Liang Wu

This study presents a novel embedded wearable Parkinson’s states detection system. This system integrates data collection using inertial measurement unit (IMU), feature extraction, machine learning classification algorithm, and Parkinson’s states warning cue into a custom printed circuit board. Fifty Parkinson’s disease (PD) patients experiencing lower limb movement disturbance and fifty age matched healthy controls were enrolled in this study, with two IMU sensors attached to the outside of each ankle and asked to perform three tasks used in clinical to evaluate the severity of PD bradykinesia. In the first phase, the performance of seven state-of-the-art machine learning algorithms to classify lower limb movement symptoms from PD patients with healthy controls was evaluated. In the second phase, the support vector machine with radial basis function classifier demonstrated the best classification performance for the left heel tapping under the knee flexion state task, leading to its selection for microcontroller implantation. The proposed wearable Parkinson’s states detection system includes the microcontroller, IMU sensor, charging module, result indicator module, battery, power switch and reset module. The total area of the printed circuit board is nearly 6.1 cm×3.5 cm, the whole detection device weights 28.2 g, the feature extraction and model classification time needed is about 0.103 seconds, and the power consumption during the active mode is 0.195 W. The embedded machine learning on local microcontroller eliminates the need for data offloading or cloud-based interference, allowing for direct and timely inference of detection result for new subjects.

With the rapid development of synthetic aperture radar (SAR) technology, SAR remote sensing has a wide range of applications in fields such as marine surveillance and sea rescue. Currently, the SAR ship detection model based on deep learning suffers from the problems of low detection in real-time and low detection accuracy. In order to solve the above problems, this paper proposes a lightweight SAR ship detection network (EGTB-Net) based on transformer and feature enhancement. Firstly, we design a novel Ghost-ECA model as the backbone network of EGTB-Net, which reduces the number of parameters of the model and enhances the ability to identify key feature information at the same time. Then, we incorporate the transformer block in the backbone network to capture long-range dependencies, enrich contextual information, and improve the network's ability to capture different types of local information. Finally, we adopt a new SIoU loss function, which is used to solve the direction problem of mismatch between the real frame and the predicted frame and improve the network's ability to localize ship targets. The experimental results on the SAR-Ship-Dataset show that the mean average precision (mAP) of the method is 94.83%, the detection speed is 61 FPS, and the model size is only 5.94 M, while the model has excellent anti-interference ability.

Alzheimer's disease (AD) is a degenerative mental disorder of the central nervous system that affects people's ability of daily life. Unfortunately, there is currently no known cure for AD. Thus, the early detection of AD plays a key role in preventing and controlling its progression. Magnetic resonance imaging (MRI)-based measures of cerebral atrophy are regarded as valid markers of the AD state. As one of representative methods for measuring brain atrophy, image registration technique has been widely adopted for AD diagnosis. However, AD detection is sensitive to the accuracy of image registration. To address this problem, an AD assistant diagnosis framework based on joint registration and classification is proposed. Specifically, in order to capture more local deformation information, we propose a novel patch-based joint brain image registration and classification network (RClaNet) to estimate the local dense deformation fields (DDF) and disease risk probability maps that explain high-risk areas for AD patients. RClaNet consists of a registration network and a classification network, in which the deformation field from registration network is fed into the classification network to enhance the prediction accuracy of the disease. Then, the exponential distance weighting method is used to obtain the global DDF and the global disease risk probability map, and it can remove grid-like artifacts by uniformly weighting method. Finally, the global classification network uses the global disease risk probability map for the early detection of AD. We evaluate the proposed method on the OASIS-3, AIBL and ADNI datasets, and experimental results show that the proposed RClaNet achieves superior registration performances than several state-of-the-art methods. Early diagnosis of AD using the global disease risk probability map also yielded competitive results. To demonstrate the generality, we also evaluate the proposed method on a COVID-19 dataset and achieve decent registration and classification results. These experiments prove that the deformation information in the registration process can be used to characterize subtle changes of degenerative diseases and further assist clinicians in diagnosis.

