If the feature number is too large, it causes the curse of the dimensionality problem. In the pathological brain detection (PBD) system, each feature may have many possible values. Therefore, to ensure several samples exist with each combination of different feature values, an enormous amount of training data is required. Otherwise, if the training samples number is fixed, the classification performance deteriorates as the feature number increases.

In this chapter, three data sets for single-slice pathological brain detection (PBD), along with their download URLs, are given. All the data sets can be downloaded from The Whole Brain Atlas from the Harvard Medical School. The inclusion criteria of three commonly used data sets are introduced. The limitation of using didactic images is explained. In the field of pattern recognition, a training set is necessary, where data are labelled using known categories. The validation set is important to optimize hyper-parameters.

This chapter briefly introduces the development of neuroimaging modalities. We give a simple introduction about pneumoencephalography (PEG), cerebral angiography, computerized tomography, and nuclear imaging (positron emission tomography and single photon emission computed tomography). Later in the chapter, the technique of magnetic resonance imaging (MRI) is considered. Projectile risk, shimming technique, and water and fat suppression are discussed, with two contrasting types (spin–lattice and spin–spin) of MRI. A method for rough interpretation of magnetic resonance images is given. Finally, we make a comparison among different imaging modalities in MRI, including diffusion tensor imaging, functional MRI, magnetic resonance angiography, and magnetic resonance spectral imaging.

In this chapter, four important wavelet families are discussed: the Daubechies wavelet family, the Coiflet wavelet family, the Morlet wavelet family, and the biorthogonal wavelet family. The wavelet display function and the "waveinfo" command are introduced so that the detailed curve shape of scaling and wavelet functions, for both decomposition and reconstruction, can be viewed in arbitrary accuracy. Several popular wavelet transform variants are presented. The ε-decimated wavelet transform chooses either odd or even index randomly.

Image segmentation is an important application of polarimetric synthetic aperture radar. This study aimed to create an 11-layer deep convolutional neural network for this task. The Pauli decomposition formed the RGB image and was used as the input. We created an 11-layer convolutional neural network (CNN). L-band data over the San Francisco bay area and C-band data over Flevoland area were employed as the dataset. For the San Francisco bay PSAR image, our method achieved an overall accuracy of 97.32%, which was at least 2% superior to four state-of-the-art approaches. We provided the confusion matrix over test area, and the kernel visualization. We compared the max pooling and average pooling. We validated by experiment that four convolution layers perform the best. Besides, our method gave better results than AlexNet. The GPU yields a 173× acceleration on the training samples, and a 181× acceleration on the test samples, compared to standard CPU. For the Flevoland PSAR image, our 11-layer CNN also gives better overall accuracy than five state-of-the-art approaches. The convolutional neural network is better than traditional classifiers and is effective in remote sensing image segmentation.

This chapter is a natural follow on from Chap. 4 , focusing on multiscale and multiresolution features. First, the development of signal processing from Fourier transform to short-time Fourier transform to wavelet analysis, is presented. The advantages and disadvantages of the three techniques are analyzed. Next, the question of why wavelet transform is the most popular feature extraction method is answered, by comparison between fingerprints and brain gyri.

Aim: Currently, identifying multiple sclerosis (MS) by human experts may come across the problem of “normal-appearing white matter”, which causes a low sensitivity. Methods: In this study, we presented a computer vision based approached to identify MS in an automatic way. This proposed method first extracted the fractional Fourier entropy map from a specified brain image. Afterwards, it sent the features to a multilayer perceptron trained by a proposed improved parameter-free Jaya algorithm. We used cost-sensitivity learning to handle the imbalanced data problem. Results: The 10 × 10-fold cross validation showed our method yielded a sensitivity of 97.40 ± 0.60%, a specificity of 97.39 ± 0.65%, and an accuracy of 97.39 ± 0.59%. Conclusions: We validated by experiments that the proposed improved Jaya performs better than plain Jaya algorithm and other latest bioinspired algorithms in terms of classification performance and training speed. In addition, our method is superior to four state-of-the-art MS identification approaches.

