ES2853354B2 - Procedure to correct the effect of light scattering in planar laser beam microscope measurements - Google Patents

Description

DESCRIPTION

Procedure to correct the effect of light scattering in planar laser beam microscope measurements

OBJECT OF THE INVENTION

The invention generally pertains to the field of optics, and more particularly to planar laser beam microscopy.

The object of the present invention is a method to reverse the effects caused by light scattering when passing through the sample in planar laser beam microscopy images.

BACKGROUND OF THE INVENTION

Studies of embryos and large biological samples through the light microscope present, unlike what happens with small samples, particular problems related to the optical properties of the tissue, absorption and light scattering, which cause a marked loss in the resolution of the images. To overcome these problems, significant improvements have been made in recent years over the planar laser beam microscopes, whose invention dates back to 1903.

A planar laser beam microscope is made up of an illumination means capable of emitting a thin layer of light in one direction, called the "illumination direction" (x), and a detection means consisting of a camera coupled to a large objective. numerical aperture arranged according to a direction called "detection direction" (z), which is perpendicular to the direction of illumination, following the original configuration of Siedentopf and Zsigmondy.

With this configuration, the camera can obtain a 2D image of the fluorescence emitted by the sample section excited by the illumination plane. If the sample is also moved in the direction of the detection axis and images are taken at different positions, a set or stack of 2D images is generated where each of the 2D images corresponds to the fluorescence emitted by the section excited by the detection plane. illumination at each position with respect to the sample. This stack of 2D images contains information about the position in z (depth of the sample according to the detection direction) obtained by moving the sample, and of the x and y positions, present in each 2D image. The 2D image stack can then be merged to generate a 3D image of the sample, as described in US 7,554,725 to Stelzer et al. Subsequently, it was proposed to rotate the sample around its own axis, usually vertical, to capture several stacks of 2D images (commonly called "angular measurements") and subsequently merge them, which allows to improve the anisotropy and the quality of the images (S Preibisch et al, Nature Methods 7 (2010)).

For a clearer understanding of this technique, Figs. 1a and 1b showing a first example of a planar laser beam microscope (100). The sample (107) is placed in a support (101) inside a cuvette (102) filled with a liquid. A beam (103) of Gaussian, Bessel, Airy or similar linear lighting falls on a cylindrical lens (104) that focuses it thanks to a lighting objective (105) to generate the sheet (106) of vertical flat lighting. This sheet (106) of vertical plane illumination falls on the sample (107) according to the direction of illumination (DI), and the fluorescent light (108) emitted by that particular plane of the sample (107) is collected by an objective (109). ) of detection oriented according to the direction of detection (DD), which is perpendicular to the direction of illumination (DI). An image corresponding to the sample portion 107 illuminated by the illumination sheet 106 is thus obtained. To obtain a complete image of the sample (107), it is possible to move the support (101) according to the detection direction (DD) to take measurements corresponding to several sections of the sample (107), or the support (101 ) can rotate around its vertical axis to allow various angular measurements to be taken according to the technique proposed by Preibisch.

Since 2015, the inventors of the present application have filed several patent applications aimed at various improvements in this type of microscope. These patent applications are the following:

PCT/ES2015/070455 entitled "Microscope and procedure for generating 3D images of a collection of samples" that describes a new microscope that combines the SPIM (Selective Plane Illumination Microscope) planar laser beam technique with the tomography technique optical projection (OPT, Optical Projection Tomography).

PCT/ES2016/070714, titled “Multiple Loading Device for Beam Microscope planar laser'' describing a multiple loading device for feeding a continuous and sequential flow of samples to a planar laser beam microscope.

PCT/ES2017/070028, entitled “Automatic objective change device for planar laser beam microscope”, which describes a device that allows the image acquisition objective of a planar laser beam microscope to be automatically changed depending on the magnification desired at all times.

PCT/ES2017/070028, entitled "Rotary objective change device for flat laser beam microscope", which describes a device where the objective change is carried out through rotations of the cuvette itself.

PCT/ES2017/070184 entitled "Sample holding device for microscope", which describes a device to be able to mount samples and measure them in flat laser beam equipment.

PCT/ES2019/070629 entitled "Microscope and planar laser beam method for large samples", which describes a system to allow large 3D images to be obtained without the need to move the sample.

In planar laser beam microscopy systems, the size of the sample determines the maximum image quality that can be obtained. As images of planes at greater depths are obtained, their resolution is affected, among other factors, by the dispersion suffered by the light on its way to the detection objective.

