EnGAN: Enhancement Generative Adversarial Network in Medical Image Segmentation

Deep learning has been widely used in medical image segmentation, although the accuracy is affected by the problems of small sample space, data imbalance, and cross-device differences. Aiming at such issues, a enhancement GAN network is proposed by using the domain transferring of the adversarial generation network to enhance the original medical images. Specifically, based on retaining the transferability of the original GAN network, a new optimizer is added to generate a sample space with a continuous distribution, which can be used as the target domain of the original image transferring. The optimizer back-propagates the labels of the supervised data set through the segmentation network and maps the discrete distribution of the labels to the continuous image distribution, which has a high similarity to the original image but improves the segmentation efficiency.On this basis, the optimized distribution is taken as the target domain, and the generator and discriminator of the GAN network are trained so that the generator can transfer the original image distribution to the target distribution. extensive experiments are conducted based on MRI, CT, and ultrasound data sets. The experimental results show that, the proposed method has a good generalization effect in medical image segmentation, even when the data set has limited sample space and data imbalance to a certain extent.

How to Reveal the Meaning of Space in Vernacular Architecture?

Vernacular architecture is a modest style of building used to maintain the balance of human relations with nature. This architectural style is specific to a region and passed down from one generation to another to embody cultural values. However, its development is currently facing globalization and modernization challenges, thereby leading to a gradual shift of this ancestral heritage to modern buildings. Change is unavoidable due to continuous evolution, however, the meaning inherent architecture buildings need to be maintained because it contains the cultural and social values of the associated local community. Furthermore, vernacular building space is a place for social activities and contains historical meaning applicable to modern buildings. Its functionality responds to changes and the needs of times while maintaining the local essence. Therefore, this research aims to determine the suitable method needed to reveal the meaning of vernacular architectural space. Data were collected from the conscious mind of space users through in-depth interviews by applying epoche, which were further reduced, categorized, and integrated to determine its meaning. The data collected through a literature review were analyzed using the content analysis method. The results showed that transcendental phenomenology is the right method to determine the meaning of vernacular architectural space. Based on the results, it is concluded that the meaning passed down from one generation to another could be expressed through the conscious experience of space users. Furthermore, transcendental phenomenology helped reveal the meaning without the intervention of the author’s knowledge, therefore it is unbiased and applicable in modern buildings.

Group feature selection for enhancing information gain in MRI reconstruction

Magnetic resonance imaging (MRI) has revolutionized the radiology. As a leading medical imaging modality, MRI not only visualizes the structures inside body, but also produces functional imaging. However, due to the slow imaging speed constrained by the MR physics, MRI cost is expensive, and patient may feel not comfortable in a scanner for a long time. Parallel MRI has accelerated the imaging speed through the sub-Nyquist sampling strategy and the missing data are interpolated by the multiple coil data acquired. Kernel learning has been used in the parallel MRI reconstruction to learn the interpolation weights and re-construct the undersampled data. However, noise and aliasing artifacts still exist in the reconstructed image and a large number of auto-calibration signal lines are needed. To further improve the kernel learning-based MRI reconstruction and accelerate the speed, this paper proposes a group feature selection strategy to improve the learning performance and enhance the reconstruction quality. An explicit kernel mapping is used for selecting a subset of features which contribute most to estimate the missing k-space data. The experimental results show that the learning behaviours can be better predicted and therefore the reconstructed image quality is improved.

Microdosimetric Investigation and a Novel Model of Radiosensitization in the Presence of Metallic Nanoparticles

