# US 2020/0402273 A1 · Holographic reconstruction method

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0402273 A1
- Document ID: US20200402273A1
- Status: Published application
- Kind: A1
- Application number: 16/907,407
- Filed: 2020-06-22
- Published: 2020-12-24
- Earliest claimed: 2019-06-22
- Assignees: Commissariat A L'energie Atomique Et Aux Energies Alternatives
- Applicants: Commissariat a l'energie atomique et aux energies alternatives
- Inventors: Lionel Herve; Cedric Allier
- Web page: https://patentgrep.com/patent/US20200402273A1

## CPC

- G03H 1/0866: {Digital holographic imaging, i.e. synthesizing holobjects from holograms}
- G01N 15/1468: {with spatial resolution of the texture or inner structure of the particle}
- G03H 1/0005: {Adaptation of holography to specific applications (holographic optical element G02B5/32; holographic scanner G02B26/106; recognition using holographic mask G06V10/88; holographic memories G11B7/0065, G11C13/042)}
- G03H 1/0443: {Digital holography, i.e. recording holograms with digital recording means (holobject computation G03H1/0866)}
- G06F 18/217: {Validation; Performance evaluation; Active pattern learning techniques}
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06T 12/10: Image preprocessing, e.g. calibration, positioning of sources or scatter correction
- G06T 12/20: Inverse problem, i.e. transformations from projection space into object space
- G06V 20/693: {Acquisition}
- G01N 2015/1454: {using phase shift or interference, e.g. for improving contrast}
- G03H 2001/005: {in microscopy, e.g. digital holographic microscope [DHM]}
- G03H 2001/0447: {In-line recording arrangement}
- G03H 2001/0452: {arranged to record an image of the object}
- G03H 2001/0816: {Iterative algorithms}
- G03H 2001/0875: {Solving phase ambiguity, e.g. phase unwrapping}
- G03H 2210/55: Having particular size, e.g. irresolvable by the eye
- G06N 20/10: using kernel methods, e.g. support vector machines [SVM]
- G06T 2211/424: Iterative
- G03H: HOLOGRAPHIC PROCESSES OR APPARATUS (holograms, e.g. point holograms, used as ordinary optical elements G02B5/32; producing stereoscopic or other three-dimensional effects G02B30/00; diffraction-grating systems G02B27/44; systems using moiré fringes G02B27/60; optical logic elements G02F3/00; stereo-photography G03B35/00; photosensitive materials or processes for photographic purposes G03C; {stereo-photographic or similar processes G03C9/00}; apparatus for processing exposed photographic materials G03D; analogue computers performing mathematical operations with the aid of optical elements G06E3/00; authentication by radiation, of concealed information carried by holograms or diffraction gratings G06K19/16; holographic storage G11B7/0065, G11C13/04; {stereoscopic or other three dimensional effects in television systems H04N13/00})
- G01N: INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES (measuring or testing apparatus or processes other than immunoassay, involving enzymes or microorganisms C12M, C12Q)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A method for observing a sample, the sample lying in a sample plane defining radial positions, parameters of the sample being defined at each radial position, the method comprising: a) illuminating the sample using a light source, emitting an incident light wave that propagates toward the sample; b) acquiring, using an image sensor, an image of the sample, said image being formed in a detection plane, the sample being placed between the light source and the image sensor; c) processing the image acquired by the image sensor, so as to obtain an image of the sample, the image of the sample corresponding to a distribution of at least one parameter of the sample describing the sample in the sample plane; wherein the processing of the acquired image comprises implementing an iterative method, followed by applying a supervised machine learning algorithm, so as to obtain an initialization image intended to initialize the iterative method.

