Blood cell counting based on U-Net++ and YOLOv5

Clinical information about a variety of disorders is available through blood cell counting, which is usually done by manual methods. However, manual methods are complex, time-consuming and susceptible to the subjective experience of inspectors. Although many efforts have been made to develop automated blood cell counting algorithms, the complexity of blood cell distribution and the highly overlapping nature of some red blood cells (RBCs) remain significant challenges that limit the improvement of analytical accuracy. Here, we proposed an end-to-end method for blood cell counting based on deep learning. Firstly, U-Net++ was used to segment the whole blood cell image into several regions of interest (ROI), and each ROI contains only one single cell or multiple overlapping cells. Subsequently, YOLOv5 was used to detect blood cells in each ROI. Specifically, we proposed several strategies, including fine classification of RBCs, adaptive adjustment for non-maximal suppression (NMS) threshold and blood cell morphology constraints to improve the accuracy of detection. Finally, the detection outcomes for each ROI were combined and superimposed. The results show that our method can effectively address the issue of high overlap and precisely segment and detect blood cells, with a 98.18% accuracy rate for blood cell counting.     

An improved U-Net for cell confluence estimation

Cell confluence is an important metric to determine the growth and the best harvest time of adherent cells. At present, the evaluation of cell confluence mainly relies on experienced labor, and thus it is not conducive to the automated cell culture. In this paper, we proposed an improved U-Net algorithm (called DU-Net) for the segmentation of adherent cells. First, the general convolution was replaced by the dilated convolution to expand the receptive fields for feature extraction. Then, the convolutional layers were combined with the batch normalization layers to reduce the dependence of the network on initialization. As a result, the segmentation accuracy and F1-score of the proposed DU-Net for adherent cells with low confluence (<50%) reached 96.94% and 93.87%, respectively, and for those with high confluence (≥50%), they reached 98.63% and 98.98%, respectively. Further, the paired t-test results showed that the proposed DU-Net was statistically superior to the traditional U-Net algorithm.     

Entropy-based active sparse subspace clustering

Multimedia Tools and Applications - Sparse Subspace Clustering (SSC) is widely used in data mining and machine learning. Some studies have been developed to add pairwise constraints as side...     

Hatching egg classification based on CNN with channel weighting and joint supervision

Multimedia Tools and Applications - Convolutional neural networks (CNNs) show state-of-the-art performance in tackling a variety of visual tasks. It is expected that a CNN can be applied to the...     

Hatching eggs classification based on deep learning

In order to realize the fertility detection and classification of hatching eggs, a method based on deep learning is proposed in this paper. The 5-days hatching eggs are divided into fertile eggs, dead eggs and infertile eggs. Firstly, we combine the transfer learning strategy with convolutional neural network (CNN). Then, we use a network of two branches. In the first branch, the dataset is pre-trained with the model trained by AlexNet network on large-scale ImageNet dataset. In the second branch, the dataset is directly trained on a multi-layer network which contains six convolutional layers and four pooling layers. The features of these two branches are combined as input to the following fully connected layer. Finally, a new model is trained on a small-scale dataset by this network and the final accuracy of our method is 99.5%. The experimental results show that the proposed method successfully solves the multi-classification problem in small-scale dataset of hatching eggs and obtains high accuracy. Also, our model has better generalization ability and can be adapted to eggs of diversity.     

Dual-demodulation large-scope high-sensitivity refractive index sensor based on twin-core PCF

In this paper, a refractive index (RI) sensor based on the twin-core photonic crystal fiber (TC-PCF) is presented. Introducing the rectangular array in the core area makes the PCF possible to obtain high birefringence and low confinement loss over the wavelength range from 0.6 μm to 1.7 μm. Therefore, the core region can enhance the interaction between the core mode and the filling material. We studied theoretically the evolution characteristics of the birefringence and operating wavelength corresponding to the strongest polarization point under the condition of filling the rectangular array with RI matching fluid range from 1.33 to 1.41. Simulation results reveal that the proposed TC-PCF has opposite evolutions of change rates between the B and wavelength, and the maximum RI sensing sensitivities of 1.809×10-2 B/RIU and 8 700 nm/RIU at low and high RI infill are obtained respectively, which means that the TC-PCF features of dual-parameter demodulation for the RI sensing can maintain a high refractive index sensing sensitivity within a large scope of RI ranging from 1.33 to 1.41. Compared with the results of single-parameter demodulation, it is an optimized method to improve the sensitivity of low refractive index sensors, which has great application potency in the field of biochemical sensing and detection.     

