Automatic acquisition of microscope image sets

Abstract

A computer-implemented method for image acquisition is provided. The method may include receiving user input to indicate (106) at least one quality condition (206) associated with a plurality of regions of interest (204) of a sample array (200) imaged using a fluorescence microscope (310). The at least one quality condition preferably includes a target signal-to-noise ratio. The method may include automatically causing the fluorescence microscope (310) to acquire at least one microscope image for each region of interest (204) using an illumination setting automatically determined (114) such that the target signal-to-noise ratio is met (108). The method may include generating (118) a dataset that can be used to generate, train, validate, and / or test a machine learning model or to optimize in another way. The dataset may include the acquired microscope images or references to the microscope images.

Description

Technical Field

[0001] The present invention generally relates to the field of data processing in microscopy applications, and more particularly to improved techniques for obtaining large sets of microscope images for use in machine learning.

Background Art

[0002] In certain microscopy applications, large sets of images are required. As an example, when the goal is to train a microscope denoising model using a technique such as that disclosed in applicant's EP21204605.6 entitled "Training a denoising model for a microscope", it is necessary to obtain sufficient training images. In particular, for large samples with different heterogeneous regions, for example in a well plate, setting up imaging is often very time-consuming because the user usually has to adjust the acquisition settings for each imaging position to find the region of interest. On the other hand, using just one set of acquisition parameters typically results in some images being too bright or too dark, making the phosphor invisible and thus not optimal results. In such scenarios, obtaining a sufficient number of images for use as training data for machine learning, particularly deep learning, can be a major bottleneck for the application.

[0003] In the field of fluorescence microscope image acquisition, research has been done to address the monotonous and time-consuming tasks of manual placement of comparable specimens and repeated initiation of the same acquisition protocol (see, for example, "Methods in Cell Biology, Volume 123, 2014, Chapter 26 “Adaptive fluorescence microscopy by online feedback image analysis”, pages 489 - 503, doi: 10.1016 / B978 - 0 - 12 - 420138 - 5.00026 - 4").

[0004] As another example, “”AutoScanJ: A Suite of ImageJ Scripts for Intelligent Microscopy” by Tosi S. et.al (18 March 2021, Frontiers in Bioinformatics, doi: 10.3389 / fbinf.2021.627626)” discloses a set of ImageJ scripts for imaging a target of interest by automatically driving a motorized microscope at corresponding positions.

[0005] U.S. Patent Application Publication No. 2021 / 0264595 discloses a method for training a convolutional neural network using computer-assisted microscope image acquisition and pre-classification of training images for biological objects of interest. This method combines an automated scanning platform, on-the-fly image analysis parallel to the scanning process, and the use of a user interface for review of pre-classification performed by software.

[0006] Conventional workflows, including feedback microscopy, have extended widely around the automation of region definition within a sample, job definition (e.g., selection of modality, phosphor, and objective lens), and focusing (e.g., using a focus map). Such features are currently provided as building blocks in many commercially available microscopes.

[0007] However, especially with regard to illumination, it remains very difficult to efficiently set all image acquisition conditions. One typical approach is to define general illumination conditions that are performed manually for the entire acquisition or for each region. That is, for large samples, it is not possible to guarantee that the intensity and quality of staining reach appropriate and homogeneous levels for all regions to be imaged. Existing tools aimed at automating data acquisition are still not sufficiently versatile and require monotonous and continuous manual adjustment.

Summary of the Invention

Problems to be Solved by the Invention

[0008] Therefore, the underlying problem of the present invention is to provide an improved microscope image acquisition technology, thereby at least partially overcoming the above-mentioned drawbacks of the prior art.

Means for Solving the Problems

[0009] In one embodiment of the present invention, a computer-implemented method for image acquisition is provided. The method may include receiving user input to indicate at least one quality condition associated with a plurality of regions of interest of a sample array imaged using a fluorescence microscope. The at least one quality condition preferably includes a target signal-to-noise ratio. The method may include automatically causing the fluorescence microscope to acquire at least one microscope image for each region of interest using an illumination setting automatically determined such that the target signal-to-noise ratio is satisfied. The method may include generating a dataset that can be used to generate, train, validate, and / or test a machine learning model or otherwise optimize it. The dataset may include the acquired microscope images or references to the microscope images.

