Microscopy system and method for image segmentation

Patent metadata

Field	Value
Publication number	US-12555239-B2
Application number	US-202117493937-A
Country	US
Kind code	B2
Filing date	Oct 5, 2021
Priority date	Oct 9, 2020
Publication date	Feb 17, 2026
Grant date	Feb 17, 2026

Abstract

A microscopy system comprises a microscope configured to capture an overview image and a computing device comprising a model trained for image segmentation, which calculates a segmentation mask based on the overview image. The computing device adjusts a pattern described by a parameterized model to the segmentation mask. An updated segmentation mask is generated using the adjusted pattern.

First claim

What is claimed is:

1 . A microscopy system, comprising
a microscope configured to capture at least a first image; and a computing device comprising a model trained for image segmentation, which is configured to calculate a segmentation mask based on at least the first image; wherein the computing device is configured to generate an updated segmentation mask in which problem areas of the segmentation mask are removed by adjusting a mathematical pattern formed by a parameterized model to the segmentation mask and generating the updated segmentation mask using the adjusted mathematical pattern, wherein the mathematical pattern includes a periodicity with repeating shapes, wherein the problem areas in the segmentation mask are deviations from a periodic pattern with repeating shapes, wherein the updated segmentation mask is free of problem areas, wherein the first image or an image calculated therewith is analyzed in order to detect a sample carrier type, wherein contextual data is stored for different sample carrier types, and wherein the contextual data pertaining to the detected sample carrier type is used to adjust the mathematical pattern.

2 . A method for image segmentation, comprising
receiving a first image; calculating a segmentation mask based on the first image using a model trained for image segmentation; generating an updated segmentation mask in which problem areas of the segmentation mask are removed by adjusting a mathematical pattern formed by a parameterized model to the segmentation mask, and generating the updated segmentation mask using the adjusted mathematical pattern, wherein a plan-view image is first calculated from the first image using calibration data and the segmentation mask is calculated from the plan-view image; or wherein the segmentation mask is calculated from the first image without first calculating a plan-view image with calibration data, and wherein the mathematical pattern described by the parameterized model is converted to a viewing angle of the first image using the calibration data before the adjusting of the mathematical pattern to the segmentation mask occurs.

3 . The method according to claim 2 ,
wherein the mathematical pattern includes a periodicity with repeating shapes, wherein the problem areas in the segmentation mask are deviations from a periodic pattern with repeating shapes, and wherein the updated segmentation mask is free of problem areas.

4 . The method according to claim 3 ,
wherein the mathematical pattern is periodic and wherein the periodic pattern comprises an arrangement of similar shapes in the form of a grid.

5 . The method according to claim 3 ,
wherein the periodic pattern describes sample receptacles of a sample carrier.

6 . The method according to claim 2 ,
wherein parameters of the parameterized model which define the mathematical pattern are calculated such that the mathematical pattern has a highest possible degree of correspondence with the segmentation mask.

7 . The method according to claim 6 ,
wherein calculation of the parameters includes an iterative adjustment in which the degree of correspondence between the mathematical pattern and the segmentation mask is maximized.

8 . The method according to claim 2 ,
wherein parameters of the parameterized model indicate one or more of the following characteristics: a uniform size of shapes of the mathematical pattern; a uniform spacing between shapes of the mathematical pattern; a slope of rows or columns of the shapes of the mathematical pattern; a class type of the shapes of the mathematical pattern; a position of the mathematical pattern relative to the segmentation mask.

9 . The method according to claim 2 ,
wherein an image classification of the first image or an image calculated therewith is calculated; wherein different mathematical patterns described by respective parameterized models are stored for different image classes; wherein, depending on a result of the image classification, an associated stored mathematical pattern is selected and used for the adjusting to the segmentation mask.

10 . The method according to claim 2 ,
wherein the first image or an image calculated therewith is analyzed in order to detect a sample carrier type; wherein contextual data is stored for different sample carrier types; wherein the contextual data pertaining to the detected sample carrier type is used to adjust the mathematical pattern.

11 . The method according to claim 10 ,
wherein the contextual data relates to at least one of: parameter starting values and parameter border values for an iterative adjustment of the parameters.

12 . The method according to claim 2 ,
wherein the adjusted mathematical pattern is used as the updated segmentation mask and output to a user or to a subsequent image processing program.

13 . The method according to claim 12 ,
wherein the adjusted mathematical pattern used as the updated segmentation mask is a vector graphic.

14 . The method according to claim 2 ,
wherein the problem areas in the segmentation mask are identified by comparing the adjusted mathematical pattern with the segmentation mask and are corrected using the adjusted mathematical pattern; and wherein the thus corrected segmentation mask is used as the updated segmentation mask.

15 . The method according to claim 2 ,
wherein the first image is an overview image.

16 . A computer program with commands which are stored on a non-transitory computer-readable medium and which, when executed by a computer, cause the execution of the method according to claim 2 .

17 . The system of claim 1 ,
wherein parameters of the parameterized model which define the mathematical pattern are calculated such that the mathematical pattern has a highest possible degree of correspondence with the segmentation mask.

18 . The system of claim 1 ,
wherein an image classification of the first image or an image calculated therewith is calculated; wherein different mathematical patterns described by respective parameterized models are stored for different image classes; wherein, depending on a result of the image classification, an associated stored mathematical pattern is selected and used for the adjusting to the segmentation mask.

19 . The system of claim 1 ,
wherein the first image is an overview image.

20 . A method for image segmentation, comprising:
receiving a first image; calculating a segmentation mask based at least on the first image using a model trained for image segmentation; generating an updated segmentation mask in which problem areas of the segmentation mask are removed by adjusting a mathematical pattern formed by a parameterized model to the segmentation mask, and generating the updated segmentation mask using the adjusted mathematical pattern; wherein contextual data is stored for different sample carrier types; and using the contextual data pertaining to a determined sample carrier type to adjust the mathematical pattern.

Assignees

• Zeiss Carl Microscopy Gmbh

Inventors

• Amthor Manuel
• Haase Daniel

Classifications

• G02B21/367

  providing an output produced by processing a plurality of individual source images, e.g. image tiling, montage, composite images, depth sectioning, image comparison · CPC title

• G06T2207/10056

  Microscopic image · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T7/11 Primary

  Region-based segmentation · CPC title