There Is No Male and Female Brain Peer Review Study

Introduction

Recent studies signal that gender may accept a substantial influence on homo cognitive functions, including emotion, memory, perception, etc., (Cahill, 2006). Men and women announced to have dissimilar means to encode memories, sense emotions, recognize faces, solve sure issues, and brand decisions. Since the brain controls noesis and behaviors, these gender-related functional differences may exist associated with the gender-specific structure of the brain (Cosgrove et al., 2007).

Diffusion tensor imaging (DTI) is an effective tool for characterizing nervus fibers compages. By computing fractional anisotropy (FA) parameters in DTI, the anisotropy of nerve fibers can exist quantitatively evaluated (Lasi et al., 2014). Differences in FA values are thought to associate with developmental processes of axon caliber, myelination, and/or fiber organization of nervus fibers pathways. By calculating FA, researchers has revealed subtle changes related to normal encephalon evolution (Westlye et al., 2009), learning (Golestani et al., 2006), and healthy crumbling (Kochunov et al., 2007). Nevertheless, existing studies are however to provide consistent results on exploring the difference of brain structure between men and women. Ingalhalikar et al. (2014) argued that the men have greater intra-hemispheric connection via the corpus callosum while women take greater interhemispheric connectivity. However, other studies reported no significant gender departure in encephalon structure (Raz et al., 2001; Salat et al., 2005). A contempo critical stance commodity suggested that more research is needed to investigate whether men and women really accept unlike brain structures (Joel and Tarrasch, 2014).

Most existing DTI studies used the group-level statistical methods such equally Tract-Based Spatial Statistics (TBSS) (Thatcher et al., 2010; Mueller et al., 2011; Shiino et al., 2017). Still, recent studies indicated that machine learning techniques may provide us with a more than powerful tool for analyzing brain images (Shen et al., 2010; Lu et al., 2017; Tang et al., 2018). Especially, deep learning can extract not-linear network structure, realize approximation of complex office, narrate distributed representation of input information, and demonstrate the powerful ability to larn the essential features of datasets based on a small size of samples (Zeng et al., 2016, 2018a; Tian et al., 2018; Wen et al., 2018). In particular, the deep convolutional neural network (CNN) uses the convolution kernels to excerpt the features of image and can find the characteristic spatial divergence in brain images, which may promise a better result than using other conventional machine learning and statistical methods (Cole et al., 2017).

In this study, we performed CNN-based analyses on the FA images and extracts the features of the hidden layers to investigate the difference between man and adult female brains. Different from normally used 2D CNN model, we innovatively proposed a 3D CNN model with a new structure including 3 hidden layers, a linear layer and a softmax layer. Each hidden layer is comprised of a convolutional layer, a batch normalization layer, an activation layer and followed past a pooling layer. This novel CNN model allows using the whole 3D brain paradigm (i.e., DTI) every bit the input to the model. The linear layer between the subconscious layers and the softmax layer reduces the number of parameters and therefore avoids over-plumbing fixtures problems.

Materials and Methods

MRI Data Acquisition and Preprocessing

The database used in this piece of work is from the Human Connectome Projection (HCP) (Van Essen et al., 2013). This open-admission database contains information from 1,065 subjects, including 490 men and 575 women. The ages range is from 22 to 36. This database represents a relatively large sample size compared to most neuroimaging studies. Using this open-admission dataset allows replication and extension of this work by other researchers.

We performed DTI data preprocessing includes format conversion, b0 image extraction, brain extraction, eddy current correction, and tensor FA calculation. The outset four steps were processed with the HCP diffusion pipeline, including diffusion weighting (bvals), management (bvecs), time series, brain mask, a file (grad_dev.nii.gz) for gradient not-linearities during model fitting, and log files of Eddy processing. In the final step nosotros use dtifit to calculate the tensors to become the FA, as well as mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) values.

The original data were too big to train the model and it would cause Resources EXAUSTED trouble while training due to the insufficient of GPU retention. The GPU nosotros used in the experiment is NVIDIAN TITAN_XP with 12G retention each. To solve the problem, we scaled the size of FA image to [58 × lxx × 58]. This procedure may lead to a better classification outcome, since a smaller size of the input paradigm can provide a larger receptive field to the CNN model. In gild to perform the image scaling, "dipy" (http://nipy.org/dipy/) was used to read the .nii data of FA. Then "ndimage" in the SciPy (http://www.scipy.org) was used to reduce the size of the information. Scaled data was written into the TFRecord files (http://world wide web.tensorflow.org) with the corresponding labels. TFRecord file format is a simple record oriented binary format that is widely used in Tensorflow application for the training data to get a high operation of input efficiency. The labels were processed into the format of 1-hot. We implemented a pipeline to read information asynchronously from TFRecord according to the interface specification provided by Tensorflow (Abadi et al., 2016). The pipeline included the reading of TFRecord files, data decoding, data type conversion, and reshape of data.

CNN Model

We did the experiments on a GPU work station, which has four NVIDIA TITAN Xp GPUs. The operation system of the GPU piece of work station was Ubutnu16.04. We used FSL to preprocess the data. The CNN model was designed using the open source machine learning framework Tensorflow (Abadi et al., 2016).