[0010] Accordingly, this method provides a workflow for automated data acquisition, particularly for automated acquisition by automated illumination drive of a large set of images. In the sense that a region of interest (ROI) can be defined and tasks can be performed in the ROI, this workflow may be somewhat similar to a feedback microscopy workflow, but the disclosed method requires minimal user intervention. In existing workflows, the user typically has to set illumination conditions for each region, such as gain and exposure time, for each phosphor or transmitted light, in advance. For example, in a multi-well plate having different dies or phosphors for each row, or in a large sample having regions (e.g., a brain), if the sample has different regions where the user desires to acquire different conditions, this is a time-consuming step. Typically, the user is not interested in setting these different conditions per se, but often has to do so in order to obtain the desired results. Embodiments of the present invention omit this indirection and enable the user to directly specify the desired quality (e.g., metric).

[0011] Furthermore, it can be very difficult for the user to discover photosensitive regions without photobleaching and damage to them. In some scenarios, such as long experiments, it is impossible for the user to continuously and appropriately adjust the acquisition settings throughout the experiment. This can be achieved by setting the conditions in the automated illumination.

[0012] A particular use of embodiments of the present invention may center around light microscopes that use wide-field light, confocal light, and transmitted light. The proposed workflow is applicable to fixed samples or live samples. In the fixed or prepared area of the sample (where photo-bleaching may be possible), images can be obtained with a higher SNR (signal-to-noise ratio) determined by the automatic illumination feature of aspects of the present invention. For live samples such as high-throughput 3D imaging of tissue culture cells, researchers typically attempt to minimize photo-bleaching and phototoxicity by minimizing energy input. In aspects of the present invention, the amount of phototoxicity can instead be a condition set by automatic illumination.

[0013] In one practical implementation, the step of automatically acquiring at least one microscopic image for each region of interest using an illumination setting automatically determined to meet a target signal-to-noise ratio can be realized using the technique disclosed in International Publication No. WO 2022 / 028694 entitled "Method for adjusting the illumination in a fluorescence microscope, and corresponding fluorescence microscope" by the applicant.

[0014] In one aspect of the present invention, at least one quality condition includes a target signal-to-noise ratio applied to the entirety of a plurality of regions of interest. Additionally or alternatively, at least one quality condition may include a target signal-to-noise ratio applied to a selected one of the plurality of regions of interest. Thus, the user can conveniently select a global target SNR and individual local target SNRs in relation to the ROI, and this method automatically acquires corresponding images using image acquisition settings to meet the target SNR settings. The target signal-to-noise ratio may be or include a specific signal-to-noise ratio value or a minimum signal-to-noise ratio value. The target signal-to-noise ratio may be applied to a selected one of a plurality of channels.

[0015] In another aspect of the present invention, at least one quality condition further includes at least one of a threshold condition based on light dose, a threshold condition based on phosphor quality, or a threshold condition based on image quality such as contrast. Thus, a particularly diverse range of image acquisition scenarios becomes possible in a user-friendly and efficient manner to achieve one or more of the following advantages. · By defining a simple set of metrics or even a single metric such as a minimum SNR, similar quality is guaranteed across the entire acquired dataset. · Enabling the user to define conditions for different ROIs. This can be done by simply selecting the channel to be imaged and the desired quality, i.e., without the need to individually set acquisition parameters for each region. This provides a particularly efficient implementation of use cases such as "desiring that ROI no.1 has an SNR of 5 in channel 3 and ROI no.2 has an SNR of 7 in channel 2". · Enabling the definition of threshold conditions based on light dose, phosphor density or quality for imaging and discovering ROIs (e.g., ensuring that all phosphors are visible in the region). The user can specify the regions to which these conditions apply or can have the system filter ROIs based on these conditions.

[0016] In a preferred usage scenario, the sample array includes a plurality of distinct samples. In particular, the sample array may be a well plate. In one aspect, the method may further include receiving user input to indicate the selection of a plurality of regions of interest. This selection can indicate all samples of the sample array, one or more individual samples of the sample array, one or more rows and / or columns of the samples of the sample array, and / or a shape surrounding one or more samples of the sample array, where the shape may optionally be one of a square, a circle, or a free form. Thus, the user is given complete control over which part of the multi-sample set to process.

