Assessing drug efficacy using optical coherence tomography data

Abstract

There is presented a computer-implemented method (100) for assessing drug efficacy, comprising receiving (102) an optical coherence tomography (OCT) data set, analysing (104) the OCT data set to obtain a plurality of metric values, wherein the plurality of metric values is comprising a metric value for one or more or all of the metrics within a group of metrics comprising a cut-off spatial frequency metric, wherein the value of the cut-off spatial frequency metric relates to a frequency separating high- and low frequency content, and a ratio metric, wherein the value of the ratio metric relates to a ratio between high- and low spatial frequency content, and assessing (106) a drug efficacy based on the plurality of metric values.

Description

[0001] ASSESSING DRUG EFFICACY USING OPTICAL COHERENCE TOMOGRAPHY DATA

[0002] FIELD OF THE INVENTION

[0003] The present invention relates to assessing drug efficacy, and more particularly to a computer-implemented method for assessing drug efficacy using an Optical Coherence Tomography (OCT) data set, a corresponding computer program product, a corresponding computer-readable medium, a method for acquiring the Optical Coherence Tomography (OCT) data set and assessing drug efficacy via the computer-implemented method, and a system for obtaining the Optical Coherence Tomography (OCT) data set and assessing drug efficacy via the computer-implemented method.

[0004] BACKGROUND OF THE INVENTION

[0005] The current protocols for developing drugs, such as anti-cancer drugs, are expensive and time consuming. Methods that could facilitate high-throughput screening could significantly reduce the time and cost towards identifying one or more candidates. However, the characterisation of the drug efficacy remains a significant bottleneck, such as in terms of time and / or costs. The characterisation typically involves a series of chemical assays testing for various metrics of the drug efficacy including changes in metabolism and markers of apoptosis, as well as fixation and histological validation, which is resource demanding in terms of time and / or costs.

[0006] Hence a method for assessing drug efficacy, which is efficient in terms of time and / or costs would be advantageous.

[0007] The application JP 2023 059459 A describes that an image processing device includes a processing unit which generates a phase differential image of a sample on the basis of at least two front OCT images obtained by imaging the sample with OCT.

[0008] The reference by Yan Feng, et al., " Optical coherence tomography for multicellular tumor spheroid category recognition and drug screening classification via multi-spatial-superficial-parameter and machine learning", Biomedical optics express, vol. 15, no. 4, 4 March 2024 (2024-03-04), page 2014, United States ISSN: 2156-7085, DOI: 10.1364 / BOE.514079, the authors developed cross-statistical, cross-screening, and composite-hyperparameter feature processing methods in conjunction with 12 machine learning models to assess changes within the multicellular tumor spheroids (MCTS) internal structure, describes optical coherencetomography for multicellular tumor spheroid category recognition and drug screening classification via multi-spatial-superficial-parameter and machine learning.

[0009] The reference by Zhang Linyi et al., " Quantifying the drug response of patient-derived organoid clusters by aggregated morphological indicators with multi-parameters based on optical coherence tomography", Biomedical optics express, vol. 14, no. 4, 29 March 2023 (2023-03-29), page 1703, the authors developed a method for label-free, continuous tracking imaging and quantitative analysis of drug efficacy using PDOs. A self-developed optical coherence tomography (OCT) system was used to monitor the morphological changes of Patient-derived organoids (PDOs) within 6 days of drug administration.

[0010] SUMMARY OF THE INVENTION

[0011] It may be seen as an object of the present invention to provide a method for assessing drug efficacy, which is efficient in terms of time and / or costs. It may be a further object of the present invention to provide an alternative to the prior art.

[0012] Thus, the above-described object and several other objects are intended to be obtained in a first aspect of the invention by providing a computer-implemented method for assessing drug efficacy, such as drug efficacy in one or more 3D tissue models, comprising

[0013] Receiving an optical coherence tomography (OCT) data set, such as a 1D data set, such as a 2D image, such as a volumetric data set,

[0014] Analysing, such as automatically analysing, the optical coherence tomography (OCT) data set to obtain a plurality of metric values, wherein the plurality of metric values is comprising a metric value for one or more or all of the metrics within a group of metrics comprising:

[0015] i. A cut-off spatial frequency metric, wherein the value of the cut-off spatial frequency metric relates to, such as is a measurement of, a frequency separating high- and low frequency content, such as obtained in a spatial frequency analysis of the received optical coherence tomography (OCT) data set, and

[0016] II. a ratio metric, wherein the value of the ratio metric relates to, such as is a measurement of, a ratio between high- and low spatial frequency content, such as obtained in a spatial frequency analysis of the received optical coherence tomography (OCT) data set,

[0017] and

[0018] Assessing a drug efficacy based on the plurality of metric values.The invention may be advantageous for providing a computer-implemented method for assessing a drug efficacy, which enables assessing the drug efficacy using an Optical Coherence Tomography (OCT) data set, such as using only an OCT data set. This may enable automating assessment of drug efficacy, such as longitudinal assessment of drug efficacy, which may be beneficial for reducing a need for human intervention during the analysis and / or increasing throughput, e.g., as compared to chemical assays (which may include a series of chemical assays testing for various metrics of the drug efficacy including changes in metabolism and markers of apoptosis, as well as fixation and histological validation).

[0019] The invention may be advantageous for presenting a cost-effective and / or time-effective method, such as relative to current protocols for developing drugs, such as anti-cancer drugs, which are expensive and time consuming. This may in turn be advantageous for facilitating high-throughput screening, which could significantly reduce the time and cost towards identifying candidate drugs.

[0020] A possible advantage of the method is that it enables analysing data for one or more 3D tissue models while dispensing with a need to necessarily analyse each cellular unit, such as each spheroid, individually.

[0021] Our invention is a novel methodology for assessing the same metrics of drug efficacy using optical imaging. Our method will significantly reduce the need for human intervention during the analysis thereby significantly increasing throughput. The method is based on optical coherence tomography images and our novel image processing protocol.

[0022] Another possible advantage may be that the method relies on one or more physics-driven metrics, and as such presents a physics-based method, such as wherein changes in the OCT data set, such as changes in one or more 3D tissue models depicted in the OCT data set, cause a physical interaction, such as a change in scattering, with the probing light utilized for OCT. A possible advantage of the method being a physics-based method, may in turn be that it will apply generally. For example, application for a new cell line and or a new drug might not need new training (as might be the case for a method based solely on machine learning).

[0023] 'Drug efficacy' is understood as is common in the art, such as a response that can be obtained with a drug, e.g., in an in vitro model, such as one or more 3D tissue models.

[0024] '3D tissue model' is understood as is common in the art, such as one or more three-dimensional (3D) tissue model where 3D cell cultures are grown in 3D space. One or more 3D tissue models may comprise scaffold-based 3D cell cultures, scaffold-free 3D cell cultures, such as one or more spheroids or one or more organoids, or a microfluidic organ-on-a-chip. A3D tissue model may comprise cellular units, such as one or more spheroids or one or more organoids, and a matrix, such as a hydrogel, wherein the cellular units are suspended in the matrix. One or more 3D tissue models comprising cellular units in the form of one or more tumour spheroids, which can be cultured in well plates, may allow for the testing of many candidate drugs simultaneously.

[0025] 'Drug efficacy in one or more 3D tissue models' is understood as is common in the art, which is in particular relevant to the extent the one or more 3D tissue models translate well between the model and humans.

[0026] 'Optical Coherence Tomography (OCT)' is understood as is common in the art. OCT may be considered beneficial as it enables performing an "optical biopsy", allowing real-time, in situ visualization of tissue microstructure and pathology without necessitating removing the sample and prior processing.

[0027] 'Data' is understood as is common in the art and may be understood to be digital data.

[0028] 'Data set,' is understood as is common in the art, such as a set of data comprising a plurality of data points, such as spatially resolved data points in 1 or 2 or 3 dimensions, such as a 1D data set (such as a depth-resolved OCT intensity profile along the z-axis at a single point in the xy-plane, which may be referred to as an A-line OCT scan), such as a 2D image (such as a cross-sectional 2D image, also called B-scans, created by laterally scanning the OCT beam and generating sequential A-scans), such as a volumetric data set (such as with sequentially generated B-scans, which results in 3D data, also called C-scans).

[0029] 'Optical coherence tomography (OCT) data set' is understood as is common in the art, such as a data set comprising data obtainable by, such as obtained by, OCT. The optical coherence tomography (OCT) data set may comprise information, such as spatially resolved data points in 1 or 2 or 3 dimensions, of one or more 3D tissue models, such as one or more organoids or one or more spheroids.

[0030] 'Receiving an optical coherence tomography (OCT) data set' is understood as is common in the art, such as encompassing the mere receipt of the information comprising the optical coherence tomography (OCT) data set, such as without necessarily (physically) acquiring the optical coherence tomography (OCT) data set.

[0031] A 'metric value' is understood as is common in the art, such as is understood to be a value of a metric, such as wherein a metric is a quantifiable statistic, indicator or property. Examples of a metric could be volume or frequency threshold. The value of a metric is understood to bethe quantified value for the metric (for an entity or system), e.g., the metric value in case of the metric being volume (e.g., of one or more spheroids in one or more 3D tissue models) could be the quantified value of X mm3(with 'X' representing an example numerical value of volume), or the metric value in case of the metric being a frequency threshold (e.g., in a spectral representation of one or more spheroids in one or more 3D tissue models) could be the quantified value of Y m-1(with 'X' representing an example numerical value of frequency or log of frequency).

[0032] A 'plurality of metric values' is understood as is common in the art. A possible advantage of having a plurality of metric values may be that it may yield more information being relevant for assessing drug efficacy, such as enabling obtaining a unique representation of a drug efficacy.

[0033] In an alternative aspect or embodiment, drug efficacy may be based on a single metric value, such as (solely) on the 'cut-off spatial frequency metric' or the 'ratio metric.

[0034] 'Analysing the optical coherence tomography (OCT) data set to obtain a plurality of metric values' is understood as is common in the art, such as wherein a processing unit receives the data set as an input and provides as an output to be obtained the plurality of metric values, such as wherein the values of each of the plurality of metric values is not known in advance of the analysis and / or wherein each of the plurality of metric values depends on the content of the optical coherence tomography (OCT) data set.