Model Design

The unremarkably used CNN structures are based on 2d images. When using a 2d CNN to process 3D MRI images, it needs to map the original paradigm from dissimilar directions to get second images, which will lose the spatial structure information of the image. In this study, nosotros designed a 3D CNN with 3D convolutional kernels, which allowed the states to extract 3D structural features from FA images. Besides, traditional CNN model usually uses several fully connected layers to connect the hidden layers and the output layer. The fully connected layer may exist decumbent to the over-fitting trouble in binary classification when the number of samples is express (like our information). To address this problem, we used a linear layer to supplant the fully continued layer. The linear layer integrates the outputs of subconscious layers (i.e., a 3D matrix comprised of multiple featuremaps) into the inputs (i.east., a 1D vector) of the output layer which is a softmax classifier. Moreover, we performed a Batch Normalization (Ioffe and Szegedy, 2015) afterward each convolution operation. The Batch Normalization is used to avoid internal covariate shift problem in training the CNN model. Therefore, our designed model is a 3D "pure" CNN (3D PCNN). The architecture of the 3D PCNN model is shown in Effigy ane. The 3D PCNN consists of three subconscious layers, a linear layer and a softmax layer. Each of the hidden layer contains a convolutional layer, a Batch Normalization layer, an activation layer, a pooling layer with several feature maps as the outputs.

www.frontiersin.org

Effigy 1. 3D PCNN compages.

Convolutional layer

The procedure of convolutional layer is to convolve the input vector I with the convolution kernel K, represented past I1000. The shape of the input vector in our 3D PCNN model was [due north, d, due west, h, c], where d, due west, h, c represent the depth, width, height and channel numbers (which is 1 for a grayscale image) of the input vector, respectively, and n is the batch size which is a hyperparameter that was set up to 45 (an empirical value) in this paper. In the first layer, the input size was 58 × 70 × 58 × 1, which was the 3D size (58 × lxx × 58) of the input paradigm plus a unmarried channel (grayscale epitome). The shape of the convolution kernel was [d grand , w k , h k , c in , c out ], where d 1000 , w k , h thousand represents the depth, width, and height of the convolution kernel, respectively. In all three hidden layers, the kernel size was set to3 × 3 × 3, which means that d k = w grand = h 1000 = 3. The c in is the number of input channels which is equal to the channel number of the input vector. The c out is the number of output channels. As each kernel has an output channel, c out is equal to the number of convolution kernels, and is besides the aforementioned as the number of input channels for the next subconscious layer. In all convolution layers, the moving step of the kernel was gear up to 1 and padding mode was to "SAME."

Batch normalization layer

Batch normalization was performed later on the convolutional layer. Batch normalization is a kind of training fox which normalizes the information of each mini-batch (with cipher mean and variance of one) in the hidden layers of the network. To convalesce the gradient internal covariate shift phenomenon and speed upward the CNN training, an Adam Slope Decent method was used to train the model (Kingma and Ba, 2015).

Activation layer

After the batch normalization operation, an activation function was used to non-linearize the convolution upshot. The activation role we used in the model was the Rectified linear unit, ReLU (Nair and Hinton, 2010).

Pooling layer

Pooling layer was added afterwards the activation layer. Pooling layers in the CNN summarize the outputs of neighboring groups of neurons in the same kernel map (Krizhevsky et al., 2012). Max-pooling method was used in this layer.

The outputs of each subconscious layer were characteristic maps, which were the features extracted from the input images to the hidden layer. The outputs from the previous subconscious layer were the inputs to the next layer. In our model, the first subconscious layer generated 32 characteristic maps, the second hidden layer produced 64 feature maps, and the third hidden layer yielded 128 feature maps. Finally, we integrated the terminal 128 characteristic maps into the input of the softmax layer through a linear layer, then got the final classification results from the softmax layer.

In our model, the input X ∈ {x (i), x (two), …, x (n)}, 10 (i) was the ith subject field's FA value. Y ∈ {y (1), y (2), …, y (n)}, y (i) was the ith bailiwick'due south characterization that were processed to one-hot vector where [1 0] represents man and [0 one] woman. We used h(θ, x) to represent the proposed 3D PCNN model. And then we had:

where ŷ represents the predicted value obtained using the 3D PCNN on a sample x.

Parameters Optimization

The initial values of the weights of the convolution kernels were random values selected from a truncated normal distribution with standard deviation of 0.1. We defined a cost function to conform these weights based on the softmax cross entropy (Dunne and Campbell, 1997):

J ( θ , x ) = - i = 1 n ŷ ( i ) log P ( ŷ ( i ) = y ( i ) | x = x ( i ) ) ( ii )

Equally such, the task of adjusting the weight value became an optimization problem with J(θ, x) as the optimization goal, where a small penalty was given if the classification result was right, and vice versa. We used the Adam Gradient Descent (Kingma and Ba, 2015) optimization algorithm to reach this goal in the model training. All parameters in the Adam algorithm were set to the empirical values recommended past Kingma and Ba (2015), i.east., learning charge per unit was α = 0.001, exponential decay rates for the moment estimates were β 1 = 0.9, β i = 0.999, ε = 10−viii.