To improve the detection rate of pulmonary nodules in early lung cancer screening, a low-dose CT pulmonary nodule detection algorithm based on 3D CNN-CapsNet (3D convolution neural network and capsule network) was presented. However, the convolution kernel size of the traditional CNN is relatively simple at each layer, and it is difficult to obtain more abundant features, which is not effective for medical images with a hierarchical structure and does not fully consider the spatial information of medical sequence data. CapsNet is a new network architecture that can be used to classify, using a group of neurons as a capsule to replace the traditional neural networks, it may be made to the attribute information and spatial feature extraction. The network structure we designed includes FCN and CapsNet. First, the convolution kernels of different sizes are used to extract features at different scales. Then enter the initial feature map to obtain the first part into the designed CapsNet to get the final classification result. Through the experimental verification of the ELCAP database, the nodule detection rate is 95.19%, the sensitivity is 92.31%, the specificity is 98.08% and the F1-score is 0.95 which are much better than other baseline methods.

Patch-based medical image registration has been well explored in recent decades. However, the patch fusion process can generate grid-like artifacts along the edge of patches for the following two reasons: firstly, in order to ensure the same size of input and output, zero-padding is used, which causes uncertainty in the edges of the output feature map during the feature extraction process; secondly, the sliding window extraction patch with different strides will result in different degrees of grid-like artifacts. In this paper, we propose an exponential-distance-weighted (EDW) method to remove grid-like artifacts. To consider the uncertainty of predictions near patch edges, we used an exponential function to convert the distance from the point in the overlapping regions to the center point of the patch into a weighting coefficient. This gave lower weights to areas near the patch edges, to decrease the uncertainty predictions. Finally, the dense displacement field was obtained by this EDW weighting method. We used the OASIS-3 dataset to evaluate the performance of our method. The experimental results show that the proposed EDW patch fusion method removed grid-like artifacts and improved the dice similarity coefficient superior to those of several state-of-the-art methods. The proposed fusion method can be used together with any patch-based registration model.

The precise segmentation of brain tumor images is a vital step towards accurate diagnosis and effective treatment of brain tumors. Magnetic Resonance Imaging (MRI) can generate brain images without tissue damage or skull artifacts, providing important discriminant information for clinicians in the study of brain tumors and other brain diseases. In this paper, we survey the field of brain tumor MRI images segmentation. Firstly, we present the commonly used databases. Then, we summarize multi-modal brain tumor MRI image segmentation methods, which are divided into three categories: conventional segmentation methods, segmentation methods based on classical machine learning methods, and segmentation methods based on deep learning methods. The principles, structures, advantages and disadvantages of typical algorithms in each method are summarized. Finally, we analyze the challenges, and suggest a prospect for future development trends.

Magnetic resonance (MR) images often suffer from random noise pollution during image acquisition and transmission, which impairs disease diagnosis by doctors or automated systems. In recent years, many noise removal algorithms with impressive performances have been proposed. In this work, inspired by the idea of deep learning, we propose a denoising method named 3D‐Parallel‐RicianNet, which will combine global and local information to remove noise in MR images. Specifically, we introduce a powerful dilated convolution residual (DCR) module to expand the receptive field of the network and to avoid the loss of global features. Then, to extract more local information and reduce the computational complexity, we design the depthwise separable convolution residual (DSCR) module to learn the channel and position information in the image, which not only reduces parameters dramatically but also improves the local denoising performance. In addition, a parallel network is constructed by fusing the features extracted from each DCR module and DSCR module to improve the efficiency and reduce the complexity for training a denoising model. Finally, a reconstruction (REC) module aims to construct the clean image through the obtained noise deviation and the given noisy image. Due to the lack of ground‐truth images in the real MR dataset, the performance of the proposed model was tested qualitatively and quantitatively on one simulated T1‐weighted MR image dataset and then expanded to four real datasets. The experimental results show that the proposed 3D‐Parallel‐RicianNet network achieves performance superior to that of several state‐of‐the‐art methods in terms of the peak signal‐to‐noise ratio, structural similarity index, and entropy metric. In particular, our method demonstrates powerful abilities in both noise suppression and structure preservation.