## Claims

### Claim 1

A method for observing a sample, the sample lying in a sample plane defining radial positions, parameters of the sample being defined at each radial position, the method comprising:
a) illuminating the sample using a light source, the light source emitting an incident light wave that propagates toward the sample;
b) acquiring, with an image sensor, an image of the sample, said image being formed in a detection plane, the sample being placed between the light source and the image sensor;
c) from the image acquired by the image sensor, obtaining an image of the sample, in the sample plane, the image of the sample in the sample plane corresponding to a distribution of at least one parameter of the sample in the sample plane;
wherein c) comprises:
(i) taking into account an initialization image, in the sample plane;
(ii) applying a holographic propagation operator to the initialization image, or to a complex image of the sample resulting from (v) of a preceding iteration, so as to provide an estimated image, the estimated image being an estimation of the image acquired by the image sensor in the detection plane;
(iii) comparing the image acquired in b) and the image estimated in (ii);
(iv) depending on the comparison, updating the parameters of the sample;
(v) provided that a convergence criterion has not been met or a preset number of iterations has not been reached, updating a complex image of the sample, in the sample plane, using the parameters of the sample updated in (iv), then reiterating (ii) to (v);
(vi) using the parameters of the sample, resulting from the last step (iv) performed, to feed a supervised machine learning algorithm, so as to update parameters of the sample;
(vii) using updated parameters of the sample resulting from (vi) to update the initialization image;
(viii) reiterating, at least once, steps (ii) to (v), using, during the first reiteration, the updated initialization image resulting from (vii);
(ix) obtaining an image of the sample from the parameters of the sample updated in step (iv) of the last iteration of steps (ii) to (v).

### Claim 2 (depends on 1)

The method of claim 1, wherein (iv) comprises computing a validity indicator, such that the parameters of the sample are updated to make the validity indicator tend toward a preset value.

### Claim 3 (depends on 2)

The method of claim 2, wherein, in (iv), the parameters of the sample are updated so as to minimize the validity indicator.

### Claim 4 (depends on 1)

The method of claim 1, wherein the supervised machine learning algorithm comprises a neural network.

### Claim 5 (depends on 4)

The method of claim 4, wherein the neural network is a convolutional neural network.

### Claim 6 (depends on 2)

The method of claim 2, wherein (iv) comprises determining a gradient of the validity indicator as a function of at least one parameter of the sample, such that the parameters of the sample are updated to decrease the validity indicator of the following iteration.

### Claim 7 (depends on 6)

The method of claim 6, wherein (iv) employs a gradient descent algorithm.

### Claim 8 (depends on 1)

The method of claim 1, wherein:
(iv) results in the determination of an image of each parameter of the sample in the sample plane;
in (vi), at least one image of a parameter of the sample forms an input layer of the supervised machine learning algorithm;
in (viii), the supervised machine learning algorithm delivers an output image, corresponding to an image of a parameter of the sample updated by the algorithm.

### Claim 9 (depends on 1)

The method of claim 1, wherein, in (ii), the estimation of the image of the sample in the detection plane comprises a convolution with a convolution kernel, the convolution kernel representing a spatial extent of the light source.

### Claim 10 (depends on 1)

The method of claim 1, wherein the parameters of the sample comprise:
a first parameter of the sample representing an absorbance of the sample;
a second parameter of the sample, representing an optical path difference, along the propagation axis of the incident light wave.

### Claim 11 (depends on 10)

The method of claim 10, wherein, in step (vi), the supervised machine learning algorithm is fed with:
an image of the first parameter of the sample, corresponding to a spatial distribution of the first parameter of the sample, which is updated in the last iteration of steps (ii) to (v) preceding step (vi);
an image of the second parameter of the sample, corresponding to a spatial distribution of the second parameter of the sample, which is updated in the last iteration of steps (ii) to (v) preceding step (vi);
and wherein the supervised machine learning algorithm allows an image of the updated second parameter of the sample to be obtained.

### Claim 12 (depends on 1)

The method of claim 1, wherein, following a step (viii), steps (vi) to (viii) are repeated at least once.

### Claim 13 (depends on 1)

The method of claim 1, wherein no image-forming optics are placed between the sample and the image sensor.

### Claim 14 (depends on 1)

The method of claim 1, wherein an optical system, such as a lens or objective, is placed between the sample and the image sensor, the optical system defining an image plane and an object plane, the method being such that, in b):
the object plane is offset from the sample plane by an object defocus distance;
and/or the image plane is offset from the detection plane by an image defocus distance.

### Claim 15 (depends on 1)

A device for observing a sample, comprising:
a light source, configured to emit an incident light wave in order to illuminate the sample;
a sample holder, configured to receive the sample;
an image sensor, configured to acquire an image of the sample when the sample is placed on the sample holder;
a processor, programmed to execute instructions allowing steps i) to ix) of a method of claim 1 to be implemented on the basis of an image acquired by the image sensor.