Cross-Validation

To ensure the independent training and testing in the cantankerous-validation. The process of cross-validation is shown in Figure 2. We implemented a two-loop nested cross-validation scheme (Varoquaux et al., 2017). We divided the data set into three parts, i.e., 80% of the data as the training set up for model training, 10% as the verification fix for parameter option, and 10% equally the testing gear up for evaluating the generalization ability of the model. To eliminate the random mistake of model preparation, nosotros run 10 fold cross validation and then took the average of classification accuracies as the final result.

www.frontiersin.org

Figure 2. Model training and nested cross validation. (A) General overview. (B) 10 fold cross validation.

Features in First Hidden Layer

CNN has an reward that information technology can extract primal features by itself (Zeng et al., 2018c). However, these features may be difficult to interpret since they are highly abstract features. Thus, in this report, we merely analyzed the features obtained in the first hidden layer, since they are the direct outputs from the convolution on the grayscale FA images. In this case, the convolution operation of the showtime layer is equivalent to applying a convolution kernel based spatial filter on the FA images. The obtained features are less abstractive than those from the 2nd and three hidden layers. In that location are totally 32 features in the first hidden layer. These features are the lowest-level features which may represent the structural features of FA images. We firstly computed the mean of voxel values across all subjects in each group (human being vs. woman) for each characteristic and and so evaluated their group-level difference using a two-sample t-exam. Besides, we as well computed the entropy on each characteristic for each individual:

where p i indicates the frequency of pixel with value i appears in the epitome. The entropy of each feature likely indicates the complexity of brain structural encoded in that feature. We likewise performed a two-sample t-test on entropy results to explore the differences between men and women. A strict Bonferroni correction was practical for multiple comparisons with the threshold of 0.05/32 = ane.56 × ten−three to remove spurious significance.

Discriminative Ability of Encephalon Regions

In order to determine which encephalon regions may play important office in gender-related brain structural differences, nosotros repeated the same 3D PCNN-based classification on each specific encephalon region. We segmented each FA image into 246 gray matter regions of interests (ROIs) according to the Man Brainnetome Atlas (Fan et al., 2016) and 48 white affair ROIs co-ordinate to the ICBM-DTI-81 White-Matter Labels Atlas (Mori et al., 2005). The nomenclature accuracy was then obtained for each ROIs. A higher accurateness indicates a more important role of that ROI in gender-related difference. A map was then obtained based on the classification accuracies of different ROIs to testify their distribution in the brain.

Comparisons With Tract Based Spatial Statistics and Support Vector Automobile

To justify the effectiveness of our method, the Tract Based Spatial Statistics (TBSS) and Support Vector Automobile (SVM) were applied to our dataset as comparisons, since these are 2 popular methods for information analysis in neuroimaging studies (Bach et al., 2014; Zeng et al., 2018b). Nosotros compared the results in following two conditions: (1) We used the SVM equally the classifier while keeping the same preprocessing process in gild to compare its results with our 3D PCNN method. Nosotros flatten each sample from the 3D FA matrix into a vector, and then fed the SVM with the vector. (two) We used the TBSS to identify the brain regions where are shown the statistically significant gender-related difference.

Results

Nomenclature Results on the Whole-Brain FA Images

Using our 3D PCNN methods on the whole-brain FA images, we can well-distinguish men and women with the classification accuracy of 93.three%. This consequence is much better than using the SVM, whose nomenclature accuracy is only 78.2%.

As comparisons, nosotros also used Doctor, Advertising, and RD to repeat the same analysis. The classification accuracy of Medico is 65.8%, AD is 69.nine%, and RD is 67.8%. All of them are lower than the classification accuracy obtained by using FA.

Feature Analysis in the First Subconscious Layer of 3D PCNN

The result of two-sample t-test of 32 features of men and women shows that at that place are 25 features had meaning gender differences including 13 features that women accept larger values and 12 features that men have larger values (see Figure iii). Interestingly, men have significantly higher entropy than women for all features (meet Figure 4).

www.frontiersin.org

Figure iii. Between-group differences of 32 features in voxel values. The mean (bar height) and standard divergence (error bars) of voxel values across all subjects in each group were evaluated for each characteristic. Their group-level difference was examined using a two-sample t-exam. Bonferroni correction was practical for multiple comparisons with the threshold equal to 0.05/32 = 1.56 × 10−3 to remove spurious significance. The features with significantly larger mean voxel values for men are marked out with*, while features with significantly larger mean voxel values for women are indicated by +.

www.frontiersin.org

Figure 4. Between-group differences of 32 features in entropy values. The mean (bar height) and standard deviation (error confined) of entropy value were computed beyond all subjects in each group for each feature. Their group-level difference was evaulated using a two-sample t-test. Bonferroni correction was applied for multiple comparisons with the threshold equal to 0.05/32 = 1.56 × 10−iii to remove spurious significance. The entropy values are significantly larger in men than in women for features.