Auger cascades generated in high atomic number nanoparticles (NPs) following ionization were considered a potential mechanism for NP radiosensitization. In this work, we investigated the microdosimetric consequences of the Auger cascades using the theory of dual radiation action (TDRA), and we propose the novel Bomb model as a general framework for describing NP-related radiosensitization. When triggered by an ionization event, the Bomb model considers the NPs that are close to a radiation sensitive cellular target, generates dense secondary electrons and kills the cells according to a probability distribution, acting like a “bomb.” TDRA plus a distance model were used as the theoretical basis for calculating the change in α of the linear-quadratic survival model and the relative biological effectiveness (RBE). We calculated these quantities for SQ20B and Hela human cancer cells under 250 kVp X-ray irradiation with the presence of gadolinium-based NPs (AGuIXTM), and 220 kVp X-ray irradiation with the presence of 50 nm gold NPs (AuNPs), respectively, and compared with existing experimental data. Geant4-based Monte Carlo (MC) simulations were used to (1) generate the electron spectrum and the phase space data of photons entering the NPs and (2) calculate the proximity functions and other related parameters for the TDRA and the Bomb model. The Auger cascade electrons had a greater proximity function than photoelectric and Compton electrons in water by up to 30%, but the resulting increases in α were smaller than those derived from experimental data. The calculated RBEs cannot explain the experimental findings. The relative increase in α predicted by TDRA was lower than the experimental result by a factor of at least 45 for SQ20B cells with AGuIX under 250 kVp X-ray irradiation, and at least four for Hela cells with AuNPs under 220 kVp X-ray irradiation. The application of the Bomb model to Hela cells with AuNPs under 220 kVp X-ray irradiation indicated that a single ionization event for NPs caused by higher energy photons has a higher probability of killing a cell. NPs that are closer to the cell nucleus are more effective for radiosensitization. Microdosimetric calculations of the RBE for cell death of the Auger electron cascade cannot explain the experimentally observed radiosensitization by AGuIX or AuNP, while the proposed Bomb model is a potential candidate for describing NP-related radiosensitization at low NP concentrations.

Dynamics and length scales in vertical convection of liquid metals

Using complementary experiments and direct numerical simulations, we study turbulent thermal convection of a liquid metal (Prandtl number $\textit {Pr}\approx 0.03$ ) in a box-shaped container, where two opposite square sidewalls are heated/cooled. The global response characteristics like the Nusselt number ${\textit {Nu}}$ and the Reynolds number $\textit {Re}$ collapse if the side height $L$ is used as the length scale rather than the distance $H$ between heated and cooled vertical plates. These results are obtained for various Rayleigh numbers $5\times 10^3\leq {\textit {Ra}}_H\leq 10^8$ (based on $H$ ) and the aspect ratios $L/H=1, 2, 3$ and $5$ . Furthermore, we present a novel method to extract the wind-based Reynolds number, which works particularly well with the experimental Doppler-velocimetry measurements along vertical lines, regardless of their horizontal positions. The extraction method is based on the two-dimensional autocorrelation of the time–space data of the vertical velocity.

Joint reconstruction framework of compressed sensing and nonlinear parallel imaging for dynamic cardiac magnetic resonance imaging

Compressed Sensing (CS) and parallel imaging are two promising techniques that accelerate the MRI acquisition process. Combining these two techniques is of great interest due to the complementary information used in each. In this study, we proposed a novel reconstruction framework that effectively combined compressed sensing and nonlinear parallel imaging technique for dynamic cardiac imaging. Specifically, the proposed method decouples the reconstruction process into two sequential steps: In the first step, a series of aliased dynamic images were reconstructed from the highly undersampled k-space data using compressed sensing; In the second step, nonlinear parallel imaging technique, i.e. nonlinear GRAPPA, was utilized to reconstruct the original dynamic images from the reconstructed k-space data obtained from the first step. In addition, we also proposed a tailored k-space down-sampling scheme that satisfies both the incoherent undersampling requirement for CS and the structured undersampling requirement for nonlinear parallel imaging. The proposed method was validated using four in vivo experiments of dynamic cardiac cine MRI with retrospective undersampling. Experimental results showed that the proposed method is superior at reducing aliasing artifacts and preserving the spatial details and temporal variations, compared with the competing k-t FOCUSS and k-t FOCUSS with sensitivity encoding methods, with the same numbers of measurements.

Next generation earth-to-space telecommand coding and synchronization: ground system design, optimization and software implementation

The Consultative Committee for Space Data Systems, followed by all national and international space agencies, has updated the Telecommand Coding and Synchronization sublayer to introduce new powerful low-density parity-check (LDPC) codes. Their large coding gains significantly improve the system performance and allow new Telecommand services and profiles with higher bit rates and volumes. In this paper, we focus on the Telecommand transmitter implementation in the Ground Station baseband segment. First, we discuss the most important blocks and we focus on the most critical one, i.e., the LDPC encoder. We present and analyze two techniques, one based on a Shift Register Adder Accumulator and the other on Winograd convolution both exploiting the block circulant nature of the LDPC matrix. We show that these techniques provide a significant complexity reduction with respect to the usual encoder mapping, thus allowing to obtain high uplink bit rates. We then discuss the choice of a proper hardware or software platform, and we show that a Central Processing Unit-based software solution is able to achieve the high bit rates requested by the new Telecommand applications. Finally, we present the results of a set of tests on the real-time software implementation of the new system, comparing the performance achievable with the different encoding options.