Optical microscopy is an important tool for biomedical research at submicron resolution. As the fine structures of biological tissues are complex and varied, it is a recognized challenge to observe them optically in three-dimensional space. Neurons with complex morphology are the basic functional units of the brain, and how to obtain their complete structure poses a great challenge to the existing technology. The diameter of the cytosol of fluorescently labeled neurons is about 10-20 μm, and the axons and numerous branching fibers extending out from the cytosol are only 0.2-0.5 μm in diameter, and most of them are projected to different brain regions throughout the brain. The cytosol and fibers differ in brightness by 2-3 orders of magnitude, and the spatial distribution is often intertwined. Detecting weak fluorescent signals on axons in the presence of interfering peripheral cytosol is like looking at small stars around a bright sun. In this case, the traditional optical chromatography method is difficult to realize.
On March 1, 2021, a team of Faculty Member Qingming Luo from the Functional Laboratory of Biomedical Photonics at the National Research Center for Optoelectronics in Wuhan, Huazhong University of Science and Technology (NRCP), China, published the article High- definition imaging using line-illumination in Nature Methods in the form of a long paper. definition imaging using line-illumination modulation microscopy, developing line-illumination modulation microscopy and realizing high-definition imaging.
Ching-Ming Luo's team proposed a new method of high-definition, high-throughput optical chromatography microscopy, Line illumination microscopy, LiMo, to achieve high-definition imaging. strong> LiMo). They cleverly use the Gaussian distribution of line illumination intensity as a natural modulation mode of illumination intensity, where different illumination intensities modulate only the signals on the focal plane, but not the out-of-focus background signals. Therefore, with multi-line detection, these signals modulated by different intensities can be recorded at once, and the same out-of-focus background signals can be removed in a single step of linear computation to obtain a clear optical tomography image of the focal plane. Theoretically derived LiMo background signals have faster attenuation coefficients compared to the classical methods used in biomedical applications, such as laser point*** scanning microscopy, two-photon excitation fluorescence microscopy, laser line*** scanning microscopy, and structured light optical tomography. This conclusion is also supported by experimental tests, where the signal-to-background ratio of the LiMo method is improved by 1-2 orders of magnitude compared to the above mentioned classical methods. This method requires only a simple multi-line probe line illumination optical path, overcoming the inherent defects of residual modulation artifacts in traditional structured light illumination imaging, and eliminating the need for multiple imaging sessions to obtain the required data, as well as having the advantage of high throughput in line scanning for a wide range of samples, which solves the problem of traditional fluorescence microscopy optical chromatography imaging methods not being able to simultaneously take into account the problems of high-resolution, high-throughput, and high-definition. It can be said that this method fully embodies the beauty of simplicity in technology, both in terms of optical path and algorithm. sectioning tomography ( HD-fMOST), which elevates whole-brain optical imaging from high-resolution to high-definition. While whole-brain optical imaging has brought unprecedented richness of detail to biomedical research in recent years, it has also created the difficulty of huge amounts of data. In order to solve this difficulty, researchers have mainly focused on the algorithmic field to seek a crack solution. Qingming Luo's team uniquely pointed out that the fundamental strategy to solve the big data dilemma should be to improve the data quality from the source, and then improve the efficiency of subsequent data processing and analysis. They used HD-fMOST to perform three-dimensional high-definition two-color imaging of the whole brain of mice with sparsely labeled neurons, and acquired 12,000 coronal images and their cellular architecture information with 0.3 0.3 1 micron voxel resolution in 5 days, which is the fastest technology to realize whole-brain optical imaging with similar voxel resolution at present. Through analysis, it was found that the effective signal of the raw data covered a 16-bit dynamic range with an average signal-to-noise ratio as high as 110, which could be directly superimposed to generate whole-brain projection maps. The high-resolution data quality enables the reconstruction of neuronal morphology to be nearly 2-fold faster even with a 5-fold increase in complexity. Results of online quantitative statistics of whole-brain distribution of specific types of neurons with an average accuracy of 95% are also given in the article, demonstrating that the high-quality raw data obtained by HD-fMOST can support online analysis. In addition, this technique realizes online lossless compression of 10 TB-level raw datasets of mouse whole brain with a compression rate of 3%, which can be directly written to a USB flash drive or uploaded to the cloud, and is expected to greatly reduce the burden of high-resolution whole-brain 3D datasets caused by data storage and transmission.
Figure 2 HD-fMOST results of high-definition 3D imaging of whole mouse brain sparsely labeled with specific neurons
In summary, LiMo microscopy proposed by the team significantly improves background suppression in fast high-resolution optical imaging. The HD-fMOST technique developed on this basis not only greatly improves the data quality of whole-brain optical imaging, but also opens up a completely new solution to the big data challenges faced in this field, which significantly improves the efficiency in data storage, transmission, processing, and analysis, and is expected to play a great role in standardized and scaled brain science research.
Dr. Qiuyuan Zhong and Professor An-An Li are tied as first authors, Dr. Qingming Luo and Professor Ching Yuan are tied as corresponding authors, and Rui Jin is the first author. Rui Jin, PhD student, Dejie Zhang, MSc, Xiangning Li, Prof. Xueyan Jia, MSc, Zhangheng Ding, PhD student, and Luo Pan, PhD, D. student, Zhou Can, Jiang Chenyu, M.S., Feng Zhao, Ph.D., Zhang Zhihong, Professor, and Gong Hui, Professor, are *** coauthors.