Classification on Each Specific ROI

TBSS could not detect whatsoever statistically significant gender-related difference in this dataset. However, using 3D PCNN, nosotros did discover gender-related differences in all ROIs in the both grayness and white matters, equally the classification accuracies (>75%) are much higher than the chance level (50%) for all ROIs. The maps of classification accuracies for different ROIs are shown in Effigy 5. The detail nomenclature results are provided in the supplement (run into Table S1 for gray thing and Table S2 for white thing). In the gray matter, the top v regions with highest nomenclature accuracies are the left precuneus (Broadman area, BA 31, 87.2%), the left postcentral gyrus (BA 1/two/3 torso region, 87.2%), the left cingulate gyrus (BA 32 subgenual area, 87.2%), the correct orbital gyrus of frontal lobe (BA xiii, 87.ane%) and the left occipital thalamus (86.nine%). In the white matter, the superlative five regions with highest classification accuracies are middle cerebellum peduncle (89.7%), knee joint of corpus callosum (88.iv%), the right anterior corona radiata (88.3%), the right superior corona radiata (86%), and the left anterior limb of internal capsule (85.4%).

www.frontiersin.org

Figure five. Maps of nomenclature accuracies for unlike ROIs in the grayness and white matter of the brain. (A) Results in 246 grayness affair regions of interests (ROIs) according to the Human Brainnetome Atlas (B) Results in 48 white thing ROIs according to the ICBM-DTI-81 White-Matter Labels Atlas.

Discussions

Classification on the Whole-Brain FA

The proposed 3D PCNN model achieved 93.iii% classification accurateness in the whole-brain FA. The high classification accuracy charge per unit indicates that the proposed model tin can accurately find the brain structure difference betwixt men and women, which is the basis of subsequent feature analysis and subreginal analysis. Virtually existing classification, regression, and other machine learning methods are shallow learning algorithms, such equally the SVM, Boosting, maximum entropy, and Logistic Regression. When complex functions demand to be expressed, the models obtained by these algorithms volition and then accept a limitation with pocket-size size of samples and express computational resources. Thus, the generalization ability will be deteriorated as nosotros demonstrated in the results from the SVM. The benefit of deep learning algorithms, using multiple layers in the artificial neural network, is that one can correspond circuitous functions with few parameters. The CNN is one of the widely used deep learning algorithms. In compared to the method similar SVM, which is just a classifier, 3D CNN is a method that tin can excerpt the 3D spatial structure features of the input prototype. Through constructing the 3D PCNN model, we extracted highly abstract features from FA images, which may, thusly, improve the classification accuracy. FA describes the partial anisotropy alphabetize, which indicates the difference between one management and others (Feldman et al., 2010). Information technology tin reflect alterations in diverse tissue properties including axonal size, axonal packing density, and degree of myelination (Chung et al., 2016). In this study, we also run the same assay using MD, AD, and RD images for comparisons. All their results are lower than that of FA, indicating that using FA is more effective to detect the structure difference betwixt men and women's brain than using other images.

Feature Analysis in the Outset Hidden Layer of 3D PCNN

The degree of the macroscopic diffusion anisotropy is often quantified by the FA (Lasi et al., 2014). Previous studies found that wider skeleton of white matter in adult female's brain merely wider region of gray affair in man'due south encephalon (Witelson et al., 1995; Zaidi, 2010; Gong et al., 2011; Menzler et al., 2011). These mean that men appear to have more gray thing, fabricated up of active neurons, while women may have more white matter for the neuronal communication betwixt different areas of the brain. Furthermore, a recent study found that men had college FA values than women in center aged to elderly (between 44 and 77 years onetime) people past using a statistical analysis (Ritchie et al., 2018). This study focuses on the immature salubrious individuals with the age range between 22 and 36 years old. The structural features extracted from 3D PCNN reflect the encephalon construction difference betwixt men and women. In the first subconscious layer of 3D PCNN model, we found 25 features that accept significant difference betwixt men and women in voxels value. Moreover, using entropy mensurate, we establish that men'south brains likely have more complex features as reflected by significantly higher entropy. These results indicated that the gender-related differences likely exist in the whole-brain range including both white and gray matters.

Nearly Discriminative Brain Regions

Using FA images from each specific brain region every bit the input to the 3D PCNN, nosotros found all tested brain regions may have gender-related deviation, though the TBSS assay cannot detect these differences. The brain regions with high classification accuracies include the left precuneus (Broadman area, BA 31, 87.2%), the left postcentral gyrus (BA 1/2/3 body region, 87.2%), the left cingulate gyrus (BA 32 subgenual area, 87.2%), the right orbital gyrus of frontal lobe (BA xiii, 87.i%), and the left occipital thalamus (86.ix%) in the grey matter, and middle cerebellum peduncle (89.7%), human knee of corpus callosum (88.4%), the right anterior corona radiata (88.iii%), the right superior corona radiata (86%), and the left anterior limb of internal capsule (85.4%).

The gender-related morphological difference at the corpus callosum has been previously reported, which may be associated with interhemispheric interaction (Sullivan et al., 2001; Luders et al., 2003; Prendergast et al., 2015). Nevertheless, likely due to the limitation of applied methods, not all previous studies have reported this divergence (Abe et al., 2002). Those likely results in the inconsistent findings were across different studies. Through 3D PCNN model, our results confirm that there is likely a morphological deviation at the human knee of corpus callosum between human being and women.

