Digital Microfluidics for Manipulation and Analysis of a Single Cell

The basic structural and functional unit of a living organism is a single cell. To understand the variability and to improve the biomedical requirement of a single cell, its analysis has become a key technique in biological and biomedical research. With a physical boundary of microchannels and microstructures, single cells are efficiently captured and analyzed, whereas electric forces sort and position single cells. Various microfluidic techniques have been exploited to manipulate single cells through hydrodynamic and electric forces. Digital microfluidics (DMF), the manipulation of individual droplets holding minute reagents and cells of interest by electric forces, has received more attention recently. Because of ease of fabrication, compactness and prospective automation, DMF has become a powerful approach for biological application. We review recent developments of various microfluidic chips for analysis of a single cell and for efficient genetic screening. In addition, perspectives to develop analysis of single cells based on DMF and emerging functionality with high throughput are discussed.     

Blocked crystallization in capped ultrathin polymer films studied by molecular simulations

The crystallization of capped ultrathin polymer films is closely dependent on film thickness and interfacial interaction. Using dynamic Monte Carlo simulations, the crystallization behaviors of polymer films confined between two substrates were investigated. The crystallization rate of confined polymers is reduced with high interfacial interactions. Above a critical strength of interfacial interaction, polymer crystallization in the thin film is inhibited within the simulation time scales. An increase in film thickness leads to a rise in critical interfacial interaction. In thicker films, the chains have more space to change conformation to form crystal stems. In addition, there are fewer absorbed segments in confined chains for the thicker films, and thus the chains have stronger ability to adjust their conformation. Therefore an increase in film thickness can cause a reduction in the entropic barrier required for the formation of crystals and thus an increase in the critical interfacial interaction. © 2018 Society of Chemical Industry     

Multiply Charged Conjugated Polyelectrolytes as a Multifunctional Interlayer for Efficient and Scalable Perovskite Solar Cells

A series of anionic conjugated polyelectrolytes (CPEs) is synthesized based on poly(fluorene-co-phenylene) by varying the side-chain ionic density from two to six per repeat units (MPS2-TMA, MPS4-TMA, and MPS6-TMA). The effect of MPS2, 4, 6-TMA as interlayers on top of a hole-extraction layer of poly(bis(4-phenyl)-2,4,6-trimethylphenylamine (PTAA) is investigated in inverted perovskite solar cells (PeSCs). Owing to the improved wettability of perovskites on hydrophobic PTAA with the CPEs, the PeSCs with CPE interlayers demonstrate a significantly enhanced device performance, with negligible device-to-device dependence relative to the reference PeSC without CPEs. By increasing the ionic density in the MPS-TMA interlayers, the wetting, interfacial defect passivation, and crystal growth of the perovskites are significantly improved without increasing the series resistance of the PeSCs. In particular, the open-circuit voltage increases from 1.06 V for the PeSC with MPS2-TMA to 1.11 V for the PeSC with MPS6-TMA. The trap densities of the PeSCs with MPS2,4,6-TMA are further analyzed using frequency-dependent capacitance measurements. Finally, a large-area (1 cm²) PeSC is successfully fabricated with MPS6-TMA, showing a power conversion efficiency of 18.38% with negligible hysteresis and a stable power output under light soaking for 60 s.     

基于深度学习的秀丽隐杆线虫显微图像分割方法

利用计算机实现自动、准确的秀丽隐杆线虫(C.elegans)的各项形态学参数分析,至关重要的是从显微图像上分割出线虫体态,但由于显微镜下的图像噪声较多,线虫边缘像素与周围环境相似,而且线虫的体态具有鞭毛和其他附着物需要分离,多方面因素导致设计一个鲁棒性的C.elegans分割算法仍然面临着挑战。针对这些问题,提出了一种基于深度学习的线虫分割方法,通过训练掩模区域卷积神经网络(Mask R-CNN)学习线虫形态特征实现自动分割。首先,通过改进多级特征池化将高级语义特征与低级边缘特征融合,结合大幅度软最大损失(LMSL)损失算法改进损失计算;然后,改进非极大值抑制;最后,引入全连接融合分支等方法对分割结果进行进一步优化。实验结果表明,相比原始的Mask R-CNN,该方法平均精确率(AP)提升了4.3个百分点,平均交并比(mIOU)提升了4个百分点。表明所提出的深度学习分割方法能够有效提高分割准确率,在显微图像中更加精确地分割出线虫体。