[0035] By 'a cut-off spatial frequency metric' may be understood a metric, wherein the value of the cut-off spatial frequency metric relates to or is (a measurement of) a spatial frequency separating (at least some or all) high frequency content form (at least some or all) low frequency content (such as of the optical coherence tomography (OCT) data set), such as in a spectral domain or a frequency domain. The cut-off spatial frequency metric may be obtained in a spatial frequency analysis. The cut-off spatial frequency metric may be a frequency separating spectral content relating to different cell stages or phases, such as proliferating cells and apoptotic cells. These cells may have differentiating characteristics, e.g., of cell size and / or nuclei size, which affects light scattering, which is exactly what is measured in acquired OCT data. Thus, the OCT data may be seen as directly reflecting, via the physics of light matter interaction, such as scattering of the cells (membrane, organelles) and nuclei, and the OCT principle, the cell stage or phase.

[0036] Analysing the optical coherence tomography (OCT) data set to obtain the cut-off spatial frequency metric may comprise identifying high- and low frequency content (such as of the optical coherence tomography (OCT) data set), including but not necessarily limited to fittingor regression. Furthermore, the cut-off spatial frequency metric may be based on at least the spectral values of the high- and low frequency content, such as wherein the cut-off spatial frequency metric at least indicates a spectral value being above a statistical metric of the low frequency content and below the statistical metric of the high frequency content, such as wherein the statistical metric is an, optionally weighted, average or median value.

[0037] 'Spectral domain' and 'Frequency domain' are understood to be interchangeable and is each understood as is common in the art, such as relating to data and / or functions arranged with respect to frequency, such as being spectrally resolved.

[0038] The 'high- and low frequency' are understood as relative to each other, such as the 'high frequency content' being higher in frequency than the 'low frequency content', and vice versa.

[0039] The 'cut-off spatial frequency metric' may in an embodiment correspond to a frequency value at a point of intersection between two Gaussian (mathematical) functions or distributions fitted to a spectral representation (such as a frequency histogram) of the optical coherence tomography (OCT) data set. The logarithm of the frequency data may replace the frequency data, for example in this embodiment.

[0040] By 'a ratio metric' may be understood a metric, wherein the value of the ratio metric relates to or is (a measurement of) a ratio between high- and low spatial frequency content, such as of the optical coherence tomography (OCT) data set (such as an area under a curve in a frequency histogram on either side of a cut-off spatial frequency metric separating high- and low-frequency content), such as obtained in a spatial frequency analysis of the received OCT data. The cut-off spatial frequency may be given as defined for the cut-off spatial frequency metric above, such as a frequency separating high- and low frequency content. The ratio metric may be a value reflecting cell size characteristics.

[0041] High- and low spatial frequency content may be an area under a curve in a frequency histogram on either side of a cut-off spatial frequency metric separating high- and low-frequency content. Alternatively, high- and low spatial frequency content may be determined based on an assignment of the total frequency content to high- and low spatial frequency (e.g., mathematical) distributions, e.g., via fitting of two Gaussian functions, wherein highland low spatial frequency content may be determined based on the content of each of these distributions, such as areas under each of two Gaussian functions.

[0042] The spectral representation, such as the histogram, may represent for each frequency, or frequency bin, a "count", which should represent how many of the given frequencies are inthe specific bins, so a sum. So it should represent how many of the pixels in the image are represented by the specific frequency bin. Alternatively or additionally, the spectral representation may represent an amplitude spectrum (such as based directly on Fourier coefficient) or a power spectrum (such as based on the squared magnitude of Fourier coefficients).

[0043] In general, an analysis for obtaining the 'cut-off spatial frequency metric' and / or the 'a ratio metric' may involve a Fourier analysis of the optical coherence tomography (OCT) data set.

[0044] 'Assessing a drug efficacy based on the plurality of metric values.' is understood as is common in the art, such as providing a measure, such as an estimate, of a drug efficacy, wherein providing said measure is partially or fully based on the plurality of metric values.

[0045] According to an embodiment, there is presented the computer-implemented method, wherein the group of metrics furthermore comprises one or more or all of:

[0046] A size metric, such as wherein the value of the size metric relates to, such as is a measurement of, one or more 3D tissue model sizes, such as one or more 3D tissue model volumes,

[0047] A fragmentation metric, such as wherein the value of the fragmentation metric relates to, such as is a measurement of, a fragmentation of one or more 3D tissue models, and / or

[0048] an optical attenuation coefficient metric, such as wherein the value of the optical attenuation coefficient relates to, such as is a measurement of, an attenuation coefficient of one or more 3D tissue models in, such as depicted in, the optical coherence tomography data.

[0049] The 'size metric' may relate to, e.g., a volume or a cross-sectional area of one or more 3D tissue models, such as one or more spheroids. It may be measured relative to a control, such as one or more spheroids in other one or more 3D tissue models, which is similar in ideally all aspects, except it is not treated with a drug.

[0050] The 'fragmentation metric' may relate to a degree of fragmentation, such as unfittable data, relative to a control sample.

[0051] The 'optical attenuation coefficient metric' is understood as is common in the art, and may be obtained along each OCT A-scan, group of OCT A-scans, or as 2D projection maps. It may be understood, that the optical attenuation relates to optical attenuation of one or more 3D tissue models in the optical coherence tomography (OCT) data set.According to an embodiment, there is presented the computer-implemented method, wherein the metrics for which the plurality of metric values is obtained enables obtaining a unique representation of a drug efficacy.

[0052] By 'a unique representation of a drug efficacy' is understood that a plurality of metric values is unique for a (corresponding) drug efficacy. This may be advantageous for enabling determining the drug efficacy from the plurality of metric values. For example, different drug efficacies may entail distinguishably (such as taking a resolution, accuracy and precision of each metric value into account) different sets of (a plurality of) metric values, such as wherein obtaining the plurality of metric values enables unambiguously determining drug efficacy.

[0053] According to an embodiment, there is presented the computer-implemented method, wherein Assessing a drug efficacy based on the plurality of metric values

[0054] is based on a predetermined correlation between predetermined sets of metric values and corresponding drug efficacies.

[0055] By utilizing a predetermined correlation, it may be possible to effectively, precisely and / or accurately determine a drug efficacy from the plurality of metric values. The predetermined correlation could be based on previously determined sets of metric values and corresponding drug efficacies. Basing an assessment of drug efficacy on a predetermined correlation could be carried out via a look-up table and / or a machine-learning algorithm (such as trained on the previously determined sets of metric values and corresponding drug efficacies).

[0056] For example, it could be possible to train a (machine-learning) algorithm using existing assays and correlation with the sets of metric values, so that the algorithm could generate assay results (drug efficacies) based on imaging and the obtained plurality of metrics alone.

[0057] According to an embodiment, there is presented the computer-implemented method, wherein the optical coherence tomography (OCT) data set is a volumetric data set.

[0058] 'Volumetric (data set)' is understood as is common in the art, such as a data set including such as data points spatially distributed in a 3D space, each data point thus comprising a metric value or a property value (such as an intensity value, a scattering value, or the like) and 3D coordinates (e.g., x, y, and z).

[0059] A volumetric data set could be beneficial, e.g., since it comprises more information than a 1D or a 2D data set, and / or since it enables 3D analysis, such as 3D segmentation and 3Dmetrics, such as volume. A volumetric data set may enable exploiting OCT's potential, such a fully exploiting OCT's potential (e.g., compared to relying on 1D data or 2D data).

[0060] According to an embodiment, there is presented the computer-implemented method, wherein the plurality of metric values is comprising a metric value for both of the metrics within a group of metrics comprising:

[0061] the cut-off spatial frequency metric, and

[0062] the ratio metric.

[0063] Utilizing both metrics may be advantageous since it enables that the assessment of drug efficacy can be made based on more information.

[0064] According to an embodiment, there is presented the computer-implemented method, wherein the plurality of metric values is comprising a metric value for all of the metrics within a group of metrics comprising:

[0065] the cut-off spatial frequency metric,

[0066] the ratio metric,

[0067] the size metric, and

[0068] the optical attenuation coefficient metric,

[0069] and wherein the plurality of metric values is optionally further comprising

[0070] the fragmentation metric.

[0071] Utilizing all 4 or 5 metrics may be advantageous since it enables that the assessment of drug efficacy can be made based on more information. Further, combining these 4 or 5 metrics might create a unique plurality of metric values which enables unambiguously identifying a drug efficacy.

[0072] According to an embodiment, there is presented the computer-implemented method, comprising:

[0073] Segmentation of the optical coherence tomography (OCT) data set.

[0074] 'Segmentation' is understood as is common in the art, such as identifying a region of interest, such as spheroids in one or more 3D tissue models.

[0075] An advantage of segmentation may be that it enables focusing on the part of, e.g., one or more 3D tissue models, where relevant information is to be gained, which may improve a signal-to-noise level, which may in turn improve accuracy and / or precision.According to an embodiment, there is presented the computer-implemented method, wherein the plurality of metric values is furthermore comprising a metric value for one or more or all of the metrics within a group of metrics comprising:

[0076] - A size metric, such as wherein the value of the size metric relates to, such as is a measurement of, one or more 3D tissue model sizes, such as one or more 3D tissue model volumes,

[0077] - A fragmentation metric, such as wherein the value of the fragmentation metric relates to, such as is a measurement of, a fragmentation of one or more 3D tissue models, and / or

[0078] an optical attenuation coefficient metric.

[0079] Adding one or more further metric values or metrics may be advantageous since it enables that the assessment of drug efficacy can be made based on more information. Further, the resulting combination of metrics might create a unique plurality of metric values which enables unambiguously identifying a drug efficacy.

[0080] According to an embodiment, there is presented the computer-implemented method, comprising:

[0081] - Analysing, such as automatically analysing, the optical coherence tomography data set to obtain the metric value for the cut-off spatial frequency metric.

[0082] An advantage of this may be that information from the OCT data set might be reflected into the metric value, e.g., as opposed to a predetermined metric value or a user-determined metric value, such as a predetermined spatial frequency, such as a predetermined metric value for the cut-off spatial frequency metric.

[0083] According to an embodiment, there is presented the computer-implemented method, comprising:

[0084] Obtaining the metric value for the cut-off spatial frequency metric in a spatial frequency analysis of the received optical coherence tomography data set.