The middle cerebellum peduncle is the encephalon area connected to the pons and receiving the inputs mainly from the pontine nuclei (Glickstein and Doron, 2008), which are the nuclei of the pons involved in motor activeness (Wiesendanger et al., 1979). Raz et al. (2001) found larger book in the cerebellum of men than women. The cerebellar cells release diffusible substances that promote the survival of thalamic neurons (Tracey et al., 1980; Hisanaga and Abrupt, 1990). Previous studies have reported gender-difference differences in the bones glucose metabolism in the thalamus of young subjects between the ages of twenty and 40 (Fujimoto et al., 2008). Abreast the thalamus and cerebellum, the postcentral gyrus was also found in our results as the brain region with high classification accuracy. Thus, there is very likely a gender-related difference in the cerebellar-thalamic-cortical circuitry. This divergence may as well be related to the reported gender differences in neurological degenerative diseases such as Parkinson's Illness (Lyons et al., 1998; Dluzen and Mcdermott, 2000; Miller and Cronin-Golomb, 2010), where the pathological changes are usually plant in the cerebellar-thalamic-cortical circuitry.

The findings of the current study as well indicated the gender-related difference in the limbic-thalamo-cortical circuitry. Anterior corona radiata is part of the limbic-thalamo-cortical circuitry and includes thalamic projections from the internal capsule to the prefrontal cortex. White matter changes in the anterior corona radiata could effect in many of the cerebral and emotion regulation disturbances (Drevets, 2001). The orbital gyrus of frontal cortex grey matter areas and cingulate gyrus take besides been reported to be associated with the emotion regulation arrangement (Fan et al., 2005). Thus, the gender-related difference in the limbic-thalamo-cortical circuitry may explain the gender differences in thalamic activation during the processing of emotional stimuli or unpleasant linguistic information concerning interpersonal difficulties as demonstrated by previous fMRI (Lee and Kondziolka, 2005; Shirao et al., 2005).

In summary, by using the designed 3D PCNN algorithm, we confirmed that the gender-related differences exist in the whole-brain FA images every bit well equally in each specific brain regions. These gender-related brain structural differences might be related to gender differences in cognition, emotional control as well as neurological disorders.

Information Availability

Publicly bachelor datasets were analyzed in this study. This data tin can be establish here: https://world wide web.humanconnectome.org/.

Writer Contributions

JX, YT, and YY contributed to the conception and design of the study. YT, JX, and YZ performed information analysis. YT and JX drafted manuscript. YT and YY participated in editing the manuscript.

Funding

JX and YZ are supported past 111 Project (No. B18059). YT is supported by grant 2016JJ4090 from the Natural Science Foundation of Hunan Province and grants 2017T100613 and 2016M592452 from the Cathay Postdoctoral Scientific discipline Foundation, China. YY is supported by the Dixon Translational Research Grants Initiative (PI: YY) from the Northwestern Memorial Foundation (NMF) and the Northwestern University Clinical and Translational Sciences (NUCATS) Establish.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of whatsoever commercial or financial relationships that could exist construed as a potential disharmonize of interest.

Supplementary Material

The Supplementary Material for this article can be constitute online at: https://www.frontiersin.org/manufactures/10.3389/fnins.2019.00185/full#supplementary-material

References

Abadi, 1000., Barham, P., Chen, J., Chen, Z., Davis, A., Dean, J., et al. (2016). "Tensorflow: a system for big-scale machine learning," in twelfth USENIX Symposium on Operating Systems Design and Implementation (OSDI 16), (Savannah, GA), 265–283.

Google Scholar

Abe, O., Aoki, S., Hayashi, N., Yamada, H., Kunimatsu, A., Mori, H., et al. (2002). Normal aging in the central nervous system: quantitative MR diffusion-tensor analysis. Neurobiol. Aging 23, 433–441. doi: x.1016/S0197-4580(01)00318-9

PubMed Abstract | CrossRef Total Text | Google Scholar

Bach, M., Laun, F. B., Leemans, A., Revenue enhancement, C. 1000., Biessels, G. J., Stieltjes, B., et al. (2014). Methodological considerations on tract-based spatial statistics (TBSS). Neuroimage 100, 358–369. doi: 10.1016/j.neuroimage.2014.06.021

PubMed Abstract | CrossRef Full Text | Google Scholar

Chung, A. W., Seunarine, K. Chiliad., and Clark, C. A. (2016). NODDI reproducibility and variability with magnetic field strength: a comparison between 1.5 T and three T. Human Brain Mapp. 37, 4550–4565. doi: x.1002/hbm.23328

PubMed Abstruse | CrossRef Full Text | Google Scholar

Cole, J. H., Tsagkrasoulis, D., Mwa, C., Steves, C., Spector, T. D., et al. (2017). Predicting brain historic period with deep learning from raw imaging information results in a reliable and heritable biomarker. Neuroimage 163, 115–124. doi: ten.1016/j.neuroimage.2017.07.059

PubMed Abstract | CrossRef Full Text | Google Scholar

Cosgrove, K. P., Mazure, C. Yard., and Staley, J. Chiliad. (2007). Evolving knowledge of sex differences in brain structure, role, and chemistry. Biol. Psychiatry 62, 847–855. doi: x.1016/j.biopsych.2007.03.001

PubMed Abstract | CrossRef Full Text | Google Scholar

Dluzen, D. E., and Mcdermott, J. 50. (2000). Gender differences in neurotoxicity of the nigrostriatal dopaminergic organization: implications for Parkinson'south illness. J. Gender Spec. Med. iii, 36–42.