[0017] In another aspect of the present invention, the method includes receiving user input indicating a plurality of tags. The generated dataset can associate the acquired microscopic images with the plurality of tags. The user input can indicate a plurality of tags related to a plurality of regions of interest. In the generated dataset, the plurality of tags can be associated with the microscopic images associated with each relevant region of interest. The plurality of tags may include at least one tag that qualifies the associated microscopic image as, in particular, one or more of ground truth, training data, validation data, or test data for a particular machine learning purpose.

[0018] The above-described tagging function of the aspects of the present invention is based on the following principles. · Each condition can generate a tag, each region can be associated with this tag, and vice versa. Thus, the corresponding mapping can be stored in the metadata of the image, thereby making the process particularly reproducible. · If the carrier already exists and is mapped in the user interface (e.g., a complete well plate with all rows and columns), each well can be automatically tagged in the metadata with the corresponding condition.

[0019] There are several uses for the tags provided in aspects of the present invention. For example, the tags can indicate the presence (above a threshold) of individual phosphors in the region of interest. In a well plate with multiple groups of tagged wells (e.g., odd wells contain one combination and even wells contain another combination), the quality condition (SNR) can be set or changed only for wells from one group or a selected subset of groups. Tagging or grouping can also be useful in scenarios where, for the reproducibility of results, multiple well plate positions often contain the same content. For example, if an experiment needs to be replicated multiple times using tags and saving the metadata of the first experiment, these conditions can be replicated in subsequent experiments.

[0020] Certain aspects of the present invention also provide a dataset, particularly a training dataset, a validation dataset, and / or a training dataset, that can be used in a machine learning model. The dataset may include a plurality of microscopic images, or references to a plurality of microscopic images. The images can be acquired using any of the methods disclosed herein.

[0021] Accordingly, embodiments may be based on or enable the use of a machine learning model or machine learning algorithm. Machine learning may refer to algorithms and statistical models that a computer system can use to perform a particular task without using explicit instructions, instead of relying on models and inferences. For example, in machine learning, data transformations inferred from the analysis of past data and / or training data may be used instead of rule-based data transformations. For example, image content may be analyzed using a machine learning model or a machine learning algorithm. To analyze image content, a machine learning model may be trained using training images as input and training content information as output. By training a machine learning model with a large number of training images and / or training sequences (e.g., words or sentences) and associated training content information (e.g., labels or annotations), the machine learning model "learns" to recognize image content, so that image content not included in the training data can be recognized using the machine learning model. The same principle may be used for other types of sensor data in the same way. By training a machine learning model with training sensor data and a desired output, the machine learning model "learns" the conversion between the sensor data and the output, which can be used to provide an output based on non-training sensor data provided to the machine learning model. The provided data (e.g., sensor data, metadata, and / or image data) may be preprocessed to obtain feature vectors used as input to the machine learning model.

[0022] A machine learning model may be trained using training input data. The above example uses a training method called "supervised learning". In supervised learning, a machine learning model is trained using a plurality of training samples, where each sample may include a plurality of input data values and a plurality of desired output values, i.e., each training sample is associated with a desired output value. By specifying both the training sample and the desired output value, the machine learning model "learns" during training what output value to provide based on input samples similar to the provided samples. In addition to supervised learning, semi-supervised learning may be used. In semi-supervised learning, some of the training samples lack the corresponding desired output values. Supervised learning may be based on a supervised learning algorithm (e.g., a classification algorithm, a regression algorithm, or a similarity learning algorithm). When the output is restricted to a limited set of values (categorical variables), i.e., when the input is classified into one of a limited set of values, a classification algorithm may be used. When the output may have any numerical value (within a range), a regression algorithm may be used. A similarity learning algorithm may be similar to both a classification algorithm and a regression algorithm, but is based on learning from examples using a similarity function that measures how similar or related two objects are. In addition to supervised or semi-supervised learning, unsupervised learning may be used to train a machine learning model. In unsupervised learning, only the input data may be supplied, and an unsupervised learning algorithm may be used to find structure in the input data (e.g., by grouping or clustering the input data, finding commonalities in the data). Clustering is the assignment of input data containing a plurality of input values to a plurality of subsets (clusters), such that the input values within the same cluster are similar according to one or more (predetermined) similarity criteria, but not similar to the input values contained in another cluster.