[0085] According to an embodiment, there is presented the computer-implemented method, comprising analysing the optical coherence tomography (OCT) data set to obtain the cut-off spatial frequency metric, comprising identifying high- and low frequency content, such as of the optical coherence tomography (OCT) data set, including fitting or regression.

[0086] According to an embodiment, there is presented the computer-implemented method, wherein the cut-off spatial frequency metric corresponds to a frequency value, such as the cut-off spatial frequency, at a point of intersection between two Gaussian functions or distributionsfitted to a spectral representation, such as a frequency histogram, such as the histogram being fitted as the combination of two Gaussian curves and where the cut-off spatial frequency metric corresponds to the frequency value of the intersection between the two Gaussian curves, of the optical coherence tomography (OCT) data set, optionally wherein the logarithm of the frequency data have replaced or represent the frequency data.

[0087] An advantage of this may be that is enables identifying a frequency separating different types of cells, such as apoptotic and proliferating cells and / or enables parameterizing the presence and distribution of different types of cells, such as apoptotic and proliferating cells. Another possible advantage is that it enables determining the cut-off spatial frequency metric in an automated and / or unambiguous manner.

[0088] According to an embodiment, there is presented the computer-implemented method, comprising:

[0089] - Analysing (104) the optical coherence tomography data set to obtain the metric value for the ratio metric.

[0090] An advantage of this may be that information from the OCT data set might be reflected into the metric value, e.g., as opposed to a predetermined metric value or a user-determined metric value, such as a predetermined ratio, such as a predetermined metric value for the ratio metric.

[0091] According to an embodiment, there is presented the computer-implemented method, comprising:

[0092] Obtaining the metric value for the ratio metric in a spatial frequency analysis of the received optical coherence tomography data set.

[0093] According to an embodiment, there is presented the computer-implemented method, wherein:

[0094] - The value of the ratio metric relates to a ratio between high- and low spatial frequency content given by areas under a curve in a frequency histogram on either side of a cut-off spatial frequency separating high- and low-frequency content, wherein the cut-off spatial frequency is the cut-off spatial frequency metric value.

[0095] An advantage of this may be that it enables determining the value of the ratio metric in an automated and / or unambiguous manner. The curve may be given by the actual values in the histogram and / or one or more fitted curves, such as curves given by one or more Gaussian functions, such as said one or more Gaussian functions being fitted to the frequencyhistogram. Another possible advantage may be that it enables a simple and / or efficient method for determining both of the the cut-off spatial frequency metric and the ratio metric, e.g., because the the cut-off spatial frequency metric value can be determined for one metric and reused for the other metric (e.g., because the value of the cut-off spatial frequency metric is identical to the cut-off spatial frequency used for determining the ratio metric).

[0096] According to an embodiment, there is presented the computer-implemented method, wherein:

[0097] - The ratio metric is based on one or more predetermined spatial frequencies.

[0098] An advantage of this may be that it enables basing the ratio metric on frequencies which are (assessed) to be (particularly) advantageous. Another advantage may be that it is efficient, e.g., because the one or more frequencies upon which the ratio metric is based need not be determined during the computer-implemented method.

[0099] Having multiple predetermined spatial frequencies enables, e.g., determining spatial frequency content in bands. For example, a first frequency threshold separates low- and high-frequency content, e.g., wherein the frequency percentage below the first frequency threshold is considered to be an expression of a ratio between high- and low-frequency content. Additionally, a second frequency threshold may provide further information, e.g., regarding frequency content between the first and second frequency threshold and frequency content above the second frequency threshold, which further information may be useful for assessing drug efficacy.

[0100] In an embodiment, the one or more predetermined spatial frequencies comprises two or more frequency thresholds, such as a first frequency threshold and a second frequency threshold.

[0101] According to an embodiment, there is presented the computer-implemented method, wherein the plurality of metric values is furthermore comprising a metric value for:

[0102] - A band metric, wherein the value of the band metric relates to, such as is a measurement of, a ratio between spatial frequency content within a spatial frequency band and spatial frequency content outside the spatial frequency band, such as obtained in a spatial frequency analysis of the received optical coherence tomography data set,

[0103] such as wherein the band of the band metric is predetermined, such as wherein the upper and lower threshold frequency limits of the band of the band metric are predetermined.An advantage of a band metric may be that it yields information about frequency content not only above and / or below a certain frequency, but also in a band, such as between finite frequency thresholds. This may yield alternative and / or additional information, which may be useful for assessing drug efficacy.

[0104] The spatial frequency content within a spatial frequency band may be understood to be spatial frequency content between an upper spatial frequency threshold and a lower spatial frequency threshold.

[0105] Optionally, the spatial frequency band is predetermined.

[0106] According to an embodiment, there is presented the computer-implemented method, wherein the plurality of metric values is furthermore comprising metric values for a plurality, such as 2 or 3 or 4 or 5 or 10 or 20 or 100 or more, of:

[0107] - A band metric, wherein the value of the band metric relates to, such as is a measurement of, a ratio between spatial frequency content within a spatial frequency band and spatial frequency content outside the spatial frequency band, such as obtained in a spatial frequency analysis of the received optical coherence tomography data set,

[0108] such as wherein the bands of the band metrics are adjoining and non-overlapping.

[0109] An advantage of adjoining and non-overlapping bands may be that no information is lost between the bands (due to the adjoining), while the bands still represent different parts of the spectrum (due to the non-overlapping).

[0110] Optionally, the spatial frequency bands of the plurality of band metrics are predetermined, such as wherein the upper and lower threshold frequency limits of each of the bands of the plurality of band metrics are predetermined.

[0111] According to an embodiment, there is presented the computer-implemented method for assessing drug efficacy in one or more 3D tissue models, such as one or more 3D tissue models comprising cellular units, such as cellular units in the form of one or more tumour spheroids.

[0112] According to a second aspect there is presented a method for assessing drug efficacy in one or more 3D tissue models, said method comprising:

[0113] Providing the one or more 3D tissue models,Acquiring an optical coherence tomography (OCT) data set, such as a 1D data, such as a 2D image, such as a volumetric data set, of the one or more 3D tissue models, and

[0114] Assessing drug efficacy in the one or more 3D tissue models according to the computer-implemented method according to the first aspect based on the optical coherence tomography (OCT) data set being received by the computer-implemented method.

[0115] 'Providing (the) 3D one or more tissue models' is to be understood as is common in the art, and may encompass and correspond to each of creating the models and getting the models supplied.

[0116] By 'acquiring an optical coherence tomography (OCT) data set' is understood that the optical coherence tomography (OCT) data set is created, such as via the physical interaction of OCT probing light with the one or more 3D tissue models.

[0117] According to an embodiment, there is presented a method for assessing drug efficacy in one or more 3D tissue models, wherein

[0118] acquiring the optical coherence tomography (OCT) data set

[0119] comprises

[0120] acquiring a plurality of time-resolved optical coherence tomography (OCT) data sets, such as a plurality of time-resolved optical coherence tomography (OCT) images, such as a plurality of time-resolved volumetric images, of the one or more 3D tissue models.

[0121] By 'a plurality of time-resolved optical coherence tomography (OCT) data sets' is understood a plurality of optical coherence tomography (OCT) data sets, which have each been obtained at a different point in time with respect to each other, and wherein the optical coherence tomography (OCT) data sets also include information regarding the time each data set was obtained (at least relative to the other one or more optical coherence tomography (OCT) data sets.

[0122] An advantage of acquiring a plurality of time-resolved optical coherence tomography (OCT) data sets may be that it enables longitudinal studies.

[0123] The different points in time, at which the plurality of optical coherence tomography (OCT) data sets have (each) been obtained, may comprise one or more points in time before addition of one or more drugs, such as addition of one or more drugs to the one or more 3Dtissue models and one or more points in time after addition of one or more drugs, such as addition of one or more drugs to the one or more 3D tissue models.

[0124] According to an embodiment, there is presented a method for assessing drug efficacy in one or more 3D tissue models, wherein said one or more 3D tissue models comprise any cultured / cultivated 3D tissue consisting of several cell types. According to an embodiment, there is presented a method for assessing drug efficacy in one or more 3D tissue models, wherein said one or more 3D tissue models comprise scaffold-based one or more 3D cell cultures, scaffold-free 3D cell cultures or a microfluidic organ-on-a-chip. According to an embodiment, there is presented a method for assessing drug efficacy in one or more 3D tissue models, wherein said one or more 3D tissue models comprise one or more organoids.

[0125] According to an embodiment, there is presented a method for assessing drug efficacy in 3D tissue models, wherein said one or more 3D tissue models comprise one or more spheroids, such as tumour spheroids.

[0126] According to an embodiment, there is presented a method for assessing drug efficacy in 3D tissue models, wherein said one or more 3D tissue models comprise a hydrogel with a refractive index differing, such as being smaller than or greater than, from a refractive index of 1.33 by at least 1 %, such as at least 2 %, such as at least 5 %, such as at least 10 %, such as at least 20 %, such as at least 25 %, such as at least 50 %, such as at least 100 %.

[0127] An advantage of a refractive index differing, such as being smaller than or greater than, from a refractive index of 1.33, may be that it goes to increase a contrast in OCT data between cells and matrix in one more 3D tissue models for which OCT data are acquired.

[0128] In case of a similarity in refractive index between hydrogel and cells, the similarity may introduce subtle heterogeneities and scattering centers within the hydrogel (matrix), potentially increasing the baseline OCT backscatter.

[0129] An example of a hydrogel with a refractive index differing from 1.33 is given by HistoGel™.

[0130] Hydrogel, such as HistoGel™, may in embodiments be an aqueous specimen-processing gel that encapsulates and stabilizes specimens in a solidified medium. Hydrogel, such as HistoGel™, may provide mechanical support and minimize spheroid deformation during handling. The term 'histogel' may generally be understood to be HistoGel™.

[0131] Hydrogel, such as HistoGel™, may in general have a refractive index, RI -1.30 - 1.35, while spheroids typically range RI ~ 1.35 - 1.39.According to an embodiment, there is presented a method for assessing drug efficacy in 3D tissue models, wherein said one or more 3D tissue models comprises a hydrogel with a refractive index differing, such as being smaller than or greater than, from a refractive index of any value within 1.35-1.39 by at least 1 %, such as at least 2 %, such as at least 5 %, such as at least 10 %, such as at least 20 %, such as at least 25 %, such as at least 50 %, such as at least 100 %.