PubMed Abstruse | Google Scholar

Drevets, Westward. C. (2001). Neuroimaging and neuropathological studies of low: implications for the cerebral-emotional features of mood disorders. Curr. Opin. Neurobiol. eleven, 240–249. doi: 10.1016/S0959-4388(00)00203-8

PubMed Abstruse | CrossRef Full Text | Google Scholar

Dunne, R. A., and Campbell, North. A. (1997). "On the pairing of the softmax activation and cantankerous-entropy penalty functions and the derivation of the softmax activation office," in Proc. 8th Aust. Conf. on the Neural Networks, (Melbourne, VIC: Citeseer), 185.

Google Scholar

Fan, J., Mccandliss, B. D., Fossella, J., Flombaum, J. I., and Posner, One thousand. I. (2005). The activation of attentional networks. Neuroimage 26, 471–479. doi: x.1016/j.neuroimage.2005.02.004

PubMed Abstract | CrossRef Full Text | Google Scholar

Fan, L., Zhuo, J., Zhang, Y., Wang, J., Chen, L., et al. (2016). The human brainnetome atlas: a new brain atlas based on connectional compages. Cereb. Cortex 26, 3508–3526. doi: ten.1093/cercor/bhw157

PubMed Abstract | CrossRef Total Text | Google Scholar

Feldman, H. M., Yeatman, J. D., Lee, Due east. S., Barde, L. H., and Gaman-Bean, S. (2010). Diffusion tensor imaging: a review for pediatric researchers and clinicians. J. Dev. Behav. Pediatr. 31, 346–356. doi: x.1097/DBP.0b013e3181dcaa8b

PubMed Abstruse | CrossRef Total Text | Google Scholar

Fujimoto, T., Matsumoto, T., Fujita, Due south., Takeuchi, Chiliad., Nakamura, Grand., Mitsuyama, Y., et al. (2008). Changes in glucose metabolism due to aging and gender-related differences in the healthy human brain. Psychiatry Res. 164, 58–72. doi: ten.1016/j.pscychresns.2006.12.014

PubMed Abstract | CrossRef Full Text | Google Scholar

Golestani, N., Molko, N., Dehaene, S., Lebihan, D., and Pallier, C. (2006). Brain structure predicts the learning of strange speech sounds. Cereb. Cortex 17, 575–582. doi: 10.1093/cercor/bhk001

PubMed Abstract | CrossRef Full Text | Google Scholar

Hisanaga, K., and Abrupt, F. R. (1990). Diffusible neurotrophic factors for thalamic neurons are released from target neocortex and non-target cerebellum. Progr. Brain Res. 82, 67–71. doi: x.1016/S0079-6123(08)62591-1

PubMed Abstract | CrossRef Total Text | Google Scholar

Ingalhalikar, M., Smith, A., Parker, D., Satterthwaite, T. D., Elliott, Thousand. A., Ruparel, K., et al. (2014). Sexual practice differences in the structural connectome of the homo encephalon. Proc. Natl. Acad. Sci. U.S.A. 111, 823–828. doi: 10.1073/pnas.1316909110

PubMed Abstract | CrossRef Full Text | Google Scholar

Ioffe, S., and Szegedy, C. (2015). "Batch normalization: accelerating deep network training past reducing internal covariate shift," in Proceedings of the 32nd International Conference on Motorcar Learning, Vol. 37 (Lille), 448–456.

Google Scholar

Joel, D., and Tarrasch, R. (2014). On the mis-presentation and misinterpretation of gender-related information: the case of Ingalhalikar'southward human connectome study. Proc. Natl. Acad. Sci. U.Due south.A. 111, E637–E637. doi: 10.1073/pnas.1323319111

PubMed Abstract | CrossRef Full Text | Google Scholar

Kingma, D. P., and Ba, L. J. (2015). "Adam: A method for stochastic optimization," in Proceedings of the third International Conference on Learning Representations (San Diego, CA).

Google Scholar

Kochunov, P. M., Thompson, P., Lancaster, J. 50., Bartzokis, G., Smith, S., Coyle, T., et al. (2007). Relationship between white thing fractional anisotropy and other indices of cerebral wellness in normal aging: tract-based spatial statistics study of crumbling. Neuroimage 35, 478–487. doi: x.1016/j.neuroimage.2006.12.021

PubMed Abstract | CrossRef Full Text | Google Scholar

Krizhevsky, A., Sutskever, I., and Hinton, One thousand. Eastward. (2012). "Imagenet nomenclature with deep convolutional neural networks," in Advances in Neural Information Processing Systems 25 (NIPS 2012) (Lake Tahoe, NV), 1097–1105.

