The high human labor demand involved in collecting paired medical imaging data severely impedes the application of deep learning methods to medical image processing tasks such as tumor segmentation. The situation is further worsened when collecting multi-modal image pairs. However, this issue can be resolved through the help of generative adversarial networks, which can be used to generate realistic images. In this work, we propose a novel framework, named TumorGAN, to generate image segmentation pairs based on unpaired adversarial training. To improve the quality of the generated images, we introduce a regional perceptual loss to enhance the performance of the discriminator. We also develop a regional $L_1$ loss to constrain the color of the imaged brain tissue. Finally, we verify the performance of TumorGAN on a public brain tumor data set, BraTS 2017. The experimental results demonstrate that the synthetic data pairs generated by our proposed method can practically improve tumor segmentation performance when applied to segmentation network training.

Background: Time-Frequency analysis of electroencephalogram (EEG) during different mental tasks received significant attention. As EEG is non-stationary, time-frequency analysis is essential to analyze brain states during different mental tasks. Further, the time-frequency information of EEG signal can be used as a feature for classification in brain-computer interface (BCI) applications. Methods: To accurately model the EEG, band-limited multiple Fourier linear combiner (BMFLC), a linear combination of truncated multiple Fourier series models is employed. A state-space model for BMFLC in combination with Kalman filter/smoother is developed to obtain accurate adaptive estimation. By virtue of construction, BMFLC with Kalman filter/smoother provides accurate time-frequency decomposition of the bandlimited signal. Results: The proposed method is computationally fast and is suitable for real-time BCI applications. To evaluate the proposed algorithm, a comparison with short-time Fourier transform (STFT) and continuous wavelet transform (CWT) for both synthesized and real EEG data is performed in this paper. The proposed method is applied to BCI Competition data IV for ERD detection in comparison with existing methods. Conclusions: Results show that the proposed algorithm can provide optimal time-frequency resolution as compared to STFT and CWT. For ERD detection, BMFLC-KF outperforms STFT and BMFLC-KS in real-time applicability with low computational requirement.

Recent works on brain functional analysis have accentuated the importance of distributed functional networks and synchronized activity between networks in mediating cognitive functions. The network perspective is important to relate mechanisms of brain functions and the basis for classifying brain states. In this work, the network patterns related to neural tasks based on synchronization measure phase-locking value (PLV) in an electroencephalogram (EEG) are analyzed. Based on network dissimilarities between the rest and motor imagery tasks, important nodes and channel pairs corresponding to motor tasks are identified. A framework is developed to identify these most reactive channel pairs that form the subject-specific functional network. The identified functional network corresponding to tasks demonstrate significant PLV variation in line with the experiment protocol. With the selection of subject-specific reactive band, these channel pairs provide even higher variation corresponding to tasks. To demonstrate the potential of the developed framework to brain-computer interface, identified network patterns are employed as features for classification of tasks. Analysis performed with the two classes (left-hand and right-hand motor imagery EEG data) showed that the proposed approach yielded better classification results compared to earlier band-power based approaches for single trial analysis.

Despite having a plethora of works, wind speed time-series forecasting capabilities are prone to errors due to their intermittent and nonstationary nature as well as the limited generalization capabilities of forecasting methods for non-Gaussian distributed data. In this article, a hybrid method that consists of elastic variational mode decomposition (eVMD) and forecasting random convolution nodes (fRCN) is proposed to forecast the Gaussian heteroscedastic wind speed time-series. The proposed eVMD algorithm gauges the nonstationary characteristics (complexity) of the wind speed signal and thereafter decomposes the signal into its intrinsic components (ICs) accordingly. The fRCN method rely on local receptive fields to extract features that contribute to the local variations and the global trend in each IC. These features are subsequently learned using extreme learning machines theories. An ensemble unit is employed to learn appropriate weightages for each forecasted IC before yielding the final forecasting values. Suitability of the proposed hybrid method for wind speed forecasting is evaluated via an actual wind speed dataset and comparing against various existing hybrid methods.