Urbanization trends in China reveal a shift in migration patterns, with an increasing number of recent migrants leaving primate cities while secondary cities emerge as attractive destinations. Given China's aging population and intensifying intercity competition for migrants, understanding the factors associated with recent migrants' intentions to leave cities at various levels becomes increasingly important. While spatial equilibrium theory implies that migrants' leaving intentions are shaped by the balance of housing, employment, and amenities, these factors vary hierarchically across city levels. This study examines how these factors differentially shape recent migrants' leaving intentions across primate, secondary, and small cities. Through binary logistic regression of national survey data, we find that recent migrants in secondary cities exhibit lower leaving intentions compared to those in primate and small cities. Further interaction analyses reveal distinct patterns: in primate cities, medium-income migrants are most likely to consider leaving; in secondary cities, rental housing status and hometown residential land ownership more strongly increase leaving intentions compared to primate cities, while medical resource accessibility more significantly reduces leaving intentions compared to small cities. Drawing on spatial equilibrium theory, our analysis suggests that secondary cities appear to achieve an optimal balance: their greater homeownership opportunities serve as a compensatory factor for their lower incomes compared to primate cities, while their superior medical amenity accessibility compensates for their higher housing costs compared to small cities. This paper contributes theoretically by bridging factors in spatial equilibrium theory with the urban hierarchy dynamics proposed by differential urbanization theory. It also offers practical insights for tailoring migration retention policies across city levels and adapting to the transformation of urbanization stages.

Using short-wave infrared wavelength advantages, we demonstrate one-photon fluorescence confocal microscopy of adult mouse brains with penetration depths up to 1.7mm. This is achieved by labeling quantum dots with 1300 nm excitation and 1700 nm emission and detecting them with a single-photon superconducting nanowire detector.

Confounding bias is a crucial problem when applying machine learning to practice, especially in clinical practice. We consider the problem of learning representations independent to multiple biases. In literature, this is mostly solved by purging the bias information from learned representations. We however expect this strategy to harm the diversity of information in the representation, and thus limiting its prospective usage (e.g., interpretation). Therefore, we propose to mitigate the bias while keeping almost all information in the latent representations, which enables us to observe and interpret them as well. To achieve this, we project latent features onto a learned vector direction, and enforce the independence between biases and projected features rather than all learned features. To interpret the mapping between projected features and input data, we propose projection-wise disentangling: a sampling and reconstruction along the learned vector direction. The proposed method was evaluated on the analysis of 3D facial shape and patient characteristics (N = 5011). Experiments showed that this conceptually simple method achieved state-of-the-art fair prediction performance and interpretability, showing its great potential for clinical applications.

Optical microscopy is a valuable tool for in vivo monitoring of biological structures and functions because of its noninvasiveness. However, imaging deep into biological tissues is challenging due to the scattering and absorption of light. Previous research has shown that the two optimal wavelength windows for high-resolution deep mouse brain imaging are around 1300 and 1700 nm. However, one-photon fluorescence imaging in the wavelength region has been highly challenging due to the poor detection efficiency of currently available detectors. To fully utilize this wavelength advantage, we demonstrated here one-photon confocal fluorescence imaging of deep mouse brains with an excitation wavelength of 1310 nm and an emission wavelength within the 1700 nm window. Fluorescence emission at 1700 nm was detected by a custom-built superconducting nanowire single-photon detector (SNSPD) optimized for detection between 1600 nm and 2000 nm with low detection noise and high detection efficiency. With the PEGylated quantum dots and SNSPD both positioned at the optimal imaging window for deep tissue penetration, we demonstrated in vivo one-photon confocal fluorescence imaging at approximately 1.7 mm below the surface of the mouse brain, through the entire cortical column and into the hippocampus region with a low-cost continuous-wave laser source and low excitation power. We further discussed the significance of the staining inhomogeneity in determining the depth limit of one-photon confocal fluorescence imaging. Our work may motivate the further development of long wavelength fluorescent probes, and inspire innovations in high-efficiency, high-gain, and low-noise long wavelength detectors for biological imaging.

This work presents a single-step deep-learning framework for longitudinal image analysis, coined Segis-Net. To optimally exploit information available in longitudinal data, this method concurrently learns a multi-class segmentation and nonlinear registration. Segmentation and registration are modeled using a convolutional neural network and optimized simultaneously for their mutual benefit. An objective function that optimizes spatial correspondence for the segmented structures across time-points is proposed. We applied Segis-Net to the analysis of white matter tracts from N=8045 longitudinal brain MRI datasets of 3249 elderly individuals. Segis-Net approach showed a significant increase in registration accuracy, spatio-temporal segmentation consistency, and reproducibility compared with two multistage pipelines. This also led to a significant reduction in the sample-size that would be required to achieve the same statistical power in analyzing tract-specific measures. Thus, we expect that Segis-Net can serve as a new reliable tool to support longitudinal imaging studies to investigate macro- and microstructural brain changes over time.

Analysis of longitudinal changes in imaging studies often involves both segmentation of structures of interest and registration of multiple timeframes. The accuracy of such analysis could benefit from a tailored framework that jointly optimizes both tasks to fully exploit the information available in the longitudinal data. Most learning- based registration algorithms, including joint optimization approaches, currently suffer from bias due to selection of a fixed reference frame and only support pairwise transformations. We here propose an analytical framework based on an unbiased learning strategy for group-wise registration that simultaneously registers images to the mean space of a group to obtain consistent segmentations. We evaluate the proposed method on longitudinal analysis of a white matter tract in a brain MRI dataset with 2-3 time-points for 3249 individuals, i.e., 8045 images in total. The reproducibility of the method is evaluated on test-retest data from 97 individuals. The results confirm that the implicit reference image is an average of the input image. In addition, the proposed framework leads to consistent segmentations and significantly lower processing bias than that of a pair-wise fixed-reference approach. This processing bias is even smaller than those obtained when translating segmentations by only one voxel, which can be attributed to subtle numerical instabilities and interpolation. Therefore, we postulate that the proposed mean-space learning strategy could be widely applied to learning-based registration tasks. In addition, this group-wise framework introduces a novel way for learning-based longitudinal studies by direct construction of an unbiased within-subject template and allowing reliable and efficient analysis of spatio-temporal imaging biomarkers.