Southern Federal University, Scientific Research Center for Neurotechnology Russian Federation, 194 Stachki ave, Rostov-on-Don, Russian Federation.
Southern Federal University, Scientific Research Center for Neurotechnology Russian Federation, 194 Stachki ave, Rostov-on-Don, Russian Federation.
J Neurosci Methods. 2021 Sep 1;361:109273. doi: 10.1016/j.jneumeth.2021.109273. Epub 2021 Jul 3.
There is a challenge to determine stereotaxic coordinates of a target structure with the accuracy, comparable to their size, when imaging the rat brain through cranial windows using confocal (multiphoton) microscopy in vivo. Some methods based on the estimation of the linear displacement from the intersections of the cerebral vessels are most often used for these purposes, but their accuracy can be improved.
A new method for converting pixel coordinates of points of interest on an image obtained in two-photon microscopy into stereotaxic ones using quadratic approximation with L regularization has been developed. A comparative analysis of several methods for converting pixel coordinates into stereotaxic ones was carried out. The current study is aimed to select a method which minimizes the error of coordinate conversion within the a priori specified threshold value.
A method for determining the stereotaxic coordinates of each pixel in an image obtained by laser scanning in two-photon and / or confocal modes with an accuracy of several tens of microns is proposed.
COMPARISON WITH EXISTING METHOD(S): It is shown that the error probability of the most common method based on calculating the points of interest coordinates as displacements relative to the selected vessels intersections can be reduced by using the quadratic approximation with L regularization. Our proposed method allows us to improve the accuracy of determining the coordinates of points of interest on 10-30 µm.
The proposed approach will be useful in research where precise positioning of microelectrodes, sensors, etc. for implantation in specified brain structures or groups of neurons determined by functional mapping is required.
当使用颅窗通过共聚焦（多光子）显微镜在体成像研究大鼠脑时，存在一个挑战，即需要以与目标结构大小相当的精度来确定其立体定向坐标。一些基于从脑血管交点估计线性位移的方法最常用于这些目的，但可以提高其准确性。
已经开发出一种新方法，用于使用具有 L 正则化的二次逼近将在双光子显微镜中获得的图像上的感兴趣点的像素坐标转换为立体定向坐标。对几种将像素坐标转换为立体定向坐标的方法进行了比较分析。本研究旨在选择一种在预先指定的阈值内最小化坐标转换误差的方法。
提出了一种用于以几十微米的精度确定通过双光子和/或共聚焦模式进行激光扫描获得的图像中每个像素的立体定向坐标的方法。
结果表明，通过使用具有 L 正则化的二次逼近，可以降低基于相对于所选血管交点计算感兴趣点坐标的最常见方法的误差概率。我们提出的方法允许我们将感兴趣点坐标的确定精度提高 10-30μm。
当需要在特定脑结构或通过功能映射确定的神经元群中植入微电极、传感器等时，需要精确定位，所提出的方法将非常有用。