[0132] An advantage of a refractive index differing, such as being smaller than or greater than, from a refractive index of any value within 1.35-1.39, may be that it goes to increase a contrast in OCT data between spheroids and matrix in one more 3D tissue models for which OCT data are acquired.

[0133] According to an embodiment, there is presented a method for assessing drug efficacy in 3D tissue models,, wherein said one or more 3D tissue models comprises a hydrogel with a refractive index differing, such as being smaller than or greater than, from a refractive index of the one or more 3D tissue models by at least 1 %, such as at least 2 %, such as at least 5 %, such as at least 10 %, such as at least 20 %, such as at least 25 %, such as at least 50 %, such as at least 100 %.

[0134] An advantage of a refractive index differing, such as being smaller than or greater than, from a refractive index of the one or more 3D tissue models, may be that it goes to increase a contrast in OCT data between the one or more 3D tissue models and matrix in one more 3D tissue models for which OCT data are acquired.

[0135] According to a third aspect there is presented a system comprising:

[0136] An optical coherence tomography (OCT) imaging apparatus, and a processing unit,

[0137] wherein the processing unit is operatively connected to the optical coherence tomography (OCT) imaging apparatus and arranged to execute a method according to the second aspect.

[0138] The optical coherence tomography (OCT) imaging apparatus may comprise a light source, such as a broad-band light source, such as one or more super-luminescent diodes (SLDs), a mirror, a lens, a spectrometer comprising a diffraction grating, a detector, such as a camera, such as a line scan camera, such as a high-speed line scan camera, and a beam splitter for splitting light from the light source into two paths: The reference arm and the sample arm. The apparatus is arranged so that in the reference arm of fixed length, light is reflected by the mirror, and so that the sample arm directs light through the lens onto the sample, where it is either backscattered or reflected. Light from both arms returns and generates aninterference spectrum, detected by the spectrometer using the diffraction grating to spatially separate wavelengths. This produces a linear spectral spectrum detected by the detector.

[0139] While the above OCT imaging apparatus may represent a specific interferometer configuration, such as spectral-domain OCT (SD-OCT) in a certain interferometer configuration, it is understood that other types of setups and / or interferometer configurations are applicable, e.g., a swept-source configuration or another configuration?

[0140] A 'processing unit' may be understood as is common in the art, such as a computer arranged to operate according to stored instructions, such as a computer program, optionally dependent on external input.

[0141] According to a fourth aspect there is presented a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the computer-implemented method according to the first aspect, such as cause the system of the third aspect, to execute a method according to the second aspect. Such a computer program product may be provided on any kind of computer readable medium or through a network.

[0142] According to an alternative fourth aspect there is presented a computer program product comprising instructions to cause the system of the third aspect, to execute a method according to the second aspect. Such a computer program product may be provided on any kind of computer readable medium or through a network.

[0143] According to a fifth aspect there is presented a computer-readable medium having stored thereon the computer program of the fourth aspect.

[0144] According to an alternative fifth aspect there is presented a computer-readable medium having stored thereon the computer program of the alternative fourth aspect.

[0145] BRIEF DESCRIPTION OF DRAWNGS

[0146] The first, second, third, fourth, and fifth aspect according to the invention will now be described in more detail with regard to the accompanying figures. The figures show one way of implementing the present invention and is not to be construed as being limiting to other possible embodiments falling within the scope of the attached claim set.Fig. 1 shows a flow-chart illustrating a computer-implemented method for assessing drug efficacy,

[0147] Fig. 2 shows an overview of the mean attenuation coefficient vector from each group / drug depicted as boxplots,

[0148] Fig. 3 shows a boxplot of % of unfittable data for each group / drug;

[0149] Fig. 4 shows an overview of the sizes plotted as a boxplot comprising each group / drug; Fig. 5 shows a histogram filtered image in the frequency domain for the drug Belinostat, Fig. 6 shows a boxplot showing the variance of threshold in each group,

[0150] Fig. 7 shows a boxplot of distribution of the ratio for each group, and

[0151] Fig. 8 shows boxplots of percentage energy (%) in three spectral bands (low-, mid-, high-band) at three cisplatin concentrations (0 µM, 25 µM ,100 µM), comparing histogel vs. no histogel.

[0152] DETAILED DISCLOSURE OF THE INVENTION

[0153] Fig. 1 shows a flow-chart illustrating a computer-implemented method 100 for assessing drug efficacy, such as drug efficacy in one or more 3D tissue models, comprising

[0154] Receiving 102 an optical coherence tomography (OCT) data set, such as a 1D data set, such as a 2D image, such as a volumetric data set,

[0155] - Analysing 104 the optical coherence tomography (OCT) data set to obtain a plurality of metric values, wherein the plurality of metric values is comprising a metric value for one or more or all of the metrics within a group of metrics comprising:

[0156] i. A cut-off spatial frequency metric, wherein the value of the cut-off spatial frequency metric relates to, such as is a measurement of, a frequency separating high- and low frequency content, such as obtained in a spatial frequency analysis, and

[0157] II. a ratio metric, wherein the value of the ratio metric relates to, such as is a measurement of, a ratio between high- and low spatial frequency content, such as obtained in a spatial frequency analysis of the received OCT data,

[0158] and

[0159] Assessing 106 a drug efficacy based on the plurality of metric values.

[0160] In the following, an example of an embodiment of the invention is provided.Acquisition of the Images

[0161] The creation and preparation of the samples were done at and by the company Bioneer. Creation of the spheroids and imaging were done within one week.

[0162] The creation and preparation of the samples were done at and by the company Bioneer. Creation of the spheroids and imaging were done within one week.

[0163] • Day 1 (Tuesday): Seeding of cells. The cell suspension created contained 5000 Fibroblasts (1BR.3. G) + 25000 cancer cells (HT29). The seeding volume was of 200 pL per well in a low-adhesive 96-well plate.

[0164] • Day 3 (Friday): Bright-field imaging was done, and the 9 different drugs were added to the wells.

[0165] • Day 6 (Monday): Plate 1 was bright-field imaged, whereas plates 2 and 3 were assayed.

[0166] After imaging was done, plate 1 was prepared for OCT imaging with embedding the spheroids in Histogel™.

[0167] Plate 4 was PFA-fixated for histology.

[0168] OCT imaging was done the same day.

[0169] The 9 drugs added on Friday were: Mercaptopurine, Docetaxel, Temsirolimus, Dasatinib, Belinostat, Bosutinib, Sunitinib, Carfilzomib, and Rucaparib. A 96-well plate assembly was utilized. We had six spheroids per group. A Vehicle is our control group. This means there has been no added drug to its wells. The concentration of each added drug was 10 µM.

[0170] The OCT system used for imaging was the Telesto II - Spectral Domain OCT system from the company Thorlabs.

[0171] Segmentation methods

[0172] To analyse the various metrics of interest, it may be beneficial to segment the spheroid from the background, as the data of interest is located here. Therefore, different algorithms may be applied to segment out the spheroid. Applicable segmentation methods may include Active Contouring with Back Image Subtraction, optionally with self-drawn box, and segmentation with K-means filtering and BLOB analysis.

[0173] No single segmentation method is necessarily optimal for segmenting all the spheroids, given their varying sizes and degrees of disintegration. Therefore, in embodiments, three differentmethods will be used in the segmentation part, and the most precise segmentation will be chosen, such as subjectively chosen.

[0174] Attenuation

[0175] The attenuation coefficient, μOCT, reflects the rate at which incident light attenuates due to absorption and scattering, depending on the medium's properties. In highly scattering or absorbing tissues, μOCTis larger, indicating the tissue's structure and organization. [22, 23] It is expected that the attenuation coefficient result will indicate a change compared to the control group, although it is uncertain in which direction this change will occur, as it depends on the mechanism of action of the drug.

[0176] Absorption

[0177] Light intensity is decreased through absorption, which changes light into a different kind of energy. When the frequency of the incoming light is at or close to the energy level of the electrons in the substance, absorption takes place. In biological tissue, it is the presence of various chromophores (molecules that can absorb light), like haemoglobin, melanin, and water, that decide the absorption spectrum. [22, 24]

[0178] The main absorbing chromophore in the samples used in this thesis is water. Experiments show that heavy water exhibits low absorption at wavelengths < 1700nm. Heavy water has a similar absorption spectrum to water, thus it can be assumed that this applies to water. The system that has been used uses a wavelength of 1310 nm and therefore the absorption is negligible.

[0025]

[0179] Scattering

[0180] Scattering is the phenomenon known where light deviates from its path when it hits a particle in the sample. There are two different types of scattering: single backscattering and multiple backscattering. Single backscattering can be described as when the light travels until it reaches a molecule that will backward scatter the light. Multiple backscattering is when the light hits several molecules before it is backward scattered. It is the optical backscattering that produces the contrast that creates the OCT images. [22, 26, 13]

[0181] The sum between absorption and scattering is the attenuation coefficient μoct= μa+ μs, and can be modelled as a single exponential decay in the form of Lambert-Beer's law using the Single-Scattering Model.

[0022] Single-Scattering Model

[0182] It is the simplest and most commonly used model describing the OCT signal. It relies on the first-order Born approximation. The first-order Born approximation makes the following assumptions

[0022] :

[0183] • The incident beam will continue to propagate ahead until a single scattering event occurs, reflecting the light back toward the source. Therefore multiple scattering events will in this model be disregarded.

[0184] • The incident light beam is only attenuated by absorption and scattering described by Lambert-Beer's law.

[0185] • The wave used is a plane wave.

[0186] • Homogeneous optical scattering is present.

[0187] • There is no noise and instrumental effects.

[0188] When taking these assumptions into account the mean OCT signal can be described by Lambert-Beer's law:

[0189] {A zy> oc Ao exp (— Jocr-z), (4)

[0190] where z depicts the depth,Oct depicts the attenuation coefficient, A(z) depicts the OCT signal amplitude and Ao depict the signals amplitude at z = 0.

[0022]

[0191] One need to note here is that the scattering coefficient psand absorption coefficient pacontrol the intensity's decay rather than the field amplitude described in equation 4.