Indeed, the fluorescence emitted by an excited plane of the sample must propagate through it a certain distance. Therefore, the further away from the detection objective this plane is, the greater the distance that the light travels through the sample, and therefore the greater the deterioration of the images. For this reason, the use of this technique is currently limited to the study of very transparent samples (previously cleared) or of a size small enough that the loss of resolution due to scattering is almost negligible.

DESCRIPTION OF THE INVENTION

The present invention describes a procedure that allows reversing the effect of dispersion of light in plane beam microscopy images in order to improve their quality. To do this, first the degree of light scattering in the sample is estimated and, subsequently, post-processing is applied to the images obtained to recover part of the lost resolution. This improvement allows the use of planar laser beam techniques to study samples whose size or limited transparency prevent their visualization with current planar laser beam techniques.

An x,y,z coordinate system is defined herein where: "x-direction " refers to the direction of illumination; the "z direction " refers to the detection direction, or sample depth direction; and the "y direction " is perpendicular to the x and z directions, that is, it is the vertical direction in the natural use position of a planar laser beam microscope as shown in the figures.

In this document, the term "depth" referring to an image or sample plane refers to the distance between the sample plane excited by the illumination sheet and the detection chamber. The greater this distance, the greater the depth of said image or plane. The depth therefore corresponds to the z-direction.

In this document, the term "Gaussian beam " refers to a monochromatic beam whose intensity profile transverse to the propagation axis follows a Gaussian function. This type of beam is the most common output of any laser light source, corresponding to the fundamental mode TEM ⁰⁰ .

The present invention describes a procedure for correcting the light scattering effect in a plane laser beam microscope that basically comprises the following steps:

to. Generation of a database of transfer functions for semi-transparent media

The objective of this step is to characterize the loss of resolution of an image when propagated in a semi-transparent medium with low dispersion. For this, simulations are carried out with the aim of estimating a transfer function H(k ^x ,k ^y ;z) for the medium that quantifies the attenuation suffered by each spectral component (k ^x , k ^y ) of an image for a depth z .

In this context, the parameters kx and ky refer to spatial frequencies in an image. These are expressed as cycles or line pairs per millimeter. In diffraction-limited microscopy techniques, the maximum frequency in the image corresponds to the theoretical Abbe limit.

The Monte Carlo method allows the simulation of light propagation in scattering media with great precision. In addition to estimating the spatial distribution of light intensity, this technique can be used to calculate the frequency response of the medium to a unit impulse. For this, a simulation of the propagation of a collimated source of the 'pencil' type is carried out, in which the photons are generated in the z direction at a point (x, y) = (0, 0) in a three-dimensional volume. Once the simulation is complete, the Fourier transform is calculated for each z-plane to obtain the frequency response of the system.

This process will be repeated for different values of the dispersion coefficient in order to create a database of transfer functions corresponding to each dispersion coefficient.

In short, in this step, through simulations using the Monte Carlo method, the transfer function of a medium for different scattering coefficients and scanning depths is calculated, creating a database of transfer functions corresponding to each scattering coefficient. and scan depth.

b. Experimental estimation of the scattering coefficient of the sample using a plane beam microscope

In order to perform the deconvolution correctly, it is necessary to use the transfer function of a medium with a dispersion coefficient equivalent to that of the sample. Since its value is unknown a priori, an in situ experimental estimate of it must be made. It is important to emphasize that the measurement will be carried out in the flat laser beam microscope with the sample mounted as a procedure prior to image acquisition.

Thus, assuming that the sample has a coefficient of dispersion relatively homogeneous, before scanning the sample and once mounted in the plane beam microscope, it is illuminated with a Gaussian beam at different depths (z) and positions (y) with respect to the camera. This can be done either simultaneously or sequentially, depending on whether the projection rate is greater than the camera's measurement rate (frames per second, fps).

The camera (without filter) captures the distribution of the light scattered by the sample as a consequence of the propagation of each of the beams, as shown in Fig. 2. The divergence of the beams in each of the images due to dispersion allows estimating the dispersion coefficient. To do this, during post-processing, intensity profiles are measured in the image and beam broadening is quantified at various points on the x-axis (beam propagation axis), as shown in Fig. 3. Minimizing the measurement From the error between the experimental curves and those of the data from the Monte Carlo simulations of the previous step, the dispersion coefficient that best fits is determined.

To increase the precision of the procedure, it is repeated for each of the measurement positions (y, z) in order to estimate an average dispersion coefficient of the sample.