[0023] Reinforcement learning is the third group of machine learning algorithms. In other words, reinforcement learning may be used to train a machine learning model. In reinforcement learning, one or more software actors (referred to as "software agents") are trained to act in their surroundings. Based on the actions taken, a reward is calculated. Reinforcement learning is based on training one or more software agents to select actions such that (as revealed by an increase in reward) the cumulative reward increases and a software agent that performs better on a given task is obtained.

[0024] Furthermore, several techniques may be applied to part of the machine learning algorithm. For example, feature representation learning may be used. In other words, the machine learning model may be trained at least in part using feature representation learning, and / or the machine learning algorithm may include a feature representation learning component. Feature representation learning algorithms, which may be referred to as representation learning algorithms, not only store information in their input but may also usefully transform the information, often as a preprocessing step before performing classification or prediction. Feature representation learning may be based on, for example, principal component analysis or cluster analysis.

[0025] In some examples, anomaly detection (i.e., outlier detection) may be used, which aims to provide the identification of input values that raise suspicion by being significantly different from most of the input or training data. In other words, the machine learning model may be trained at least in part using anomaly detection, and / or the machine learning algorithm may include an anomaly detection component.

[0026] In some examples, the machine learning algorithm may use a decision tree as a prediction model. In other words, the machine learning model may be based on a decision tree. In a decision tree, an observation regarding an item (e.g., a set of input values) may be represented by a branch of the decision tree, and an output value corresponding to this item may be represented by a leaf of the decision tree. The decision tree may support both discrete and continuous values as output values. When discrete values are used, the decision tree may be represented as a classification tree, and when continuous values are used, the decision tree may be represented as a regression tree.

[0027] Correlation rules are another technique that can be used in machine learning algorithms. In other words, the machine learning model may be based on one or more correlation rules. Correlation rules are created by identifying relationships between variables in a large amount of data. The machine learning algorithm may identify and / or utilize one or more correlative rules that represent knowledge derived from the data. These rules may be used, for example, to store, manipulate, or apply knowledge.

[0028] Machine learning algorithms are typically based on a machine learning model. In other words, the term "machine learning algorithm" may represent a set of instructions that can be used to create, train, or use a machine learning model. The term "machine learning model" may represent a data structure and / or set of rules that represents learned knowledge (e.g., based on training performed by a machine learning algorithm). In an embodiment, the usage of a machine learning algorithm may mean the usage of one underlying machine learning model (or multiple underlying machine learning models). The usage of a machine learning model may mean that the machine learning model and / or the set of data structures / rules that are the machine learning model are trained by a machine learning algorithm.

[0029] For example, the machine learning model may be an artificial neural network (ANN). An ANN is a system influenced by biological neural networks, such as those found in the retina or brain. An ANN includes a plurality of interconnected nodes and a plurality of junctions between the nodes, so-called edges. Usually, there are three types of nodes, namely input nodes that receive input values, hidden nodes that are connected to other nodes (only), and output nodes that provide output values. Each node may represent an artificial neuron. Each edge may transmit information from one node to another. The output of a node may be defined as a (non-linear) function of its input (e.g., the sum of its inputs). The input of a node may be used in a function based on the "weights" of the edges or nodes that provide the input. The weights of the nodes and / or edges may be adjusted during the learning process. In other words, training an artificial neural network may include adjusting the weights of the nodes and / or edges of the artificial neural network to obtain a desired output for a given input.

[0030] Alternatively, the machine learning model may be a support vector machine, a random forest model, or a gradient boosting model. A support vector machine (i.e., a support vector network) is a supervised learning model with an associated learning algorithm that can be used to analyze data (e.g., in classification or regression analysis). A support vector machine may be trained by providing an input with a plurality of training input values belonging to one of two categories. A support vector machine may be trained to assign new input values to one of two categories. Alternatively, the machine learning model may be a Bayesian network, which is a probabilistic directed acyclic graphical model. A Bayesian network may represent a set of probabilistic variables and their conditional dependencies using a directed acyclic graph. Alternatively, the machine learning model may be based on a genetic algorithm, which is a heuristic method that mimics the process of search algorithms and natural selection.

[0031] The use cases that can be considered for the results generated by the embodiments of the present invention are diverse. For example, the model can be trained to detect a specific cell phenotype or target specific cell organelles and compare them at different SNRs. This can be useful for creating a model to evaluate the phototoxicity of an assay or for creating a simpler model by using simple cell detection and segmentation (measurements such as the number of cells and the area of cell nuclei, or whether some markers are displayed).