[0192] Therefore the detected signal can be described with the equation:

[0193] (I(z)) oc Io exp (- / JOCT-Z'), (5)

[0194] When applying the signal scattering model to an OCT image, a correction of sensitivity roll-off and Confocal Point Spread Function (CPSF) is needed, because they produce a depthdependent response.

[0022] Their contribution to the OCT signal can be written as:

[0195] 4(z) oc CPSFiz}' + Roll-Off zper) + Attenuation(z), (6)

[0196] where zperis the depth in percentage and z is the depth.Sensitivity Roll-off

[0197] Sensitivity roll-off describes the loss of sensitivity with depth in OCT systems due to the limitations of the spectrometer. The spectrometer has a fixed pixel size, which results in a consistent sampling rate of the interference fringes across the spectrum. At low fringe frequencies, the sensitivity is high because the system can measure many samples per period. However, as we move to higher fringe frequencies, the number of samples per period decreases. This reduced sampling frequency results in less contrast between the highs and lows of the fringes, as we are not sampling them adequately.

[0198] Insufficient spectral sampling by the spectrometer leads to lower signal-to-noise ratios at greater depths, resulting in reduced contrast in the image due to a mismatch between high-frequency data and the sampling frequency, a characteristic of the system's response. [22, 27]

[0199] The sensitivity roll-off can for the OCT amplitude be expressed as:

[0022]

[0200] sin (rzper Roll-Off (,3pef

[0201]

[0202] where the depth is described as the percentage of the imaging range,

[0203]

[0204] where ZD is the maximum imaging depth of the system, zo is the distance between the tissue boundary and the interferometer's zero-delay position.

[0205] The data we acquire have been stored in decibels, by multiplying the 10-log of the IIFT converted signal with 20. Thus, the data we acquire is in decibels. Therefore, will the equation used for the correction be translated to:

[0206] (8)

[0207]

[0208] where sine will be defined as sin(x) / x, and the depth in pixels on the A-scan divided by the overall imaging depth in pixels can be used to define zper. For our system, the overall imaging depth will be 1024 pixels, a is a free fitting parameter for off-set and s is the spectral ratio.

[0209] Measuring the sensitivity Roll-off parameters

[0210] We collect measurements of a reflector at different depths to measure the sensitivity roll-off for the video-rate (28 kHz) setting used for imaging.

[0027]

[0211] As a reflector, a white tile with a flat surface was used. When measuring the reflection, the confocal arm is kept fixed, so the focus of light is fixed at the reflector's surface.

[0212] To determine the sensitivity roll-off parameters, the length of the reference arm was adjusted by equal increments of 0.2 mm until the end of the FOV at 3.6 mm. In each step, a B-scan was obtained. These B-scans are what we use in the further analysis to determine the value of the sensitivity Roll-off parameters.

[0213] 18 B-scan were taken over the z-axis length of 1024 pixels. Ten A-scans per B-scan were averaged to enhance the accuracy of the resulting A-scan.

[0214] The maximum peak represents the amplitude in dB at a specific depth. These data points were fitted into the model shown in equation 8. An amplitude loss over depth can be observed. This is exclusively because of the sensitivity roll-off.

[0027] The spectral ratio s was

[0215] found to be 1.1019. The parameter was used for the following data processing, when correcting for the sensitivity roll-off.

[0216] Confocal Point Spread Function

[0217] When correcting for the Confocal Point Spread Function (CPSF) you correct for effects the resulting signal gets from the light's distribution. If the intensity of the light is plotted as it travels through the system, the intensity will go up at the focal plane and down afterward. If the signal is not corrected for CPSF the attenuation will be minimized and therefore the true attenuation in the sample cannot be depicted.

[0022]

[0218] The CPSF can be described as:

[0022] (9)

[0219]

[0220] where zt is the focal point relative to the tissue boundary, n is the average refractive index of the medium along the beam, and Z is the Rayleigh length of the Gaussian beam.

[0221] Also for the CPSF we need to take into consideration that we are working with data in decibels. Therefore will the correction used look like:

[0222]

[0223] where z is the depth in mm from the surface of the sample.

[0224] The attenuation coefficient a

[0225]

[0226] The attenuation coefficient is or the mean attenuation coefficient is an example of the optical attenuation coefficient metric.

[0227] To find the mean attenuation coefficient through the spheroid the following algorithm was created. It has 5 steps: Initialization, correction, inserting the mask, filtering, and finding the attenuation coefficient.

[0228] • Step 1: Initialization: The data is loaded.

[0229] • Step 2: Correction: In this step, the data is corrected for the CPSF and sensitivity rolloff. To correct for them equations 8 and 10 are used and subtracted from the data.

[0230] • Step 3: Inserting the mask: After inserting a segmentation mask, the values outside of the mask are set to NaN.

[0231] • Step 4: Filtering: Speckle must be removed using a median filter since it can cause errors when computing the attenuation coefficient, as we are working with single scattering only.

[0232] • Step 5: Finding the attenuation coefficient: To find the attenuation Lambert- Beer's law is used, see equation 5. The goal is to identify the mean attenuation coefficient in the spheroid; to do this, we will loop through each A-scan in the B-scan, determine its slope, and append it to a vector. Because we are only interested in valid data (data / =NaN) before fitting, we must eliminate the NaN values from the A-scan. When finding the slope we set some criteria that need to be fulfilled:

[0233] o When plotting the A-scan, at least two peaks are required; otherwise, there is insufficient

[0234] o data for an appropriate fit.

[0235] o If the length of the A-scan is less than 10, its slope will be set to NaN, as a satisfactory fit cannot be achieved.

[0236] If the slope is positive, its value is set to NaN, as this is a sign of disintegration. After we have found the slopes of each A-scan we remove the first and last 25 indices to remove edge artifacts. The vector containing all slopes through the spheroid gets saved for further analysis. We find the mean slope of only the valid data and times it with (-0.2). This is due to the fact that we have data in dB on the y-axis and must convert the slope to the same unit. This is the resulting mean attenuation coefficient.

[0237] age of 'unfittable' data

[0238] The Percentage of 'unfittable' data is an example of the fragmentation metric.

[0239] As mentioned in the section on " The attenuation coefficient algorithm", when the data did not fit the criterion, NaN was added to the vector containing all the slopes. This metric was intended to determine the degree of spheroid disintegration; the more unfittable data a spheroid held, the more disintegrated it was. This percentage would be a translation between what we can observe in the images and what data tells us.

[0240] To find the percentage of NaN values the Matlab function ISNAN was used. This creates a logical vector where the NaN values are set to 1 and not NaN values are set to zero.

[0241] 1 TF = isnan ( att_vec_short ); %Returns a logical vector with 1 for NaN values 2 per_att = ( sum (TF ) / length ( att_vec_short ) ) * 100;

[0242]

[0243] 3 per_vec = [per_vec; per_att]; %Saves the value for further analysis

[0244] where att_vec_short is the vector containing all attenuation coefficients through the spheroid, after removing the first and last 25 indices.

[0245] Size

[0246] The spheroid size is an example of the size metric. This metric shows if the drugs had an impact on the size of the spheroids. The masks created for each spheroid were used toindicate this, as they are logical masks. This means that the pixel value will be one where we have spheroid and the rest is set to zero. The size is then converted from pixels to microns.

[0247] To do the conversion, the axis dimensions in pixels and millimeters were retrieved through image data from one of the spheroids. The conversion dimensions are the same for all spheroids.

[0248] 1 conv = (x_im / x_pix) * ( ( z_im / z_pix) / RI );

[0249] 2

[0250]

[0251] %Finding the sum of pixels for where the spheroid is

[0252] 3 C = sum (mask, ' all ' );

[0253] 4 C = (C*conv) *1000; %Converting to microns

[0254] 5 sum_vec = [sum_vec, C ]; % Saving the sum for further analysis

[0255] analysis - such as for cut-off

[0256]

[0257] metric and ratio metric

[0258] Frequency analysis holds promise in providing a statistical metric identifying a change in metabolism, and the presence of apoptotic and proliferating cells. To look at the different frequencies we use a 2D Fourier transform. This will make the image go from the spatial domain where we look at pixel intensity to the frequency domain where we look at frequencies. In the frequency domain, you can get a spectral image.

[0259] Fourier Transform

[0260] The 2D Fourier transform is an extension of the ID Fourier Transform, that is used in signal processing to extract and separate a signal to a sum of complex exponentials (complex oscillations). When using the 2D Fourier transform in image processing, the principle is the same, but because the pixels have real intensities, their complex oscillation will come in pairs.

[0261] Discrete Fourier transform

[0262] Af-1 A'-l

[0263]

[0264] m=0 n=0

[0265] where m and n are spatial coordinates; p and q are frequency coordinates.

[0266] The Discrete Fourier transform is a complex function, as it sums complex exponential functions at different frequencies, to represent the image in the frequency domain. Thisresults in a complex image, and we therefore need to display the amplitude and phase separately.

[0267] When displaying the spectral image of the amplitudes, with the function FFTSHIFT, the low frequencies will be located in the middle of the image and the high frequencies at the border of the image.

[0268] In time-domain imaging, a smooth surface is indicated by low-frequency components, reflecting gradual intensity changes between adjacent pixels, thus demonstrating slower transitions in pixel intensity values. The high frequencies represent a rapid change between two parts in the image, e.g. going from a dark area to a brighter area. [28, 29, 30]

[0269] This means the higher frequencies represent the edges in the image and the lower frequencies represent the information, thus the image will appear blurred when removing information about the low frequencies.

[0270] Finding the two groups

[0271] During the initial exploration of the frequency data, the histogram was plotted. This was done to get an overview of the frequency distribution of the spheroids and to see if there was a pattern. The histogram was plotted in log scale and only the real values were visualized. During the evaluation of the histogram, it was discovered that they had two humps. A hypothesis was set up that the two humps represented respectively the proliferating cells and the apoptotic cells. The histogram was fit as the combination of two Gaussian curves. Where and if the two Gaussians had an intersection the frequency value of this intersection was found and used as a threshold to discriminate between low and high frequencies. The frequencies to the left of the threshold are considered low frequencies, and the frequencies to the right of the threshold are considered high frequencies. If no intersection could be found the threshold would have the value NaN.

[0272] After the threshold was found the percentage of low and high frequencies was calculated and the ratio between the high and low frequencies was calculated (%high %low ).

