BackgroundStroke rehabilitation with different exercise paradigms has been investigated, but which one is more effective in facilitating motor recovery and up-regulating brain neurotrophic factor (BDNF) after brain ischemia would be interesting to clinicians and patients. Voluntary exercise, forced exercise, and involuntary muscle movement caused by functional electrical stimulation (FES) have been individually demonstrated effective as stroke rehabilitation intervention. The aim of this study was to investigate the effects of these three common interventions on brain BDNF changes and motor recovery levels using a rat ischemic stroke model.Methodology/Principal FindingsOne hundred and seventeen Sprague-Dawley rats were randomly distributed into four groups: Control (Con), Voluntary exercise of wheel running (V-Ex), Forced exercise of treadmill running (F-Ex), and Involuntary exercise of FES (I-Ex) with implanted electrodes placed in two hind limb muscles on the affected side to mimic gait-like walking pattern during stimulation. Ischemic stroke was induced in all rats with the middle cerebral artery occlusion/reperfusion model and fifty-seven rats had motor deficits after stroke. Twenty-four hours after reperfusion, rats were arranged to their intervention programs. De Ryck's behavioral test was conducted daily during the 7-day intervention as an evaluation tool of motor recovery. Serum corticosterone concentration and BDNF levels in the hippocampus, striatum, and cortex were measured after the rats were sacrificed. V-Ex had significantly better motor recovery in the behavioral test. V-Ex also had significantly higher hippocampal BDNF concentration than F-Ex and Con. F-Ex had significantly higher serum corticosterone level than other groups.Conclusion/SignificanceVoluntary exercise is the most effective intervention in upregulating the hippocampal BDNF level, and facilitating motor recovery. Rats that exercised voluntarily also showed less corticosterone stress response than other groups. The results also suggested that the forced exercise group was the least preferred intervention with high stress, low brain BDNF levels and less motor recovery.

This paper reviews the second challenge on spectral reconstruction from RGB images, i.e., the recovery of wholescene hyperspectral (HS) information from a 3-channel RGB image. As in the previous challenge, two tracks were provided: (i) a "Clean" track where HS images are estimated from noise-free RGBs, the RGB images are themselves calculated numerically using the ground-truth HS images and supplied spectral sensitivity functions (ii) a "Real World" track, simulating capture by an uncalibrated and unknown camera, where the HS images are recovered from noisy JPEG-compressed RGB images. A new, larger-than-ever, natural hyperspectral image data set is presented, containing a total of 510 HS images. The Clean and Real World tracks had 103 and 78 registered participants respectively, with 14 teams competing in the final testing phase. A description of the proposed methods, alongside their challenge scores and an extensive evaluation of top performing methods is also provided. They gauge the state-of-the-art in spectral reconstruction from an RGB image. arXiv:2005.03412v1 [eess.IV] 7 May 2020

Physical training is necessary for effective rehabilitation in the early poststroke period. Animal studies commonly use fixed training intensity throughout rehabilitation and without adapting it to the animals' recovered motor ability. This study investigated the correlation between training intensity and rehabilitation efficacy by using a focal ischemic stroke rat model. Eighty male Sprague-Dawley rats were induced with middle cerebral artery occlusion/reperfusion surgery. Sixty rats with successful stroke were then randomly assigned into four groups: control (CG, n = 15), low intensity (LG, n = 15), gradually increased intensity (GIG, n = 15), and high intensity (HG, n = 15). Behavioral tests were conducted daily to evaluate motor function recovery. Stress level and neural recovery were evaluated via plasma corticosterone and brain-derived neurotrophic factor (BDNF) concentration, respectively. GIG rats significantly (P < 0.05) recovered motor function and produced higher hippocampal BDNF (112.87 ± 25.18 ng/g). GIG and LG rats exhibited similar stress levels (540.63 ± 117.40 nM/L and 508.07 ± 161.30 nM/L, resp.), which were significantly lower (P < 0.05) than that (716.90 ± 156.48 nM/L) of HG rats. Training with gradually increased intensity achieved better recovery with lower stress. Our observations indicate that a training protocol that includes gradually increasing training intensity should be considered in both animal and clinical studies for better stroke recovery.