[0032] Embodiments of the present invention also provide a data processing device, particularly a computer device, or a control unit for a microscope, including means for implementing any of the methods disclosed herein. A computer program having program code for implementing any of the methods disclosed herein when executed on a processor is also provided.

[0033] This disclosure can be better understood by referring to the following drawings.

Brief Description of the Drawings

[0034]

Figure 1

Figure 2

Figure 3

Modes for Carrying Out the Invention

[0035] Here, exemplary embodiments of the present invention will be described that provide an autonomous microscope image acquisition workflow for automatic illumination drive, along with the final steps for big data acquisition and tagging for machine learning purposes.

[0036] As shown illustratively in FIG. 2, FIG. 1 shows an embodiment of an automatic illumination workflow 100 executed in relation to samples within a well plate 200. In particular, the illustrated well plate is merely an example of a sample set that may include any collection of slides, Petri dishes, and the like.

[0037] In step 102 of FIG. 1, one or more ROIs are selected as indicated by user selection. For example, the user can request a full sample scan or selection of wells within the well plate (e.g., by row or column). As shown by ROI 204a in FIG. 2, the user can also individually tag well 202. The user can select wells using a specific closed shape (e.g., a column 204b in a row, a square, a rectangle 204c, a circle, a free form, etc.). The ROIs can be tagged independently, and the corresponding tags for the acquired images within the regions are stored as metadata associated with the acquired images.

[0038] The following are examples of how data is labeled according to tags defined by the user when the experiment is conducted. · In the first example, regions 1, 2, 3 are acquired with high SNR and tagged as "ground truth". · In the second example, regions 4, 5, 6 are acquired at multiple SNRs and tagged as "input data". · In the third example, as shown in FIG. 2, rows from A to D are acquired with different SRNs 206. · In the fourth example, rows A and B of the well plate are acquired with high SNR with phosphors X1 and X2 and labeled as "protein A" and "protein B". · In the fifth example, rows C and D of the well plate are acquired with phosphors X2 and X3 having low phototoxic illumination conditions and labeled as "protein B" and "protein C".

[0039] Returning to FIG. 1, at step 104, the experiment to be executed can be defined. The definition of such an experiment can help to specify aspects such as the definition of settings that are generally applicable to the entire sample, the selection of one or more modalities (e.g., wide field (WF), confocal laser scanning microscope (CLSM), transmitted light, differential interference contrast (DIC) microscope, RGB, or any other modality), the selection of a phosphor for an image for each region of interest (ROI) which can vary for each ROI, field of view (FOV), resolution, frame period, objective lens, z-range, time series, or a combination thereof.

[0040] As described above, according to an embodiment of the present invention, a tag can be associated with ROI 204. Each tag defined for each ROI 204 can have its own experiment definition.

[0041] At step 106, the conditions to be applied to each ROI are defined. Exemplary conditions can be selected from one or more of the following. · Achievement of a specific SNR · Presence of a specific phosphor or combination of phosphors having a predetermined SNR · A specific density of photons and / or energy of a specific phosphor, for example, to detect a fluorescence threshold or to avoid phototoxicity

[0042] It is also possible to repeat one task by changing one parameter (e.g., SNR, fluorescence threshold, etc.).

[0043] Those skilled in the art will understand that the sequence of steps 102, 104, and 106 shown in FIG. 1 is not essential and can be changed.

[0044] At step 108, automatic acquisition is executed. The microscope moves to the regions respectively defined at step 110 and, at step 112, loads the conditions (defined at step 106). Additional parameters such as autofocus, drift correction, or changes in the object can be considered.

[0045] In step 114, automatic illumination is performed where the conditions defined in step 106 are applied to the specified phosphor and / or modality. In one embodiment, the automatic illumination function is performed as disclosed in International Publication No. WO 2022 / 028694 entitled "Method for adjusting the illumination in a fluorescence microscope, and corresponding fluorescence microscope" by the applicant. In step 116, one or more images are acquired.

[0046] When an image is acquired for the current region 204, process 100 moves to the next region 204 and is repeated until all ROIs 204 have been processed.

[0047] The output of process 100 in step 118 is a set of acquired images with auxiliary information, preferably stored in the metadata associated with the images, which may then be used for various technical tasks including training, validation, or optimization in other ways of a machine learning model. In addition to the acquired images (or references to the images), the output data set may include one of a plurality of parameters selected for the map of the well plate with ROIs in the case of automatic illumination, tags, additional metadata, and well plates.