[0273] Whether the low frequencies contained the apoptotic cells or the proliferating cells is unsure but a hypothesis was created that a frequency analysis could be a metric to parameterize the presence and distribution of apoptotic and proliferating cells.

[0274] The threshold is an example of the cut-off spatial frequency metric.The ratio between the high and low frequencies is an example of the ratio metric.

[0275] Fig. 5 shows a histogram filtered image in the frequency domain for a segmented spheroid (labelled Belinostat0067) treated with the drug Belinostat, wherein two Gaussian functions, Gaussianl (as indicated with a dashed line) and Gaussian2 (as indicated with a dotted line), have been fitted to the histogram data, and wherein the intersection between the Gaussian functions is identified as the Threshold (as indicated with the full-drawn, vertical line between the " Log Frequency"-axis values 18 and 20).

[0276] Frequency analysis - for band metric

[0277] Fourier analysis - for band metric

[0278] Structural features in OCT images can be characterized in spatial terms and frequency domain using the two-dimensional Discrete Fourier Transform (2D DFT). This mathematical technique decomposes an image into a weighted sum of sinusoidal basis functions, each representing a specific spatial frequency

[0035] .

[0279] The 2D DFT of an image f m, n) of size M x N is defined as:

[0280]

[0281] m=0 n=0

[0282] Here, m and n are the special coordinates in the images, u and v are the frequency coordinates, Mand N are the number of rows and columns in the image, and j is the imaginary unit.

[0283] The output F(u, v) is a complex-valued matrix containing both magnitude, which is how strongly a given frequency contributes, and phase information, which is the special alignment of those features

[0035] .

[0284] To quantify frequency content, the power spectral density (PSD) is computed as the squared magnitude of the Fourier coefficients:

[0285] P{u, v) = |F(u, v) |2This measures how much energy is present at each spatial frequency. The total energy in the frequency domain is defined as the sum of the squared magnitudes of each Fourier coefficients:

[0286] M-l N-l

[0287] E = |F(iz, v)|2

[0288]

[0289] m=0 n=0

[0290] This yields a frequency-based decomposition of the image, allowing quantification of how much each spatial scale contributes to the overall structural composition. In the context of band metric(s), energy contributions from predefined frequency bands are calculated as partial sums of the total energy and expressed as percentages, enabling band-wise structural characterization of spheroids.

[0291] To interpret this information, the frequency domain is divided into three radial bands: low-, mid-, and high-frequencies. Low frequencies, concentrated near the center of the frequency spectrum, correspond to smooth, large-scale structures. High frequencies are found toward the periphery and represent rapid intensity changes associated with edges and fine details.

[0292] The inverse 2D DFT is used to transform filtered frequency bands back into the spatial domain, enabling visualization and localization of structures associated with low-, mid-, and high-frequency content:

[0293] M — 1 N ~~ 1

[0294] 1n

[0295] f(m, n) — y y F(u, v)eJ

[0296]

[0297] u= 0 v=0

[0298] Frequency analysis is applied as a quantitative method to assess how the spatial organization of spheroids responds to cisplatin exposure. A shift toward higher frequencies may indicate increased heterogeneity or microstructural degradation, reflecting treatment-induced architectural changes.

[0299] Frequency content - for band metric

[0300] A MATLAB pipeline was developed to characterize structural differences in OCT images of spheroids. The workflow implements frequency-based analysis, following the approach described in the preceding section (" Fourier analysis - for band metric"').1. Data initialization and loading:

[0301] The 3D OCT volume was loaded, and physical voxel dimensions were extracted from the file.

[0302] 2. Extract axial slice:

[0303] A single 2D axial slice was extracted at the middle of the spheroid.

[0304] 3. Background subtraction:

[0305] The background level was estimated from the bottom-right corner of the image. The mean intensity value was subtracted from the entire image. Negative values were clipped to zero, and the image was normalized.

[0306] 4. Spheroid segmentation:

[0307] Here, the Otsu's thresholding method was used, which finds an optimal intensity cutoff separating foreground from background. Small noise regions (<500 pixels) were removed, and internal holes were filled. The largest connected component was retained, assuming it corresponds to the spheroid. Morphological operations (imopen and imclose) were applied to smooth the mask boundary, and a slight dilation ensured that peripheral edge features were included in the analysis.

[0308] 5. Frequency domain transformation:

[0309] The 2D DFT was applied to the masked image using "fft2," followed by "fftshift" to center the Direct Current (DC) component.

[0310] 6. DC component and noise removal:

[0311] A circular mask removed the DC-spike within a radius of 0.003 cycles / pixel. A low-pass filter removed frequencies above 0.1 cycles / pixel to suppress noise.

[0312] 7. Band separation and inverse FFT:

[0313] The frequency domain was divided into three radial bands using two cutoff frequencies, rl = 0.01 and r2 = 0.05, where r denotes the radial spatial frequency (in cycles / pixel):• Low-band: r ≤ 0.01

[0314] • Mid-band: 0.01 < r < 0.05

[0315] • High-band: r > 0.05

[0316] Then the inverse DFT transformed each band back to the spatial domain to localize its structural features.

[0317] 8. Visualization of images (Pure-band map and RGB composite):

[0318] A winner-takes-all map assigned each pixel to its dominant band (blue = low-band, green = mid-band, red = high-band), while an RGB composite visualized relative contributions per pixel.

[0319] 9. Computation of energy and pixel count:

[0320] Absolute energy per band was computed, and percentages were used to normalize spheroid size and OCT signal intensity, enabling consistent comparison across groups. Pixel counts were extracted for each frequency band to support size control.

[0321] The study comprised 60 spheroids embedded in histogel and 60 non-embedded spheroids, said spheroids comprising colorectal cancer cells (HT29) and fibroblasts (1BR3G), and wherein 120 OCT images were acquired with a using the Thorlabs Telesto II Spectral-Domain OCT system and analyzed.

[0322] Results

[0323] Attenuation

[0324] Fig. 2 shows an overview of the mean attenuation coefficient vector from each group / drug depicted as boxplots.

[0325] Another, similar study (comprising 60 spheroids embedded in histogel and 60 non-embedded spheroids, said spheroids comprising colorectal cancer cells (HT29) and fibroblasts (1BR3G), and wherein 120 OCT images were acquired with a using the Thorlabs Telesto II Spectral-Domain OCT system and analyzed), confirmed via one-way ANOVA analysis a significant overall dose effect (with cisplatin concentrations of 0 pM, 25 pM and 100 pM) on attenuation coefficient, / JOCT, for each of hydrogel (histogel) embedded spheroids and non-hydrogel (non-histogel) embedded spheroids.

[0326] Percentage of unfittable dataFig. 3 shows a boxplot of % of unfittable data for each group / drug.

[0327] Size

[0328] Fig. 4 shows an overview of the sizes plotted as a boxplot comprising each group / drug.

[0329] Another, similar study (comprising 60 spheroids embedded in histogel and 60 non-embedded spheroids, said spheroids comprising colorectal cancer cells (HT29) and fibroblasts (1BR3G), and wherein 120 OCT images were acquired with a using the Thorlabs Telesto II Spectral-Domain OCT system and analyzed) indicated via three-way ANOVA analysis a highly significant effect on spheroid size of varying the cisplatin concentration (with cisplatin concentrations of 0 pM, 25 pM and 100 pM) indicating that increasing drug dose strongly influenced the spheroid size. However, changes in volume depended on whether a hydrogel (histogel) matrix surrounded the spheroids.

[0330] Frequency content

[0331] Fig. 6 shows a boxplot showing the variance of threshold (such as value of cut-off spatial frequency metric) in each group.

[0332] Fig. 7 shows a boxplot of distribution of the ratio (such as the value of ratio metric) for each group.

[0333] Frequency content - band metric

[0334] As explained in the preceding section " Frequency content - for band metric", two radial cutoffs (rl and r2) were chosen to divide the three frequency-bands (low-, mid-, high-band) so that the three filtered outputs exhibited clear frequency-band separation.

[0335] The spatial frequency r (in cycles / pixel) was defined as the inverse of the spatial wavelength λ (in pixel), such that r = 1 / λ. The selected cutoff pair rl = 1 / 100 = 0.01 cycles / pixel and r2 = 1 / 20 = 0.05 cycles / pixel produced outputs in which low-, mid-, and high-frequency bands were clearly separated. This made these cutoffs ideal for further analysis and aligned them with the relevant spatial scales, corresponding to whole spheroids (~100 pm) and individual cells (~20 jum).

[0336] The defined frequency bands were then used to determine an appropriate cutoff frequency for DC component removal (Section " Frequency content - for band metric"') to prevent thezero-frequency spike from overwhelming the spheroid's low-frequency content. A circular mask of radii rDC = 0.001, 0.003, and 0.006 cycles / pixel was applied in the 2D fourier domain.

[0337] With rDC = 0.001 cycles / pixel, only the central spike was filtered, causing all pixels to appear as low-frequency in the pure-band map. In contrast, rDC = 0.006 cycles / pixel removed the DC spike as well as potentially signal-bearing low-frequency coefficients, leading to a midfrequency ring. At rDC = 0.003 cycles / pixel, the zero-frequency term was eliminated while preserving the nearest actual peak at |fx| = 0.004. This cutoff retained relevant large-scale structural information and was therefore applied when generating frequency-band images to ensure optimal results.

[0338] With these parameters established, the frequency algorithm was applied to all spheroids enabling quantification of how histogel and cisplatin influenced spectral energy across all samples.

[0339] A three-way ANOVA on absolute spectral energy was used to evaluate the main effects of embedding (histogel vs. no histogel), cisplatin concentration (0 μM, 25 μM, 100 μM), and frequency band (low-, mid-, high-band), as well as all two- and three-way interactions.

[0340] The following section examines the percentage energy distributed across in the low-, mid-, and high-bands at 0 μM, 25 μM, and 100 μM cisplatin concentrations (Fig. 8). When comparing histogel embedded and non-embedded spheroids, the boxplots revealed how increasing cisplatin dose might systematically redistribute spectral energy across the three bands.

[0341] Fig. 8 shows boxplots of percentage energy (%) in three spectral bands (low-, mid-, high-band) at three cisplatin concentrations (0 µM, 25 µM ,100 µM), comparing histogel vs. no histogel.