Google Scholar

Lasi, S., Szczepankiewicz, F., Eriksson, South., Nilsson, M., and Topgaard, D. (2014). Microanisotropy imaging: quantification of microscopic diffusion anisotropy and orientational social club parameter by diffusion MRI with magic-bending spinning of the q-vector. Front. Phys. ii:eleven. doi: 10.3389/fphy.2014.00011

CrossRef Full Text | Google Scholar

Lu, Y., Yi, S., Zeng, North., Liu, Y., and Zhang, Y. (2017). Identification of rice diseases using deep convolutional neural networks. Neurocomputing 267, 378–384. doi: 10.1016/j.neucom.2017.06.023

CrossRef Full Text | Google Scholar

Luders, E., Rex, D. E., Narr, Thou. L., Woods, R. P., Jancke, L., Thompson, P. M., et al. (2003). Relationships between sulcal asymmetries and corpus callosum size: gender and handedness effects. Cereb. Cortex 13, 1084–1093. doi: 10.1093/cercor/13.ten.1084

PubMed Abstract | CrossRef Total Text | Google Scholar

Lyons, K. Due east., Hubble, J. P., Troster, A. I., Pahwa, R., and Koller, West. C. (1998). Gender differences in Parkinson's illness. Clin. Neuropharmacol. 21, 118–121.

PubMed Abstruse | Google Scholar

Menzler, M., Belke, G., Wehrmann, E., Cracow, G., Lengler, U., Jansen, A., et al. (2011). Men and women are different: diffusion tensor imaging reveals sexual dimorphism in the microstructure of the thalamus, corpus callosum and cingulum. Neuroimage 54, 2557–2562. doi: x.1016/j.neuroimage.2010.11.029

PubMed Abstract | CrossRef Full Text | Google Scholar

Mori, S., Al, S. M., Wakana, S., and Zijl, V. (2005). MRI Atlas of Human being White Matter. London, UK: Elsevier/Academic Printing.

Google Scholar

Mueller, One thousand, Anwander, A, Möller, H. Eastward., Horstmann, A., Lepsien, J., Busse, F., et al. (2011). Sexual practice-dependent influences of obesity on cerebral white thing investigated by diffusion-tensor imaging. PloS I vi:e18544. doi: 10.1371/journal.pone.0018544

PubMed Abstract | CrossRef Total Text | Google Scholar

Nair, Five., and Hinton, Grand. E. (2010). "Rectified linear units improve restricted boltzmann machines," in Proceedings of the 27th International Conference on Car Learning (ICML-10) (Haifa), 807–814.

Google Scholar

Prendergast, D. M., Ardekani, B., Ikuta, T., John, M., Peters, B., Derosse, P., et al. (2015). Age and sex activity effects on corpus callosum morphology across the lifespan. Human Encephalon Mapp. 36, 2691–2702. doi: 10.1002/hbm.22800

PubMed Abstruse | CrossRef Full Text | Google Scholar

Raz, N., Gunning-Dixon, F., Head, D., Williamson, A., and Acker, J. D. (2001). Age and sex differences in the cerebellum and the ventral pons: a prospective MR study of healthy adults. Am. J. Neuroradiol. 22, 1161–1167.

PubMed Abstruse | Google Scholar

Ritchie, S. J., Cox, Southward. R., Shen, X., Lombardo, M. V., Reus, Fifty. M., Alloza, C., et al. (2018). Sexual practice differences in the developed human brain: evidence from 5216 UK Biobank participants. Cereb. Cortex 28, 2959–2975. doi: 10.1093/cercor/bhy109

PubMed Abstract | CrossRef Full Text | Google Scholar

Salat, D. H., Tuch, D. S., Greve, D. North., Van Der Kouwe, A. J., Hevelone, N. D., Zaleta, A. K., et al. (2005). Age-related alterations in white matter microstructure measured past diffusion tensor imaging. Neurobiol. Aging 26, 1215–1227. doi: 10.1016/j.neurobiolaging.2004.09.017

PubMed Abstract | CrossRef Full Text | Google Scholar

Shen, H., Wang, L., Liu, Y., and Hu, D. (2010). Discriminative analysis of resting-state functional connectivity patterns of schizophrenia using depression dimensional embedding of fMRI. Neuroimage 49, 3110–3121. doi: 10.1016/j.neuroimage.2009.11.011

PubMed Abstract | CrossRef Full Text | Google Scholar

Shiino, A., Chen, Y. Westward., Tanigaki, Thousand., Yamada, A., Vigers, P., Watanabe, T., et al. (2017). Sex activity-related divergence in human being white matter volumes studied: inspection of the corpus callosum and other white matter by VBM. Sci. Rep. 7:39818. doi: 10.1038/srep39818

PubMed Abstract | CrossRef Full Text | Google Scholar

Shirao, N., Okamoto, Y., Okada, G., Ueda, Yard., and Yamawaki, S. (2005). Gender differences in brain action toward unpleasant linguistic stimuli concerning interpersonal relationships: an fMRI study. Eur. Curvation. Psychiatr. Clin. 255, 327–333. doi: 10.1007/s00406-005-0566-ten

PubMed Abstruse | CrossRef Full Text | Google Scholar

Sullivan, E. 5., Rosenbloom, M. J., Desmond, J. E., and Pfefferbaum, A. (2001). Sex differences in corpus callosum size: human relationship to age and intracranial size. Neurobiol. Aging 22, 603–611. doi: ten.1016/S0197-4580(01)00232-9

PubMed Abstruse | CrossRef Total Text | Google Scholar

Tang, Y., Liu, B., Yang, Y., Wang, C.-M., Meng, L., Tang, B.-S., et al. (2018). Identifying mild-moderate Parkinson's disease using whole-brain functional connectivity. Clin. Nerophysiol. 129, 2507–2516. doi: 10.1016/j.clinph.2018.09.006