[0048] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items and may be abbreviated as " / ".

[0049] Although some aspects have been described in the context of an apparatus, it is clear that these aspects also represent a description of corresponding methods, where a block or device corresponds to a method step or a feature of a method step. Similarly, aspects described in the context of method steps also represent a description of corresponding blocks or items or features of a corresponding apparatus.

[0050] Some embodiments relate to a microscope that includes a system as described in connection with one or more of FIGS. 1-2. Alternatively, the microscope may be part of a system as described in connection with one or more of FIGS. 1-2, or may be connected to a system as described in connection with one or more of FIGS. 1-2. FIG. 3 shows a schematic diagram of a system 300 configured to implement the methods described herein. System 300 includes a microscope 310 and a computer system 320. Microscope 310 is configured to image and is connected to computer system 320. Computer system 320 is configured to implement at least a portion of the methods described herein. Computer system 320 may be configured to execute a machine learning algorithm. Computer system 320 and microscope 310 may be separate entities, or may be integrated within a common housing. Computer system 320 may be part of the central processing system of microscope 310, and / or computer system 320 may be part of a subordinate component of microscope 310, such as a sensor, actuator, camera, or illumination unit of microscope 310.

[0051] The computer system 320 may be a local computer device (e.g., a personal computer, laptop, tablet computer, or mobile phone) having one or more processors and one or more storage devices, or a distributed computer system (e.g., a cloud computing system having one or more processors and one or more storage devices distributed across various locations, e.g., local clients and / or one or more remote server farms and / or data centers). The computer system 320 may include any circuitry or combination of circuitry. In one embodiment, the computer system 320 may include one or more processors, which can be of any type. As used herein, a processor may mean any type of arithmetic circuit, such as, for example, a microscope or microscope component (camera) or any other type of processor or processing circuit, e.g., a microprocessor, a microcontroller, a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a graphics processor, a digital signal processor (DSP), a multi-core processor, a field programmable gate array (FPGA), but is not limited thereto. Other types of circuitry that may be included in the computer system 320 may be custom circuitry, application specific integrated circuits (ASIC), etc., e.g., this may be one or more circuits (such as communication circuits) used in wireless devices such as mobile phones, tablet computers, laptop computers, two-way radios, and similar electronic systems. The computer system 320 may include one or more storage devices that may include one or more memory elements suitable for a particular application, e.g., main memory in the form of random access memory (RAM), one or more hard drives and / or one or more drives for handling removable media such as compact discs (CDs), flash memory cards, digital video discs (DVDs), etc.The computer system 320 can also include a display device, one or more speakers, and a keyboard and / or a controller, which can include a mouse, a trackball, a touch screen, a voice recognition device, or any other device that enables a system user to input information to and receive information from the computer system 320.

[0052] Some or all of the method steps may be performed by (or using) a hardware device such as a processor, a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, any one or more of the highly important method steps may be performed by such a device.

[0053] Depending on specific implementation requirements, embodiments of the present invention may be implemented in hardware or software. This implementation is executable using a non-transitory storage medium, which is a digital storage medium such as, for example, a floppy disk, a DVD, a Blu-Ray, a CD, a ROM, a PROM, an EPROM, an EEPROM, or a flash memory, and stores electronically readable control signals so as to cooperate with (or be able to cooperate with) a programmable computer system such that each method is implemented. Thus, the digital storage medium may be computer-readable.

[0054] Some embodiments of the present invention include a data carrier having electronically readable control signals that can cooperate with a programmable computer system such that any of the methods described herein are implemented.

[0055] Generally, embodiments of the present invention can be implemented as a computer program product comprising program code, which operates to implement any of the methods when the computer program product is executed on a computer. The program code may be stored, for example, on a machine-readable carrier.

[0056] Another embodiment includes a computer program for implementing any of the methods described herein, stored on a machine-readable carrier.

[0057] Thus, in other words, embodiments of the present invention are computer programs having program code for implementing any of the methods described herein when the computer program is executed on a computer.

