measures, namely, accuracy, recall, precision, F-measure, G-mean, and identification time. The best LID performance is observed when the MFCC based on STD and PCA features with 119 feature dimensions is used with OGA-ELM as the classifier. The experimental results show that the proposed MFCC method achieves 99.38% accuracy using the first dataset. Additionally, it achieves accuracies of up to 97.60%, 96.80%, and 91.20% using the second dataset with durations of 30, 10, and 3 s, respectively. The proposed MFCC method exhibits the fastest computational time in all experiments, requiring only a few seconds to identify languages. Using a data reduction technique can substantially speed up the computational time, overcome resource limitations, and improve LID performance.

     

ML-CGAN: Conditional Generative Adversarial Network with a Meta-learner Structure for High-Quality Image Generation with Few Training Data

Since generative adversarial network (GAN) can learn data distribution and generate new samples based on the learned data distribution, it has become a research hotspot in the area of deep learning and cognitive computation. The learning of GAN heavily depends on a large set of training data. However, in many real-world applications, it is difficult to acquire a large number of data as needed.  In this paper, we propose a novel generative adversarial network called ML-CGAN for generating authentic and diverse images with few training data. Particularly, ML-CGAN consists of two modules: the conditional generative adversarial network (CGAN) backbone and the meta-learner structure. The CGAN backbone is applied to generate images, while the meta-learner structure is an auxiliary network to provide deconvolutional weights for the generator of the CGAN backbone.  Qualitative and quantitative experimental results on the MNIST, Fashion MNIST, CelebA and CIFAR-10 data sets demonstrate the superiority of ML-CGAN over state-of-the-art models. Specifically, the results show that the meta-learner structure can learn prior knowledge and transfer it to the new tasks, which is beneficial for generating authentic and diverse images in the new tasks with few training data.

     

Diagnosis of acute cerebral infarction using diffusion-weighted imaging by low field (0.2 T) magnetic resonance image]

The purpose of this study is to confirm the diagnosis of acute cerebral infarction on diffusion-weighted imaging using low field (0.2 T) magnetic resonance image(MRI). Acute cerebral infarctions in 51 patients were examined on diffusion-weighted imaging using low field MRI within 48 hours after clinical symptoms. Diffusion-weighted imaging was examined using line scan method. Twenty-four cases were cortical infarction, and twenty-two cases were perforating infarction. In five cases out of 51 cases, ischemic regions were not detected as abnormal high signal intensity area on diffusion-weighted imaging. Four cases of no abnormal detection were transient ischemic attack, and the other one was a perforating infarction. The earliest detection time in cortical infarction cases was 1 hour and 20 minutes. On the other hand, the earliest detection time in perforating infarction cases was 3 hours. Detective ability for acute cerebral infarction on diffusion-weighted imaging by low field MRI was depending on both size and lesion of infarction. That is to say, either small size or brain stem infarction was hard to detect. Thin slice and vertical slice examination for the infarction may improve to diagnose in low field MRI. Our conclusion is acute cerebral infarction was able to be diagnosed on diffusion-weighted imaging by low field as well as high field MRI.     

Real-Time Lane Detection by Using Biologically Inspired Attention Mechanism to Learn Contextual Information

Background State-of-the-art lane detection methods have achieved prominent performance in complex scenarios, but many limits have also existed. For example, only a fixed number of lanes can be detected, and the cost of detection time is unaffordable in many cases. Methods Inspired by human vision, attention mechanism makes network learning more concerned features. In this paper, we propose a real-time lane detection method by using attention mechanism. The network proposed consists of three modules: an encoder module that extracts the feature of lanes; the instance feature maps of lanes are predicted by two decoder modules, namely binary decoder and embeddable decoder. In the encoder, we use the biologically inspired attention to extract features, which contain many details of the target area. The correlation between the features obtained from the convolutions and that extracted by the attention is established to learn the contextual information. In the decoder, the contextual information is fused with the features from up-sampling, to compensate for the lost detailed information. Binary decoder classifies all the pixels into lane or background. Embeddable decoder obtains the distinguishable lanes. And then, the outputs of the binary decoder serve as one of the inputs to the embeddable decoder to guiding the generation of exact pixel points on the lanes. Results Comparative experiments on two benchmarks (TuSimple and Caltech lanes datasets) show that the proposed method is independent of lane number and lane pattern. It can handle an indefinite number of lanes and run at 10ms in the TuSimple dataset. Conclusions Experiments verify that our method outperforms a lot of state-of-the-art methods while maintaining a real-time performance.