[0342] In the low-frequency band, histogel-embedded spheroids exhibited a U-shaped response to increasing cisplatin concentrations. The median fell from 46 % at 0 μM to 38 % at 25 μM before rising to 52 % at 100 μM. Non-embedded spheroids followed the same pattern (59 % 54 % 58 %), but with substantially less variability.

[0343] In the mid-frequency band, histogel-embedded spheroids reached a peak median of about 42 % at 25 μM, and then declined to 34 % at 100 μM, whereas non-embedded spheroids showed a similar but less pronounced peak (29 % 33 % 30 %).In the high-frequency band, histogel-embedded spheroids' medians steadily decreased (22 % 20 % 15 %), while non-embedded spheroids' medians were almost unchanged (12 % 12.5 % 11 %).

[0344] To investigate this further, a two-way ANOVA was conducted to evaluate the Condition x Band interaction, testing whether cisplatin dose differentially affected percent spectral energy across low-, mid-, and high-bands.

[0345] The main effect of cisplatin dose on percentage spectral energy was highly significant in both unlabeled and mCherry-labeled spheroids (p < 0.01), indicating that increasing drug concentration systematically redistributed energy across the low-, mid-, and high-bands.

[0346] To pinpoint dose increments that made these shifts, Welch's t-tests were performed within each band, comparing 0 μM to 25 μM, 0 μM to 100 μM, and 25 μM to 100 μM.

[0347] In the low-frequency band for both histogel-embedded and non-embedded spheroids, the percentage energy shifts were significant between 0 μM → 25 μM and 25 μM → 100 μM at a 1%-significant level. Notably, nonembedded spheroids exhibited a decrease in low-band energy across 0 μM → 100 μM, whereas histogel-embedded spheroids showed an increase.

[0348] In the mid-frequency band, all three comparisons reach significance (0 μM → 25 μM, 0 μM → 100 μM, 25 μM →100 μM), regardless of embedding, showing a consistent redistribution.

[0349] In the high-frequency band in histogel-embedded spheroids, significant changes occurred at higher doses: 25 μM → 100 μM and 0 μM →100 μM. And for non-embedded spheroids, there was a 1%-significant level at 25 μM → 100 μM.

[0350] It has also been demonstrated from comparison of histology and OCT spatial-frequency maps for control vs. cisplatin-treated spheroids (although not the same spheroids) embedded in histogel that OCT frequency maps generally corresponded well to histological features, providing a robust framework for interpreting the spectral analyses.

[0351] Discussion - band metric

[0352] Cisplatin induced consistent, dose-dependent spectral shifts in both embedded and nonembedded spheroids. Mid-frequency energy peaked at 25 uM, suggesting the emergence of apoptotic microstructures.In contrast, low-frequency energy followed a U-shaped pattern, rebounding at 100 μM which could be in line with large-scale structural breakdown during late-stage necrosis.

[0353] High-frequency energy declined significantly at higher doses. The dose-dependent changes suggested that mid-band energy may be a biomarker of cisplatin-related microstructural alterations.

[0354] Thus, increasing cisplatin doses produced dose-dependent spectral shifts regardless of embedding condition. Low-frequency energy declined from 0 μM, then partially rebounded at 100 μM, mid-frequency energy peaked at 25 μM, and high-frequency energy fell at 100 μM, highlighting the spectrum's sensitivity to cisplatin.

[0355] While histogel embedding amplified the variability and magnitude of these shifts, the overall response pattern remained robust and consistent across mounting conditions.

[0356] These dose-dependent spectral changes highlighted the sensitivity of the OCT spectrum to cisplatin-induced structural dynamics.

[0357] Although the present invention has been described in connection with the specified embodiments, it should not be construed as being in any way limited to the presented examples. The scope of the present invention is set out by the accompanying claim set. In the context of the claims, the terms "comprising" or "comprises" do not exclude other possible elements or steps. Also, the mentioning of references such as "a" or "an" etc. should not be construed as excluding a plurality. The use of reference signs in the claims with respect to elements indicated in the figures shall also not be construed as limiting the scope of the invention. Furthermore, individual features mentioned in different claims, may possibly be advantageously combined, and the mentioning of these features in different claims does not exclude that a combination of features is not possible and advantageous.

[0358] REFERENCES

[0359]

[0013] James G. Fujimoto and Wolfgang Drexler. " Introduction to OCT". In: Springer International Publishing, Jan. 2015, pp. 3-64. ISBN: 9783319064192. DOI: 10 . 1007 / 978 - 3 - 319 - 06419-2_1.

[0360]

[0022] Peijun Gong et al. " Parametric imaging of attenuation by optical coherence tomography: review of models, methods, and clinical translation". In: Journal of Biomedical Optics 25 (04 Apr. 2020), p. 1. ISSN: 15602281. DOI: 10.1117 / 1.jbo.25.4.040901.

[0023] Shuang Chang and Audrey K. Bowden. " Review of methods and applications of attenuation coefficient measurements with optical coherence tomography". In: Journal of Biomedical Optics 24 (9 Sept. 2019), p. 1. ISSN: 15602281. DOI: 10 . 1117 / 1 . JBO . 24 . 9 . 090901. URL: https: / / pmc.ncbi.nlm.nih.gov / articles / PMC6997582 / .

[0361]

[0024] Andor. What Happens When Light is Absorbed, online, unknown unknown. URL: https: / / andor.oxinst.com / learning / view / article / absorption-of-light.

[0362]

[0025] V. M. Kodach et al. " Quantitative comparison of the OCT imaging depth at 1300 nm and 1600 nm". In: Biomedical Optics Express 1 (1 Aug. 2010), p. 176. ISSN: 2156-7085. DOI: 10.1364 / BOE. 1. 000176. URL: https: / / pu bmed.ncbi.nlm.nih.gov / 21258456 / .

[0363]

[0026] Vania Bastos Silva et al. " Signal-carrying speckle in Optical Coherence Tomography: a methodological review on biomedical applications". In: Journal of Biomedical Optics 27 (03 Aug. 2021). ISSN: 15602281. DOI: 10 . 1117 / 1 . jbo . 27 . 3 . 030901. URL: https: / / arxiv.org / abs / 2108.13109v1.

[0364]

[0027] Morgane Zimmer. Investigation of single and multiple scattering effects in OCT. Master Thesis. 2021.

[0365]

[0028] Unknown. Fourier transform. Last accessed on May 16, 2024. unknown.

[0366] URL: https: / / vincmazet.github.io / bip / filtering / fou rier.html.

[0367]

[0029] UniHeidelberg. 3.4 2D-DFT: Application to Images — Image Analysis Class 2013.

[0368] Youtube. Accessed last on May 16, 2024. May 2013. URL: https: / / www.youtube.com / watch?v=-3QRMlzlP9E.

[0369]

[0030] First Principles of Computer Vision - Columbia University. Image Filtering in Frequency Domain — Image Processing II. Youtube. Accessed last on May 16, 2024. Mar. 2021.

[0370] URL:https: / / www.voutube.com / watch?v=OOu5KP3Gvx08dist=TLPQMDQwMzIwMiRpWZOML:

[0371]

[0372] = 3.

[0373]

[0035] Github. Fourier transform — Basics of Image Processing. Vincent Mazet (Universite de Strasbourg), https: / / vincmazet.github.io / bip / filtering / fourier.html.CLAUSES

[0374] There is furthermore presented a computer-implemented method, a system, a computer program product, and a computer-readable medium according to the clauses 1-15 below, which clauses may be combined with any of the preceding embodiments and / or any of the appended claims:

[0375] 1. A computer-implemented method (100) for assessing drug efficacy, such as drug efficacy in one or more 3D tissue models, comprising

[0376] Receiving (102) an optical coherence tomography data set, such as a 1D data set, such as a 2D image such as a volumetric data set,

[0377] - Analysing (104) the optical coherence tomography data set to obtain a plurality of metric values, wherein the plurality of metric values is comprising a metric value for one or more or all of the metrics within a group of metrics comprising:

[0378] i. A cut-off spatial frequency metric, wherein the value of the cut-off spatial frequency metric relates to, such as is a measurement of, a frequency separating high- and low frequency content, such as obtained in a spatial frequency analysis of the received optical coherence tomography data set, and

[0379] II. a ratio metric, wherein the value of the ratio metric relates to, such as is a measurement of, a ratio between high- and low spatial frequency content, such as obtained in a spatial frequency analysis of the received optical coherence tomography data set, and

[0380] - Assessing (106) a drug efficacy based on the plurality of metric values.

[0381] 2. The computer-implemented method (100) according to clause 1, wherein the group of metrics furthermore comprises one or more or all of:

[0382] - A size metric, such as wherein the value of the size metric relates to, such as is a measurement of, one or more 3D tissue model sizes, such as one or more 3D tissue model volumes,

[0383] - A fragmentation metric, such as wherein the value of the fragmentation metric relates to, such as is a measurement of, a fragmentation of one or more 3D tissue models, and / or

[0384] an optical attenuation coefficient metric.3. The computer-implemented method (100) according to any of the preceding clauses, wherein the metrics for which the plurality of metric values is obtained enables obtaining a unique representation of a drug efficacy.

[0385] 4. A computer-implemented method (100) according to any of the preceding clauses, wherein

[0386] - Assessing a drug efficacy based on the plurality of metric values

[0387] is based on a predetermined correlation between predetermined sets of metric values and corresponding drug efficacies.

[0388] 5. A computer-implemented method (100) according to any of the preceding clauses, wherein the optical coherence tomography data set is a volumetric data set.

[0389] 6. A computer-implemented method (100) according to any of the preceding clauses, wherein the plurality of metric values is comprising a metric value for both of the metrics within a group of metrics comprising:

[0390] the cut-off spatial frequency metric, and

[0391] the ratio metric.

[0392] 7. A computer-implemented method (100) according to clause 2, and optionally additionally according to any of clauses 3-6, wherein the plurality of metric values is comprising a metric value for all of the metrics within a group of metrics comprising:

[0393] the cut-off spatial frequency metric,

[0394] the ratio metric,

[0395] the size metric, and

[0396] the optical attenuation coefficient metric,

[0397] and wherein the plurality of metric values is optionally further comprising

[0398] the fragmentation metric.

[0399] 8. A computer-implemented method (100) according to any of the preceding clauses, comprising:

[0400] Segmentation of the optical coherence tomography data set.