[0058] Thus, another embodiment of the present invention is a storage medium (or data carrier or computer-readable medium) including a stored computer program for implementing any of the methods described herein when executed by a processor. The data carrier, digital storage medium or storage medium is typically tangible and / or non-transitory. Another embodiment of the present invention is an apparatus as described herein including a processor and a storage medium.

[0059] Thus, another embodiment of the present invention is a data stream or signal sequence representing a computer program for implementing any of the methods described herein. The data stream or signal sequence may be configured, for example, to be transferred via a data communication connection, such as the Internet.

[0060] Another embodiment includes processing means, such as a computer or programmable logic device configured or adapted to implement any of the methods described herein.

[0061] Another embodiment includes a computer having an installed computer program for implementing any of the methods described herein.

[0062] Another embodiment of the invention includes an apparatus or system configured to transfer (e.g., electronically or optically) a computer program for implementing any of the methods described herein to a receiver. The receiver may be, for example, a computer, a mobile device, a storage device, etc. The apparatus or system may include, for example, a file server for transferring the computer program to the receiver.

[0063] In some embodiments, a programmable logic device (e.g., a field programmable gate array) may be used to perform some or all of the functions of the methods described herein. In some embodiments, the field programmable gate array may cooperate with a microprocessor to implement any of the methods described herein. Generally, preferably, the method is implemented by any hardware device.

Claims

1. A method for computer-implemented image acquisition, the method comprising: Receiving user input to indicate (106) at least one quality condition (206), including a target signal-to-noise ratio, associated with a plurality of regions of interest (204) of a sample array (200) imaged using a fluorescence microscope (310); Automatically causing the fluorescence microscope (310) to automatically acquire at least one microscope image for each region of interest (204) using an illumination setting automatically determined (114) such that the target signal-to-noise ratio is satisfied (108); Generating (118) a dataset comprising the acquired microscope images or references to the microscope images, which can be used to generate, train, validate and / or test a machine learning model; A method comprising at least.

2. The at least one quality condition (206) includes a target signal-to-noise ratio applied to the entirety of the plurality of regions of interest, The method according to claim 1.

3. The at least one quality condition (206) includes a target signal-to-noise ratio applied to one selected from the plurality of regions of interest (204), The method according to claim 1 or 2.

4. The target signal-to-noise ratio is a specific signal-to-noise ratio value or a minimum signal-to-noise ratio value, The method according to any one of claims 1 to 3.

5. The target signal-to-noise ratio is applied to one selected channel out of a plurality of channels, The method according to any one of claims 1 to 4.

6. The at least one quality condition (206) A threshold condition based on light dose, A threshold condition based on phosphor quality, or A threshold condition based on image quality Further includes at least one of, The method according to any one of claims 1 to 5.

7. The sample array (200) includes a plurality of separate samples (202), Optionally, the sample array (200) is a well plate, The method according to any one of claims 1 to 6.

8. The method includes a further step of receiving user input to indicate (102) the selection of the plurality of regions of interest (204), The selection is All samples of the sample array (200), One or more individual samples (204a) of the sample array (200), One or more rows and / or columns of the sample (204b) of the sample array (200), Optionally, a shape that encloses one or more samples (204c) of the sample array (200), which is one of a square, a circle, or a free form indicating The method according to claim 7.

9. The method includes a further step of receiving user input indicating a plurality of tags, In the generated dataset, the acquired microscopic images are associated with the plurality of tags, The method according to any one of claims 1 to 8.

10. The user input indicates the plurality of tags related to the plurality of regions of interest, In the generated dataset, the plurality of tags are associated with the microscopic images associated with each associated region of interest, The method according to claim 9.

11. The plurality of tags include at least one tag that qualifies the associated microscopic image as one or more of, in particular, ground truth, training data, validation data, or test data, for a specific machine learning purpose, The method according to claim 9 or 10.

12. A data processing device, in particular a control unit for a computer device (320) or a microscope (310), including means for implementing the method according to any one of claims 1 to 11, A data processing device.

13. A computer program, having program code for implementing the method according to any one of claims 1 to 11 when executed on a processor, A computer program.

14. A dataset that can be used in a machine learning model, in particular a training dataset, a validation dataset, and / or a training dataset, including a plurality of microscopic images obtained using the method according to any one of claims 1 to 11, or references to the plurality of microscopic images, A dataset.

15. A microscope (310), in particular a fluorescence microscope, configured for use in the method according to any one of claims 1 to 11, A microscope (310).

Images