Inverse problems are essential to imaging applications. In this letter, we propose a model-based deep learning network, named FISTA-Net, by combining the merits of interpretability and generality of the model-based Fast Iterative Shrinkage/Thresholding Algorithm (FISTA) and strong regularization and tuning-free advantages of the data-driven neural network. By unfolding the FISTA into a deep network, the architecture of FISTA-Net consists of multiple gradient descent, proximal mapping, and momentum modules in cascade. Different from FISTA, the gradient matrix in FISTA-Net can be updated during iteration and a proximal operator network is developed for nonlinear thresholding which can be learned through end-to-end training. Key parameters of FISTA-Net including the gradient step size, thresholding value and momentum scalar are tuning-free and learned from training data rather than hand-crafted. We further impose positive and monotonous constraints on these parameters to ensure they converge properly. The experimental results, evaluated both visually and quantitatively, show that the FISTA-Net can optimize parameters for different imaging tasks, i.e. Electromagnetic Tomography (EMT) and X-ray Computational Tomography (X-ray CT). It outperforms the state-of-the-art model-based and deep learning methods and exhibits good generalization ability over other competitive learning-based approaches under different noise levels.

Electrical impedance spectroscopy (EIS), as an effective analytical technique for electrochemical system, has shown a wide application for food quality and safety assessment recently. Individual differences of livestock cause high variation in quality of raw meat and fish and their commercialized products. Therefore, in order to obtain the definite quality information and ensure the quality of each product, a fast and on-line detection technology is demanded to be developed to monitor product processing. EIS has advantages of being fast, nondestructive, inexpensive, and easily implemented and shows potential to develop on-line detecting instrument to replace traditional methods to realize time, cost, skilled persons saving and further quality grading. This review outlines the fundamental theories and two common measurement methods of EIS applied to biological tissue, summarizes its application specifically for quality assessment of meat and fish, and discusses challenges and future trends of EIS technology applied for meat and fish quality assessment.

Imaging the bio-impedance distribution of the brain can provide initial diagnosis of acute stroke. This paper presents a compact and non-radiative tomographic modality, i.e. multi-frequency Electromagnetic Tomography (mfEMT), for the initial diagnosis of acute stroke. The mfEMT system consists of 12 channels of gradiometer coils with adjustable sensitivity and excitation frequency. To solve the image reconstruction problem of mfEMT, we propose an enhanced Frequency-Constrained Sparse Bayesian Learning (FC-SBL) to simultaneously reconstruct the conductivity distribution at all frequencies. Based on the Multiple Measurement Vector (MMV) model in the Sparse Bayesian Learning (SBL) framework, FC-SBL can recover the underlying distribution pattern of conductivity among multiple images by exploiting the frequency constraint information. A realistic 3D head model was established to simulate stroke detection scenarios, showing the capability of mfEMT to penetrate the highly resistive skull and improved image quality with FC-SBL. Both simulations and experiments showed that the proposed FC-SBL method is robust to noisy data for image reconstruction problems of mfEMT compared to the single measurement vector model, which is promising to detect acute strokes in the brain region with enhanced spatial resolution and in a baseline-free manner.

In order to detect the QRS complexes locally in the microcontroller-based embedded system, a novel algorithm with lower computation burden is proposed by phase space reconstruction and box-scoring calculation. The method depends on the geometrical property of the ECG signal phase portraits. According to the time-delay method, one segment of the sampled 1-D ECG signal is embedded into a 2-D phase space. The phase space is divided into 2 ^M × 2 ^M grids, and each grid is called a box. The box gridding is implemented by numerically truncating the high M-bits from the digital data. The phase portraits of the sampled ECG signal in the phase space are therefore divided by these boxes. Since the trajectory points corresponding to the R-waves are located relatively far away from the diagonal line, boxes that crossed by the trajectory are flagged and valued according to its distance to the diagonal line. Correspondingly, the box-scoring of the ECG segment is obtained by summarizing the values of the flagged boxes. The location of the R-waves can then be found by finding peak values in the box-scoring time series. The proposed algorithm is suitable for real-time QRS complex detection in microcontroller-based wearable ECG acquisition devices.

When implementing temporary vibration monitoring of mechanical equipment, it is hoped that the installation and disassembly of the sensor will not change the original structure of the equipment. In that case, non-intrusive vibration sensors are preferable. This paper presents a motion induced eddy current sensor that can non-intrusively acquire vibration signal from the surface of metal objects. The sensor consists of a cylindrical permanent magnet and a pancake coil coaxially placed under it. The permanent magnet is used to generate a static magnetic field and the coil is used to pick up the eddy current signal induced by vibration. The results of principle experiments indicate that the output signal of the sensor has a linear relationship with the vibration acceleration/velocity when the measured object is a non-ferromagnetic/ferromagnetic metallic specimen. Two application measurement experiments are also done. When the sensor is placed on the metallic panel of a weighting scale, the ballistocardiogram (BCG) of the human body can be effectively obtained. Another experiment is carried out on a rotor test bed. When the sensor is placed above the bearing support with a fixed lift distance, satisfying vibration signals can be obtained compared with that from the conventional piezoelectric accelerometer. The present works show that the simple-structure, low-cost, non-intrusive vibration sensor (MIECS) can be applied in various fields.

In this paper, the human body communication (HBC) and level crossing sampling (LCS) are combined to design electronics for a wearable ECG. The ECG signals acquired by capacitively coupled electrodes are sampled with LCS in place of conventional synchronous sampling. In order to transmit signals through HBC at low frequencies (100 kHz, 1 MHz), an electric field sensor with high input impedance is adopted as the front end of the HBC receiver. The HBC channel gain is enhanced by more than 30 dB with the electric field sensor. An LCS structure based on the send-on-delta concept is implemented with several discrete components to convert the ECG signals into binary impulses. The converted impulses are modulated by an on-off keying (OOK) modulator and then transmitted via the human body to the receiver. A prototype ECG waist belt is developed with commercially available components and experimentally evaluated. The results indicate that the acquired ECG waveforms exhibit good agreement with regular Ag/AgCl ECG methods. The heartbeat detection using a technique based on the Kadane's algorithm and the power consumption performance of the proposed system are also discussed.