Aim: Multiple sclerosis is a severe brain and/or spinal cord disease. It may lead to a wide range of symptoms. Hence, the early diagnosis and treatment is quite important. Method: This study proposed a 14-layer convolutional neural network, combined with three advanced techniques: batch normalization, dropout, and stochastic pooling. The output of the stochastic pooling was obtained via sampling from a multinomial distribution formed from the activations of each pooling region. In addition, we used data augmentation method to enhance the training set. In total 10 runs were implemented with the hold-out randomly set for each run. Results: The results showed that our 14-layer CNN secured a sensitivity of 98.77 ± 0.35%, a specificity of 98.76 ± 0.58%, and an accuracy of 98.77 ± 0.39%. Conclusion: Our results were compared with CNN using maximum pooling and average pooling. The comparison shows stochastic pooling gives better performance than other two pooling methods. Furthermore, we compared our proposed method with six state-of-the-art approaches, including five traditional artificial intelligence methods and one deep learning method. The comparison shows our method is superior to all other six state-of-the-art approaches.

Handcrafted features play an important role in PBD systems. In this chapter, some widely used features, used in PBD, are introduced. First, common 2D and 3D shape features are presented. Then, statistical measures (mean, variance, standard deviation, median, skewness, and kurtosis) and statistical plots (radar chart, pie chart, bar chart, wind rose, etc.) are discussed. The development of image moments is given, from raw moments, to central moments, to normalized central moments, to Hu moments. Besides this, a comparison between a Zernike moment and pseudo Zernike moment is provided. Gray-level co-occurrence matrix and Haralick features are discussed. Further, the standard Fourier transform, and its two important variants (sine and cosine transform), are explained. The fractional Fourier transform, and its three fast discrete implementation methods, are presented. Finally, three important entropy features are compared, i.e., Shannon entropy, Tsallis entropy, and Renyi entropy. Readers will see, in subsequent chapters, that most brain feature extraction methods practically used are in fact combinations of simple features introduced in this chapter.

Aim: Emotion recognition based on facial expression is an important field in affective computing. Current emotion recognition systems may suffer from two shortcomings: translation in facial image may deteriorate the recognition performance, and the classifier is not robust. Method: To solve above two problems, our team proposed a novel intelligent emotion recognition system. Our method used stationary wavelet entropy to extract features, and employed a single hidden layer feedforward neural network as the classifier. To prevent the training of the classifier fall into local optimum points, we introduced the Jaya algorithm. Results: The simulation results over a 20-subject 700-image dataset showed our algorithm reached an overall accuracy of 96.80 +/- 0.14%. Conclusion: This proposed approach performs better than five state-of-the-art approaches in terms of overall accuracy. Besides, the db4 wavelet performs the best among other whole db wavelet family. The 4-level wavelet decomposition is superior to other levels. In the future, we shall test other advanced features and training algorithms. (C) 2017 Elsevier B.V. All rights reserved.

BACKGROUNDThe number of patients with Alzheimer's disease is increasing rapidly every year. Scholars often use computer vision and machine learning methods to develop an automatic diagnosis system.OBJECTIVEIn this study, we developed a novel machine learning system that can make diagnoses automatically from brain magnetic resonance images.METHODSFirst, the brain imaging was processed, including skull stripping and spatial normalization. Second, one axial slice was selected from the volumetric image, and stationary wavelet entropy (SWE) was done to extract the texture features. Third, a single-hidden-layer neural network was used as the classifier. Finally, a predator-prey particle swarm optimization was proposed to train the weights and biases of the classifier.RESULTSOur method used 4-level decomposition and yielded 13 SWE features. The classification yielded an overall accuracy of 92.73±1.03%, a sensitivity of 92.69±1.29%, and a specificity of 92.78±1.51%. The area under the curve is 0.95±0.02. Additionally, this method only cost 0.88 s to identify a subject in online stage, after its volumetric image is preprocessed.CONCLUSIONIn terms of classification performance, our method performs better than 10 state-of-the-art approaches and the performance of human observers. Therefore, this proposed method is effective in the detection of Alzheimer's disease.