PubMed Abstract | CrossRef Full Text | Google Scholar

Thatcher, D. 50., Pajtek, Due south., Chung, T., Terwilliger, R. A., and Clark, D. B. (2010). Gender differences in the relationship between white matter organization and adolescent substance utilise disorders. Drug Alcohol. Depend. 110, 55–61. doi: 10.1016/j.drugalcdep.2010.02.004

PubMed Abstract | CrossRef Full Text | Google Scholar

Tian, R., Yang, Y., and Van Der Helm FCT Dewald, JPA. (2018). A novel approach for modeling neural responses to joint perturbations using the NARMAX method and a hierarchical neural network. Front. Comput. Neurosci. 12:96. doi: x.3389/fncom.2018.00096

PubMed Abstract | CrossRef Total Text | Google Scholar

Tracey, D., Asanuma, C. G., Jones, East., and Porter, R. (1980). Thalamic relay to motor cortex: Afferent pathways from brain stem, cerebellum, and spinal cord in monkeys. J. Neurophysiol. 44, 532–554. doi: 10.1152/jn.1980.44.3.532

PubMed Abstract | CrossRef Total Text | Google Scholar

Van Essen, D. C., Smith, Southward. M., Barch, D. M., Behrens, T. E., Yacoub, Due east., Ugurbil, K., et al. (2013). The WU-Minn human connectome project: an overview. Neuroimage eighty, 62–79. doi: 10.1016/j.neuroimage.2013.05.041

PubMed Abstract | CrossRef Total Text | Google Scholar

Varoquaux, G., Raamana, P. R., Engemann, D. A., Hoyos-Idrobo, A., Schwartz, Y., and Thirion, B. (2017). Assessing and tuning brain decoders: cross-validation, caveats, and guidelines. NeuroImage 145, 166–179. doi: 10.1016/j.neuroimage.2016.x.038

PubMed Abstract | CrossRef Full Text | Google Scholar

Wen, D., Wei, Z., Zhou, Y., Li, Chiliad., Zhang, Ten., and Han, Westward. (2018). Deep learning methods to process fMRI data and their application in the diagnosis of cognitive impairment: a brief overview and our opinion. Front end. Neuroinform. 12:23. doi: ten.3389/fninf.2018.00023

PubMed Abstract | CrossRef Full Text | Google Scholar

Westlye, L. T., Walhovd, Chiliad. B., Dale, A. Grand., Bjørnerud, A., Due-Tønnessen, P., Engvig, A., et al. (2009). Life-bridge changes of the human brain white matter: diffusion tensor imaging (DTI) and volumetry. Cereb. Cortex twenty, 2055–2068. doi: 10.1093/cercor/bhp280

PubMed Abstract | CrossRef Full Text | Google Scholar

Wiesendanger, R., Wiesendanger, M., and Ru, D. (1979). An anatomical investigation of the corticopontine projection in the primate (Macaca fascicularis and Saimiri sciureus)—II. The project from frontal and parietal association areas. Neuroscience four, 747–765. doi: x.1016/0306-4522(79)90004-6

CrossRef Full Text | Google Scholar

Witelson, S. F., Glezer, I. I., and Kigar, D. L. (1995). Women accept greater density of neurons in posterior temporal cortex. J. Neurosci. 15, 3418–3428. doi: 10.1523/JNEUROSCI.15-05-03418.1995

PubMed Abstract | CrossRef Full Text | Google Scholar

Zeng, L. L., Wang, H., Hu, P., Yang, B., Pu, W., Shen, H., et al. (2018a). Multi-site diagnostic classification of schizophrenia using discriminant deep learning with functional connectivity MRI. Ebiomedicine 30, 74–85. doi: x.1016/j.ebiom.2018.03.017

PubMed Abstract | CrossRef Full Text | Google Scholar

Zeng, Due north., Qiu, H., Wang, Z., Liu, W., Zhang, H., and Li, Y. (2018b). A new switching-delayed-PSO-based optimized SVM algorithm for diagnosis of Alzheimer's affliction. Neurocomputing 320, 195–202. doi: 10.1016/j.neucom.2018.09.001

CrossRef Full Text | Google Scholar

Zeng, Northward., Wang, Z., Zhang, H., Liu, W., and Alsaadi, F. East. (2016). Deep conventionalities networks for quantitative analysis of a gold immunochromatographic strip. Cogn. Comput. 8, 684–692. doi: ten.1007/s12559-016-9404-ten

CrossRef Full Text | Google Scholar

Zeng, Northward., Zhang, H., Vocal, B., Liu, W., Li, Y., and Dobaie, A. M. (2018c). Facial expression recognition via learning deep thin autoencoders. Neurocomputing 273, 643–649. doi: x.1016/j.neucom.2017.08.043

CrossRef Full Text | Google Scholar

mcconnellbutragreake72.blogspot.com

Source: https://www.frontiersin.org/articles/10.3389/fnins.2019.00185/full

0 Response to "There Is No Male and Female Brain Peer Review Study"

Отправить комментарий

Iklan Atas Artikel

Iklan Tengah Artikel 1

Iklan Tengah Artikel 2

Iklan Bawah Artikel