paper_authors: Hyunsoo Choi, Seungman Choi, Peter Menart, Angshuman Deka, Zubin Jacob for:* 这个论文的目的是开发一种新的随机子衰减图像重构算法（SSRI），以优化低信号响应率（SNR）和分辨率较低的光学成像系统。methods:* 该算法利用了常见的成像设备，使其在实际应用中易于适应。results:* 对于多种挑战性的场景，如非常低的SNR水平和较大的相对亮度比，SSRI算法表现出色，超过了已知的理德兹逊-卢西（Richardson-Lucy）减 convolution和CLEAN算法。* SSRI算法在实验图像中成功地估计了点源的位置、亮度和数量，并且在SNR水平低于1.2和子衰减范围内表现出80%-40%的成功率，位偏误在2.5像素以下。Abstract
Overcoming the diffraction limit and addressing low Signal-to-Noise Ratio (SNR) scenarios have posed significant challenges to optical imaging systems in applications such as medical diagnosis, remote sensing, and astronomical observations. In this study, we introduce a novel Stochastic Sub-Rayleigh Imaging (SSRI) algorithm capable of localizing point sources and estimating their positions, brightness, and number in low SNR conditions and within the diffraction limit. The SSRI algorithm utilizes conventional imaging devices, facilitating practical and adaptable solutions for real-world applications. Through extensive experimentation, we demonstrate that our proposed method outperforms established algorithms, such as Richardson-Lucy deconvolution and CLEAN, in various challenging scenarios, including extremely low SNR conditions and large relative brightness ratios. We achieved between 40% and 80% success rate in estimating the number of point sources in experimental images with SNR less than 1.2 and sub-Rayleigh separations, with mean position errors less than 2.5 pixels. In the same conditions, the Richardson-Lucy and CLEAN algorithms correctly estimated the number of sources between 0% and 10% of the time, with mean position errors greater than 5 pixels. Notably, SSRI consistently performs well even in the sub-Rayleigh region, offering a benchmark for assessing future quantum superresolution techniques. In conclusion, the SSRI algorithm presents a significant advance in overcoming diffraction limitations in optical imaging systems, particularly under low SNR conditions, with potential widespread impact across multiple fields like biomedical microscopy and astronomical imaging.
摘要
超过 diffraction limit 和低信号噪比 (SNR) 场景下，光学成像系统在医疗诊断、远程探测和天文观测等领域中受到了重大挑战。在这种研究中，我们介绍了一种新的 Stochastic Sub-Rayleigh Imaging（SSRI）算法，能够在低 SNR 条件下和 diffraction limit 内 Localize 点源并估计其位置、亮度和数量。SSRI 算法可以使用普通的成像设备，提供了实用和适应的解决方案。经过广泛的实验，我们证明了我们的提posed方法在各种挑战性enario中都能够超过Richardson-Lucy 混合和 CLEAN 算法，包括 extremely low SNR 条件下和大relative brightness ratio。我们在实验图像中成功地估计了40%到80%的点源数量，位置误差在2.5 pix 左右，而Richardson-Lucy 和 CLEAN 算法只能在0%到10%的时间内正确地估计点源数量，位置误差大于5 pix。特别是，SSRI 在 sub-Rayleigh 区域中表现良好，为未来 quantum superresolution 技术的评估提供了标准。综上所述，SSRI 算法在光学成像系统中超过 diffraction limit 的能力，特别是在低 SNR 条件下，具有广泛的应用前景，如生物微scopy 和天文成像。
NeuroQuantify – An Image Analysis Software for Detection and Quantification of Neurons and Neurites using Deep Learning
paper_authors: Ka My Dang, Yi Jia Zhang, Tianchen Zhang, Chao Wang, Anton Sinner, Piero Coronica, Joyce K. S. Poon
for: 研究neuronal networks的发展和neuron growth的量化信息
methods: 使用深度学习自动分类 cells和neurites
results: 可以快速和高效地分类 cells和neurites，并提供neurite length和orientation的量化信息Abstract
The segmentation of cells and neurites in microscopy images of neuronal networks provides valuable quantitative information about neuron growth and neuronal differentiation, including the number of cells, neurites, neurite length and neurite orientation. This information is essential for assessing the development of neuronal networks in response to extracellular stimuli, which is useful for studying neuronal structures, for example, the study of neurodegenerative diseases and pharmaceuticals. However, automatic and accurate analysis of neuronal structures from phase contrast images has remained challenging. To address this, we have developed NeuroQuantify, an open-source software that uses deep learning to efficiently and quickly segment cells and neurites in phase contrast microscopy images. NeuroQuantify offers several key features: (i) automatic detection of cells and neurites; (ii) post-processing of the images for the quantitative neurite length measurement based on segmentation of phase contrast microscopy images, and (iii) identification of neurite orientations. The user-friendly NeuroQuantify software can be installed and freely downloaded from GitHub https://github.com/StanleyZ0528/neural-image-segmentation.