Spectral analysis of photo-induced delayed luminescence from mesenchymal stem cells for label-free cell viability assessment

In this work, the spectral properties of the photo-induced delayed luminescence (DL) from mesenchymal stem cells (MSCs) and their correlation with cell viability were investigated using the single photon counting combined with band-pass filters. The results show that the DL of MSCs has a broad spectral distribution, which covers from 300 nm to 650 nm at least. The DL spectrum is not evenly distributed, but mainly distributed in the range from 400 nm to 550 nm. In addition, the DL spectral distribution remains stable during the DL decay process. Compared with the DL spectra of MSCs with high viability (>80%), those of MSCs with low viability (<30%) show a significant red-shift, referring to the increase in the proportion of 572—650 nm band and the decrease in the proportions of both 315—436 nm band and 413—500 nm band. Furthermore, the degree of the DL spectral red-shift exhibits a monotonous change as MSCs’ viability decreases, and thus can be used as an important indicator for the cell viability assessment.     

Intermodal interference based refractive index sensor employing elliptical core photonic crystal fiber

A refractive index (RI) sensor based on elliptical core photonic crystal fiber (EC-PCF) has been proposed. The asymmetric elliptical core introduces the polarization-dependent characteristics of the fiber core modes. The performances of intermodal interference between the intrinsic polarization fiber core modes are investigated by contrast in two interferometers based on the Mach-Zehnder (M-Z) and Sagnac interference model. In addition, the RI sensing characteristics of the two interferometers are studied by successively filling the three layers air holes closest to the elliptical core in the cladding. The results show that the M-Z interference between LP01 and LP11 mode in the same polarized direction is featured with the incremental RI sensing sensitivity as the decreasing interference length, and the infilled scope around the elliptical core has a weak correlation with the RI sensing sensitivity. Due to the high birefringence of LP11 mode, the Sagnac interferometer has better RI sensing performance, the maximum RI sensing sensitivity of 12 000 nm/RIU is achieved under the innermost one layer air holes infilled with RI matching liquid of RI=1.39 at the pre-setting EC-PCF length of 12 cm, which is two orders of magnitude higher than the M-Z interferometer with the same fiber length. The series of theoretical optimized analysis would provide guidance for the applications in the field of biochemical sensing.     

一种基于嵌入式平台的非接触式心率监测系统研究

心率的长期监测对心血管疾病的预防和治疗具有重要意义。当前心率检测常用的监护仪、心电图机、智能手表和运动手环等均属于接触式测量装置,长期佩戴易产生压痕,甚至给使用者带来不适。在非接触测量方面,远程光电容积脉搏波描记法(remote photoplethysmography, rPPG)可以通过分析面部视频获取心率,是一种很有潜力的心率长期监测方法。目前,绝大多数对于rPPG的研究都在使用电脑做数据分析,体积过大不易摆放,难以满足医学临床和家庭日常使用的需求。针对这一问题,本文尝试在嵌入式平台上依据rPPG原理实现心率监测。监测系统主要由树莓派4B开发板、相机和触摸屏组成。采用AdaBoost算法实现人脸识别与追踪,选取额头和脸颊作为感兴趣区域(range of interest, ROI),利用巴特沃斯带通滤波去噪,根据POS模型提取BVP波形,对来自不同ROI的BVP波形做盲源分离得到最终的脉搏波,最后利用能谱分析计算心率。实验结果表明本文所研究的系统具有与PC端相同的心率检测准确性和鲁棒性。本文的研究成果可以为心率长期监测设备的小型化和普及做出自己的贡献,也可以为智慧医疗中的远程监测...     

Resource Allocation Strategy of SWIPT Relay Under General Interference

The interference often results in a low rate of the wireless relay systems. The recent research on radio frequency signal energy harvesting makes it possible to utilize the interference energy. Based on the widely applicable time switching relay operation strategy and decode-and-forward (DF) relay modes, this paper studies the resource allocation strategy of the simultaneous wireless information and power transfer relay system under the general interference. The h₂ method is proposed for resource allocation. It establishes a resource allocation coordinate system, and divides the energy harvesting area and the DF area by using the h₂ method. Through derivation, the relationship between the non-interruption probability of the system and the line h₂ is found. Aiming at maximizing the non-interruption probability, the golden splitting method is used to solve the optimal value. The numerical simulation results demonstrate that the h₂ method can effectively improve the non-interruption probability of the system.