No abstract

No abstract

Monitoring the neural activities from the ischemic penumbra provides critical information on neurological recovery after stroke. The purpose of this study is to evaluate the temporal alterations of neural activities using electroencephalography (EEG) from the acute phase to the chronic phase, and to compare EEG with the degree of post-stroke motor function recovery in a rat model of focal ischemic stroke. Male Sprague-Dawley rats were subjected to 90 min transient middle cerebral artery occlusion surgery followed by reperfusion for seven days (n = 58). The EEG signals were recorded at the pre-stroke phase (0 h), acute phase (3, 6 h), subacute phase (12, 24, 48, 72 h) and chronic phase (96, 120, 144, 168 h) (n = 8). This study analyzed post-stroke seizures and polymorphic delta activities (PDAs) and calculated quantitative EEG parameters such as the alpha-to-delta ratio (ADR). The ADR represented the ratio between alpha power and delta power, which indicated how fast the EEG activities were. Forelimb and hindlimb motor functions were measured by De Ryck's test and the beam walking test, respectively. In the acute phase, delta power increased fourfold with the occurrence of PDAs, and the histological staining showed that the infarct was limited to the striatum and secondary sensory cortex. In the subacute phase, the alpha power reduced to 50% of the baseline, and the infarct progressed to the forelimb cortical region. ADRs reduced from 0.23 ± 0.09 to 0.04 ± 0.01 at 3 h in the acute phase and gradually recovered to 0.22 ± 0.08 at 168 h in the chronic phase. In the comparison of correlations between the EEG parameters and the limb motor function from the acute phase to the chronic phase, ADRs were found to have the highest correlation coefficients with the beam walking test (r = 0.9524, p < 0.05) and De Ryck's test (r = 0.8077, p < 0.05). This study measured EEG activities after focal cerebral ischemia and showed that functional recovery was closely correlated with the neural activities in the penumbra. Longitudinal EEG monitoring at different phases after a stroke can provide information on the neural activities, which are well correlated with the motor function recovery.

Convolutional neural networks (CNNs) have been widely applied in hyperspectral imagery (HSI) classification. However, their classification performance might be limited by the scarcity of labeled data to be used for training and validation. In this paper, we propose a novel lightweight shuffled group convolutional neural network (abbreviated as SG-CNN) to achieve efficient training with a limited training dataset in HSI classification. SG-CNN consists of SG conv units that employ conventional and atrous convolution in different groups, followed by channel shuffle operation and shortcut connection. In this way, SG-CNNs have less trainable parameters, whilst they can still be accurately and efficiently trained with fewer labeled samples. Transfer learning between different HSI datasets is also applied on the SG-CNN to further improve the classification accuracy. To evaluate the effectiveness of SG-CNNs for HSI classification, experiments have been conducted on three public HSI datasets pretrained on HSIs from different sensors. SG-CNNs with different levels of complexity were tested, and their classification results were compared with fine-tuned ShuffleNet2, ResNeXt, and their original counterparts. The experimental results demonstrate that SG-CNNs can achieve competitive classification performance when the amount of labeled data for training is poor, as well as efficiently providing satisfying classification results.

Jointly using spatial and spectral information has been widely applied to hyperspectral image (HSI) classification. Especially, convolutional neural networks (CNN) have gained attention in recent years due to their detailed representation of features. However, most of CNN-based HSI classification methods mainly use patches as input classifier. This limits the range of use for spatial neighbor information and reduces processing efficiency in training and testing. To overcome this problem, we propose an image-based classification framework that is efficient and straightforward. Based on this framework, we propose a multiscale spatial-spectral CNN for HSIs (HyMSCN) to integrate both multiple receptive fields fused features and multiscale spatial features at different levels. The fused features are exploited using a lightweight block called the multiple receptive field feature block (MRFF), which contains various types of dilation convolution. By fusing multiple receptive field features and multiscale spatial features, the HyMSCN has comprehensive feature representation for classification. Experimental results from three real hyperspectral images prove the efficiency of the proposed framework. The proposed method also achieves superior performance for HSI classification.