摘要
segmenation of cells and neurites in microscopy images of neuronal networks provides valuable quantitative information about neuron growth and neuronal differentiation, including the number of cells, neurites, neurite length and neurite orientation. This information is essential for assessing the development of neuronal networks in response to extracellular stimuli, which is useful for studying neuronal structures, for example, the study of neurodegenerative diseases and pharmaceuticals. However, automatic and accurate analysis of neuronal structures from phase contrast images has remained challenging. To address this, we have developed NeuroQuantify, an open-source software that uses deep learning to efficiently and quickly segment cells and neurites in phase contrast microscopy images. NeuroQuantify offers several key features: (i) automatic detection of cells and neurites; (ii) post-processing of the images for the quantitative neurite length measurement based on segmentation of phase contrast microscopy images, and (iii) identification of neurite orientations. The user-friendly NeuroQuantify software can be installed and freely downloaded from GitHub https://github.com/StanleyZ0528/neural-image-segmentation.Here's the word-for-word translation of the text into Simplified Chinese: cells 和 neurites 的分 segmentation 在 neuronal networks 的 microscopy 图像中提供了有价值的量化信息，包括细胞数量、辐化长度、辐化方向等。这些信息对于研究 neuronal networks 的发展响应 extracellular stimuli 非常重要，这些信息可以用于研究 neuronal structures，例如研究 neuodegenerative diseases 和 pharmaceuticals。然而，从 phase contrast 图像中自动和准确地分析 neuronal structures 一直是一个挑战。为解决这个问题，我们已经开发了 NeuroQuantify，一个开源的软件，使用深度学习来快速和高效地分 segmentation 细胞和 neurites 在 phase contrast microscopy 图像中。NeuroQuantify 提供了多个关键特性： (i) 自动检测细胞和辐化； (ii) 根据 segmentation 的图像进行后处理，以获取辐化长度的量化测量；以及 (iii) 辐化方向的识别。用户友好的 NeuroQuantify 软件可以在 GitHub 上免费下载 https://github.com/StanleyZ0528/neural-image-segmentation。
Impact of Data Synthesis Strategies for the Classification of Craniosynostosis
paper_authors: Matthias Schaufelberger, Reinald Peter Kühle, Andreas Wachter, Frederic Weichel, Niclas Hagen, Friedemann Ringwald, Urs Eisenmann, Jürgen Hoffmann, Michael Engel, Christian Freudlsperger, Werner Nahm
results: 组合 SSM 和 GAN 达到了高于 0.96 的准确率和高于 0.95 的 F1 分数在未看到的测试集上。与训练在临床数据上的差异小于 0.01。包括第二个图像模式可以提高分类性能。So the three key points are:1. The paper is written to assess and classify craniosynostosis using photogrammetric surface scans.2. The methods used include three different synthetic data sources: a statistical shape model, a generative adversarial network, and image-based principal component analysis.3. The results show that a combination of these synthetic data sources can achieve high accuracy and F1 score (over 0.95 and 0.96 respectively) on unseen test sets, with little difference between training on synthetic data and clinical data.Abstract
Introduction: Photogrammetric surface scans provide a radiation-free option to assess and classify craniosynostosis. Due to the low prevalence of craniosynostosis and high patient restrictions, clinical data is rare. Synthetic data could support or even replace clinical data for the classification of craniosynostosis, but this has never been studied systematically. Methods: We test the combinations of three different synthetic data sources: a statistical shape model (SSM), a generative adversarial network (GAN), and image-based principal component analysis for a convolutional neural network (CNN)-based classification of craniosynostosis. The CNN is trained only on synthetic data, but validated and tested on clinical data. Results: The combination of a SSM and a GAN achieved an accuracy of more than 0.96 and a F1-score of more than 0.95 on the unseen test set. The difference to training on clinical data was smaller than 0.01. Including a second image modality improved classification performance for all data sources. Conclusion: Without a single clinical training sample, a CNN was able to classify head deformities as accurate as if it was trained on clinical data. Using multiple data sources was key for a good classification based on synthetic data alone. Synthetic data might play an important future role in the assessment of craniosynostosis.
摘要
方法：我们测试了三种不同的生成数据源：统计学形态模型（SSM）、生成对抗网络（GAN）和图像基于主成分分析（PCA），用于基于 convolutional neural network（CNN）的颅部缺陷分类。CNN只在生成数据上训练，但VALIDATE和测试在临床数据上进行验证。结果：SSM和GAN的组合实现了在未看到的测试集上的准确率高于0.96和F1分数高于0.95。与训练在临床数据上的差异小于0.01。包括第二个图像特征提高了所有数据源的分类性能。结论：没有任何临床训练样本，CNN仍可以准确地将头形缺陷分类为临床数据。使用多种数据源是针对synthetic数据alone的分类的关键。synthetic数据可能在未来对颅部缺陷的评估中扮演一个重要的角色。