Distance determination of a sample plane in a microscope system

A computer-implemented method for a microscope system, wherein the microscope system includes a sample stage with a placement surface for holding a sample carrier, wherein the sample stage is displaceable along at least one direction of extent of a sample plane, and wherein the microscope system comprises an overview camera with a non-telecentric objective for producing digital images, wherein the overview camera is directed at the sample stage, the method comprising:

• storing at least two recorded digital images of the sample stage at different viewing angles;
• associating corresponding structures in the at least two recorded digital images using a trained machine-learning-based system to identify corresponding structures of the sample carrier placed into the sample stage in the two recorded digital images, wherein the trained machine-learning-based system comprises a trained reference model that is trainable by way of a set of annotated training images of sample carriers in a manner such, and the trained machine-learning-based system is thus adapted such, that corresponding structures in the at least two recorded digital images are associable with one another;
• and
• determining a distance of a reference point of the sample carrier from a reference point of the overview camera based on the different viewing angles onto the sample stage and a pixel distance of the at least two recorded digital images with respect to one another using the associated corresponding structures contained in the at least two recorded digital images,
• wherein the trained machine-learning-based system is configured to classify individual pixels of the at least two recorded images, and
• wherein an optimization of a target function having a plurality of corresponding structures of the at least two recorded digital images is used as input data, wherein a result of the target function indicates the pixel distance.

A non-transitory computer-readable medium for a microscope system, wherein the microscope system includes a sample stage with a placement surface for holding a sample carrier, wherein the sample stage is displaceable along at least one direction of extent of a sample plane, and wherein the microscope system includes an overview camera with a non-telecentric objective for producing digital images, wherein the overview camera is directed at the sample stage, wherein the non-transitory computer-readable medium has program instructions stored thereon, wherein the program instructions, when executed by one or more computers and/or control units and cause the one or the plurality of computers and/or control units to:

• store at least two recorded digital images of the sample stage at different viewing angles;
• associate corresponding structures in the at least two recorded digital images by using a trained machine-learning-based system for identifying corresponding structures of the sample carrier that has been placed into the sample stage in the at least two recorded digital images, wherein the trained machine-learning-based system has a trained reference model that is trainable by way of a set of annotated training images of sample carriers in a manner such, and the trained machine-learning-based system is thus adapted such, that corresponding structures in the at least two recorded digital images are able to be associated with one another;
• and
• determine a distance of a reference point of the sample carrier from a reference point of the overview camera based on the different viewing angles onto the sample stage and a pixel distance of the at least two recorded digital images with respect to one another using the associated corresponding structures contained in the at least two recorded digital images,
• wherein the trained machine-learning-based system is configured to classify individual pixels of the at least two recorded images, and
• wherein an optimization of a target function having a plurality of corresponding structures of the at least two recorded digital images is used as input data, wherein a result of the target function indicates the pixel distance.

A distance determination system for a microscope system, the distance determination system comprising:

• a sample stage having a placement surface configured to hold a sample carrier, wherein the sample stage is displaceable along at least one direction of extent of a sample plane;
• an overview camera with a non-telecentric objective configured to produce digital images, wherein the overview camera is directed at the sample stage;
• an evaluation unit connected to the overview camera, the evaluation unit including
• a storage system configured to store at least two recorded digital images of the sample stage at different viewing angles;
• and
• a trained machine-learning-based system configured to identify corresponding structures of the sample carrier placed into the sample stage in the two recorded digital images, wherein the trained machine-learning-based system comprises a trained reference model that is trainable by way of a set of annotated training images of sample carriers in a manner such, and the trained machine-learning-based system is thus adapted such, that corresponding structures in the at least two recorded digital images are associable with one another;
• and
• a distance determination unit configured to determine a distance of a reference point of the sample carrier from a reference point of the overview camera based on the different viewing angles onto the sample stage and a pixel distance of the two recorded digital images with respect to one another using the associated corresponding structures contained in the at least two recorded digital images,
• wherein the trained machine-learning-based system is configured to output a reference coordinate of one of the corresponding structures, and
• wherein an optimization of a target function having a plurality of corresponding structures of the at least two recorded digital images is used as input data, wherein a result of the target function indicates the pixel distance.