In short, this step comprises experimentally estimating, using the plane laser beam microscope, the scattering coefficient of a sample assuming that it is uniform throughout the sample, where the estimation comprises the following steps:

b1) mount the sample on the flat laser beam microscope;

b2) illuminate each sample plane with a Gaussian beam; b3) capture images of the Gaussian beam propagation in the sample and repeat the process in different positions; and

b4) estimate the scattering coefficient of the sample by measuring the divergence of the beams caused by the scattering;

c. sample scan

The sample is now scanned normally by the planar laser beam microscope, obtaining a set of images corresponding to different sections of the sample. That is, the taking of images of the sample is carried out in a plane laser beam microscope following a conventional acquisition protocol, either by lighting with a flat beam or with digital scanning.

The images obtained will present artifacts caused by scattering, especially in those corresponding to scanning planes located at greater depths with respect to the detection objective.

d. Correction of the effect of dispersion in the scan images

In order to reverse and reduce the effects of dispersion in the images obtained, the deconvolution of each image must be calculated with its corresponding medium transfer function, calculated in step a. To do this, the simulation database will be searched for the volume that corresponds to the dispersion coefficient estimated in step b, and the convolution with the transfer function for said volume will be calculated for each image.

That is, each image of the sample obtained in the scan is deconvolved with the transfer function of the database that corresponds to the dispersion coefficient of the sample and the depth of scanning of each image, thus reversing in each image the effect of light scattering.

The deconvolution can be performed with multiple methods, preferably Wiener deconvolution.

BRIEF DESCRIPTION OF THE FIGURES

Figs. 1a and 1b respectively show a perspective view and a top view of the main elements of a conventional planar laser beam microscope.

Fig. 2 experimental measurement scheme to estimate the dispersion coefficient of the sample.

Fig. 3 shows the procedure for estimating the dispersion coefficient from the experimental measurements.

Fig.4 shows a simplified flow diagram of the process of the present invention.

PREFERRED EMBODIMENT OF THE INVENTION

A complete procedure is described below. In any study carried out with the proposed method, it is assumed that the Monte Carlo simulations have been carried out previously, for this reason there is a database with the spatial information to estimate the dispersion coefficient and a database with the functions for a wide range of media and sample sizes.

Thus, Fig. 2 shows a scheme of the assembly of the invention where the illumination objective (OI), the measurement positions (PM), the detection objective (OD), the sample (M), the measurements of beam scatter (MD), an image analysis (AI) block, a bank of Monte Carlo (MC) simulations, and the transfer function for the estimated scattering coefficient (TF(^s)), where is the dispersion coefficient.

Fig. 3 shows a diagram of the dispersion coefficient estimation procedure where the experimental dispersion measure (MDE) and the dispersion measure in Monte Carlo simulations (MD-MC) can be seen for different values of ^s. The graphs show the value of the parameter y against the intensity (I)

Fig. 4 thus shows a complete flow diagram of the process of the present invention. In Fig. 4 the following symbols and parameters appear:

LS - Plane Beam Microscope

H - Data transfer function

I ^mc - simulated intensity

lo - Raw Unprocessed Image

I - Image after deconvolution

Under these conditions, the steps for the procedure will be as follows:

1. Mounting the sample in the microscope

2. Image capture without filter on the detection objective and illumination of the sample with a Gaussian beam. Images are captured at different z -depths and positions on the y- axis to estimate the mean scattering coefficient of the sample.

3. Image acquisition of the sample using the user-preferred conventional planar beam protocol.

4. Calculation of the deconvolution of the images obtained in step 3 with the transfer function of the medium for the dispersion coefficient estimated experimentally in step 2.

Claims (2)

1. Procedure to correct the effect of light scattering in measurements of a plane laser beam microscope, characterized in that it comprises the following steps: a) calculate, by means of simulations using the Monte Carlo method, the transfer function of a semi-transparent medium for different scattering coefficients and scan depths, creating a database of transfer functions corresponding to each scattering coefficient and scan depth; b) experimentally estimate, using the plane laser beam microscope, the dispersion coefficient of a sample assuming that it is uniform throughout the sample, where the estimation comprises the following steps: b1) mount the sample on the flat laser beam microscope; b2) illuminate each sample plane with a Gaussian beam; b3) capture images of the Gaussian beam propagation in the sample and repeat the process in different positions; and b4) estimate the scattering coefficient of the sample by measuring the divergence of the beams caused by the scattering; c) normally scan the sample using the flat laser beam microscope, obtaining a set of images corresponding to different sections of the sample; and d) deconvolve each image of the sample obtained in the scan with the transfer function of the database that corresponds to the dispersion coefficient of the sample and the scanning depth of each image, thus reversing the effect of dispersion in each image of the light. 2. Method according to claim 1, wherein the deconvolution is performed using the Wiener method.