Cerebral microbleeds (CMBs) are the small foci of chronic blood products. CMBs are closely related to many diseases such as dementia, siderosis, ageing, etc. A data balance method is used for their identification, since CMB voxels in collected brain images are usually about 2000 times less common than non-CMB voxels.

Alzheimer's disease (AD) is a progressive brain disease. The goal of this study is to provide a new computer-vision based technique to detect it in an efficient way. The brain-imaging data of 98 AD patients and 98 healthy controls was collected using data augmentation method. Then, convolutional neural network (CNN) was used, CNN is the most successful tool in deep learning. An 8-layer CNN was created with optimal structure obtained by experiences. Three activation functions (AFs): sigmoid, rectified linear unit (ReLU), and leaky ReLU. The three pooling-functions were also tested: average pooling, max pooling, and stochastic pooling. The numerical experiments demonstrated that leaky ReLU and max pooling gave the greatest result in terms of performance. It achieved a sensitivity of 97.96%, a specificity of 97.35%, and an accuracy of 97.65%, respectively. In addition, the proposed approach was compared with eight state-of-the-art approaches. The method increased the classification accuracy by approximately 5% compared to state-of-the-art methods.

Magnetic resonance imaging (MRI) belongs to an imaging modality that helps doctors in diagnosis. MRI is commonly used for brain because it can generate clearer images in soft tissues than CT and PET. Early detection of brain diseases is dependent on brain MRIs to a great extent. However, brain MRIs contain too much information, manual interpretation requires too much time of the doctors. Therefore, in this paper, we proposed a novel computer aided diagnosis (CAD) system that can classify brain MRIs to be pathological or healthy automatically and accurately. It can provide a second opinion for the doctors and improve their efficiency. Firstly, we extracted 7 Hu moment invariants and 7 wavelet entropies from each brain MRI to form the feature vector. Then, we employed three improved extreme learning machines'(ELM) as the classifier% they are kernel based ELM (K-ELM), bat algorithm optimized ELM (BA-ELM) and weighted ELM (W-ELM). Open dataset was used in experiment. The statistics were obtained using 10 x 10 fold cross validation. The results revealed that W-ELM performed the best among the three classifiers with sensitivity of 96.93%, specificity of 96.67% and overall accuracy of 96.89% in merely 0.0026 seconds. Therefore, our proposed method is effective for pathological brain detection.

This chapter gives the latest training methods for training the weights and biases of feed-forward neural networks (FNNs). Note that the training is not pure optimization; hence, the training should be over the validation set. The traditional back propagation scheme, performed by the gradient descent method, and its variants, are reviewed. Later, 10 global optimization methods are compared, including the genetic algorithm, simulate annealing, the tabu search, the artificial immune system, particle swarm optimization, artificial bee colony, the firefly algorithm, ant colony optimization, biogeography-based optimization, and the Jaya algorithm.

This work is aimed to develop the key algorithms used in medical robots, which can detect alcohol use disorder from structural magnetic resonance imaging of brains. We enrolled 30 alcoholic participants and 30 nonalcoholic participants. In the algorithm stage, we suggested to use Hu moment invariant to extract global features, and use single-hidden layer neural-network as the classifier. Afterwards, we proposed a novel predator prey adaptive inertia chaotic particle swarm optimization algorithm to train the classifier. The ten-fold stratified cross validation showed that our method achieves a sensitivity of 90.67 +/- 3.16%, a specificity of 91.33 +/- 3.06%, and an accuracy of 91.00 +/- 1.41%. Our results are better than genetic algorithm, firefly algorithm, and particle swarm optimization. This proposes algorithm is effective in alcoholism detection. It can be installed on medical robots. (C) 2017 Elsevier Ltd. All rights reserved.

Facial emotion recognition is important in many academic and industrial applications. In this paper, our team proposed a novel facial emotion recognition method. First, we used discrete wavelet transform to extract wavelet coefficients from facial images. Second, principal component analysis was utilized to reduce the features. Third, a single-hidden-layer neural network was used as the classifier. Finally and most importantly, we introduced the cat swarm optimization to train the weights and biases of the classifier. The ten-fold stratified cross validation showed cat swarm optimization method achieved an overall accuracy of 89.49 ± 0.76