[0401] 9. A method for assessing drug efficacy in one or more 3D tissue models, said method comprising:

[0402] Providing the one more 3D tissue models,

[0403] - Acquiring an optical coherence tomography data set, such as a 1D data, such as a 2D image, such as a volumetric data set, of the one or more 3D tissue models, and- Assessing drug efficacy in the one or more 3D tissue models according to the computer-implemented method (100) according to any of the preceding clauses based on the optical coherence tomography data set being received by the computer-implemented method.

[0404] 10. A method according to clause 9, wherein

[0405] acquiring the optical coherence tomography data set

[0406] comprises

[0407] acquiring a plurality of time-resolved optical coherence tomography data sets, such as a plurality of time-resolved optical coherence tomography images, such as a plurality of time-resolved volumetric images, of the one or more 3D tissue models.

[0408] 11. A method according to any of clauses 9-10, wherein said one or more 3D tissue models comprise spheroids.

[0409] 12. A method according to any of clauses 9-11, wherein said one or more 3D tissue models comprises a hydrogel with a refractive index differing, such as being smaller than or greater than, from a refractive index of 1.33 by at least 1 %, such as at least 2 %, such as at least 5 %, such as at least 10 %, such as at least 20 %, such as at least 25 %, such as at least 50 %, such as at least 100 %.

[0410] 13. A system comprising:

[0411] - An optical coherence tomography imaging apparatus, and

[0412] a processing unit,

[0413] wherein the processing unit is operatively connected to the optical coherence tomography imaging apparatus and arranged to execute a method according to any of clauses 9-12.

[0414] 14. A computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the computer-implemented method (100) according to any of clauses 1-8, such as cause the system of clause 13, to execute a method according to any of clauses 9-12.

[0415] 15. A computer-readable medium having stored thereon the computer program of clause 14.

Claims

Claims1. A computer-implemented method (100) for assessing drug efficacy, such as drug efficacy in one or more 3D tissue models, comprisingReceiving (102) an optical coherence tomography data set, such as a 1D data set, such as a 2D image, such as a volumetric data set,- Analysing (104) the optical coherence tomography data set to obtain a plurality of metric values, wherein the plurality of metric values is comprising a metric value for one or more or all of the metrics within a group of metrics comprising:i. A cut-off spatial frequency metric, wherein the value of the cut-off spatial frequency metric relates to, such as is a measurement of, a frequency separating high- and low frequency content, such as obtained in a spatial frequency analysis of the received optical coherence tomography data set, andII. a ratio metric, wherein the value of the ratio metric relates to, such as is a measurement of, a ratio between high- and low spatial frequency content, such as obtained in a spatial frequency analysis of the received optical coherence tomography data set, and- Assessing (106) a drug efficacy based on the plurality of metric values.

2. The computer-implemented method (100) according to claim 1, wherein the group of metrics furthermore comprises one or more or all of:- A size metric, such as wherein the value of the size metric relates to, such as is a measurement of, one or more 3D tissue model sizes, such as one or more 3D tissue model volumes,- A fragmentation metric, such as wherein the value of the fragmentation metric relates to, such as is a measurement of, a fragmentation of one or more 3D tissue models, and / oran optical attenuation coefficient metric.

3. The computer-implemented method (100) according to any of the preceding claims, wherein the metrics for which the plurality of metric values is obtained enables obtaining a unique representation of a drug efficacy.

4. A computer-implemented method (100) according to any of the preceding claims, wherein- Assessing a drug efficacy based on the plurality of metric valuesis based on a predetermined correlation between predetermined sets of metric values and corresponding drug efficacies.

5. A computer-implemented method (100) according to any of the preceding claims, wherein the optical coherence tomography data set is a volumetric data set.

6. A computer-implemented method (100) according to any of the preceding claims, wherein the plurality of metric values is comprising a metric value for both of the metrics within a group of metrics comprising:the cut-off spatial frequency metric, andthe ratio metric.

7. A computer-implemented method (100) according to claim 2, and optionally additionally according to any of claims 3-6, wherein the plurality of metric values is comprising a metric value for all of the metrics within a group of metrics comprising:the cut-off spatial frequency metric,the ratio metric,the size metric, andthe optical attenuation coefficient metric,and wherein the plurality of metric values is optionally further comprisingthe fragmentation metric.

8. A computer-implemented method (100) according to any of the preceding claims, comprising:Segmentation of the optical coherence tomography data set.

9. The computer-implemented method (100) according to any of the preceding claims, wherein the plurality of metric values is furthermore comprising a metric value for one or more or all of the metrics within a group of metrics comprising:- A size metric, such as wherein the value of the size metric relates to, such as is a measurement of, one or more 3D tissue model sizes, such as one or more 3D tissue model volumes,- A fragmentation metric, such as wherein the value of the fragmentation metric relates to, such as is a measurement of, a fragmentation of one or more 3D tissue models, and / oran optical attenuation coefficient metric.

10. A computer-implemented method (100) according to any of the preceding claims, comprising:- Analysing (104) the optical coherence tomography data set to obtain the metric value for the cut-off spatial frequency metric.

11. A computer-implemented method (100) according to any of the preceding claims, comprising:Obtaining the metric value for the cut-off spatial frequency metric in a spatial frequency analysis of the received optical coherence tomography data set.

12. A computer-implemented method (100) according to any of the preceding claims, comprising analysing the optical coherence tomography data set to obtain the cut-off spatial frequency metric, comprising identifying high- and low frequency content, such as of the optical coherence tomography data set, including fitting or regression.

13. A computer-implemented method (100) according to any of the preceding claims, wherein the cut-off spatial frequency metric corresponds to a frequency value, such as the cut-off spatial frequency, at a point of intersection between two Gaussian functions or distributions fitted to a spectral representation, such as a frequency histogram, such as the histogram being fitted as the combination of two Gaussian curves and where the cut-off spatial frequency metric corresponds to the frequency value of the intersection between the two Gaussian curves, of the optical coherence tomography data set, optionally wherein the logarithm of the frequency data have replaced or represent the frequency data.

14. A computer-implemented method (100) according to any of the preceding claims, comprising:- Analysing (104) the optical coherence tomography data set to obtain the metric value for the ratio metric.

15. A computer-implemented method (100) according to any of the preceding claims, comprising:Obtaining the metric value for the ratio metric in a spatial frequency analysis of the received optical coherence tomography data set.

16. A computer-implemented method (100) according to any of the preceding claims, wherein:- The value of the ratio metric relates to a ratio between high- and low spatial frequency content given by areas under a curve in a frequency histogram on either side of a cut-off spatial frequency separating high- and low-frequencycontent, wherein the cut-off spatial frequency is the cut-off spatial frequency metric value.

17. A computer-implemented method (100) according to any of the preceding claims, wherein:- The ratio metric is based on one or more predetermined spatial frequencies.

18. The computer-implemented method (100) according to any of the preceding claims, wherein the plurality of metric values is furthermore comprising a metric value for:- A band metric, wherein the value of the band metric relates to, such as is a measurement of, a ratio between spatial frequency content within a spatial frequency band and spatial frequency content outside the spatial frequency band, such as obtained in a spatial frequency analysis of the received optical coherence tomography data set,such as wherein the band of the band metric is predetermined, such as wherein the upper and lower threshold frequency limits of the band of the band metric are predetermined.

19. The computer-implemented method (100) according to any of the preceding claims, wherein the plurality of metric values is furthermore comprising metric values for a plurality, such as 2 or 3 or 4 or 5 or 10 or 20 or 100 or more, of:- A band metric, wherein the value of the band metric relates to, such as is a measurement of, a ratio between spatial frequency content within a spatial frequency band and spatial frequency content outside the spatial frequency band, such as obtained in a spatial frequency analysis of the received optical coherence tomography data set,such as wherein the bands of the band metrics are adjoining and non-overlapping.

20. A computer-implemented method (100) according to any of the preceding claims for assessing drug efficacy in one or more 3D tissue models, such as one or more 3D tissue models comprising cellular units, such as cellular units in the form of one or more tumour spheroids.

21. A method for assessing drug efficacy in one or more 3D tissue models, said method comprising:Providing the one more 3D tissue models,- Acquiring an optical coherence tomography data set, such as a 1D data, such as a 2D image, such as a volumetric data set, of the one or more 3D tissue models, and- Assessing drug efficacy in the one or more 3D tissue models according to the computer-implemented method (100) according to any of the preceding claims based on the optical coherence tomography data set being received by the computer-implemented method.

22. A method according to claim 21, whereinacquiring the optical coherence tomography data setcomprisesacquiring a plurality of time-resolved optical coherence tomography data sets, such as a plurality of time-resolved optical coherence tomography images, such as a plurality of time-resolved volumetric images, of the one or more 3D tissue models.

23. A method according to any of claims 21-22, wherein said one or more 3D tissue models comprise spheroids, such as tumour spheroids.

24. A method according to any of claims 21-23, wherein said one or more 3D tissue models comprises a hydrogel with a refractive index differing, such as being smaller than or greater than, from a refractive index of 1.33 by at least 1 %, such as at least 2 %, such as at least 5 %, such as at least 10 %, such as at least 20 %, such as at least 25 %, such as at least 50 %, such as at least 100 %.

25. A method according to any of claims 21-24, wherein said one or more 3D tissue models comprises a hydrogel with a refractive index differing, such as being smaller than or greater than, from a refractive index of any value within 1.35-1.39 by at least 1 %, such as at least 2 %, such as at least 5 %, such as at least 10 %, such as at least 20 %, such as at least 25 %, such as at least 50 %, such as at least 100 %.

26. A method according to any of claims 21-25, wherein said one or more 3D tissue models comprises a hydrogel with a refractive index differing, such as being smaller than or greater than, from a refractive index of the one or more 3D tissue models by at least 1 %, such as at least 2 %, such as at least 5 %, such as at least 10 %, such as at least 20 %, such as at least 25 %, such as at least 50 %, such as at least 100 %.

27. A system comprising:- An optical coherence tomography imaging apparatus, anda processing unit,wherein the processing unit is operatively connected to the optical coherence tomography imaging apparatus and arranged to execute a method according to any of claims 21-26.

28. A computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the computer-implemented method (100) according to any of claims 1-20, such as cause the system of claim 27, to execute a method according to any of claims 21-26.

29. A computer-readable medium having stored thereon the computer program of claim 28.

Images