Method for determining the presence of at least one datum representative of malignant cell proliferation within an image, corresponding device and program

Abstract

The disclosure relates to a method for detecting the presence of at least one datum representative of malignant cell proliferation within a histopathology image, the method being implemented within an electronic processing device. Such a method comprises: - dividing up (S01) an image representative of a tissue section, delivering a set of tiles of predetermined size, each tile having a position within the image representative of the tissue section; - determining (S02), for each tile of the set of tiles, a datum representative of a number of malignant cells and a datum representative of a total number of cells, these data being inserted into a first matrix Mp and into a second matrix Mt, respectively, each element of which is associated with one tile position within the image representative of the tissue section; - detecting (S03) the presence of malignant cell proliferation within the histopathology image, this involving computing data clusters of the first matrix Mp and of the second matrix Mt with a parallel-computing algorithm.

Description

[0001] Description

[0002] TITLE: Method for determining the presence of at least one data item representative of malignant cell proliferation within an image, device and corresponding program.

[0003] Domain

[0004] The disclosure relates to the field of quantifying malignant cell proliferation. The disclosure relates more particularly to a method of determining a degree of malignant cell proliferation within representative images of biolabeled tissues taken from one or more patients.

[0005] Prior art

[0006] Accurate quantification of nuclear biomarkers plays a critical role in the therapeutic decision-making process in oncology. This approach relies on the detection and counting of specifically labeled cells using an antigen-antibody reaction applied to tissue samples taken from patients. Once this technique is applied, tumor cells become visually discernible, allowing medical pathologists to perform an accurate count. For example, for the biomarker "KI67," tumor cells are stained in two variations: "brown" for positive results (cancer cells) and "blue" for negative results. By combining these two pieces of information, the pathologist can calculate a proliferation index, a crucial measure for assessing cellular activity.It is recalled that Ki67 is a protein associated with cell proliferation, and its immunostaining is commonly used in pathology to assess the rate of cell proliferation.

[0007] This proliferation index, for example based on the biomarker KI67, reflects the percentage of positively stained tumor cells compared to the total number of stained cells in specific areas of the sample, commonly called "hotspots". These "hotspots" correspond to areas of the image in which a significant number of cells are present, generally defined by a threshold (e.g., greater than 500 cells). This approach ensures that the analyzed regions are representative of the cellular activity in the tumor, avoiding potential biases related to the selection of less characteristic areas.

[0008] The proliferation index obtained from this quantification allows oncologists to better understand the aggressive nature of the tumor and make more informed treatment decisions. Highly proliferative tumors are often associated with faster tumor growth, which can influence the decision to use more aggressive treatments such as chemotherapy. On the other hand, tumors with a low proliferation index could potentially respond to more targeted treatments.

[0009] In order to calculate the proliferation index, especially of the KI67 biomarker, more accurately and efficiently, modern approaches based on artificial intelligence and machine learning have been developed. These approaches are essentially based on the global detection of areas of interest in labeled histopathological images (e.g., KI67). In particular, deep learning methods are used to identify potential proliferation areas in labeled histopathological images. Thus, the process of training a convolutional neural network (CNN) to detect hotspots in tissue images typically includes the following steps:

[0010] Data acquisition and preprocessing: Labeled histopathological images are acquired and manually annotated by practitioners;

[0011] CNN training: CNN models are trained on the annotated images. The training phase may involve using mutual information from different image types (e.g., H&E and Ki67) of the entire slide to identify hotspots of marker expression (e.g., KI67);

[0012] Identification of regions of interest (ROI): The trained CNN(s) highlight regions with a high density of positively labeled cells, which are potential hotspots, which helps identify areas of interest for further analysis and quantification.

[0013] Calculation of proliferation index: After identifying hotspots, other CNNs are used to segment tumor cells and calculate the proliferation index in these hotspots.

[0014] In these methods, deep neural networks are supposed to: highlight regions with a high density of relevant cells; and eliminate any coloring artifacts that could mislead the segmentation phase. The problem with this approach lies mainly in the large amount of computation required to identify global areas and generally on the error rates that can be due to weak learning or too small a training sample.

[0015] This technique aims to improve the situation.

[0016] Summary of the invention

[0017] Thus the disclosure relates to a method for detecting the presence of at least one data item representative of malignant cell proliferation within a histopathological image, said method being implemented within an electronic processing device.

[0018] This method comprises at least one iteration of the following steps: cutting the image representative of the tissue section, delivering a set of tiles of predetermined size, each tile having a position within the image representative of the tissue section; determining, for each tile of the set of tiles, data representative of a number of malignant cells and data representative of a total number of cells, these data being inserted respectively into a first matrix Mp and into a second matrix Mt, each element of which is associated with a tile position within the image representative of the tissue section; detecting the presence of malignant cell proliferation within said histopathological image, said detection implementing calculations of groupings of data from the first matrix Mp and the second matrix Mt, according to an algorithm for distributing the calculations in parallel.

[0019] Thus, unlike existing methods, which use artificial intelligence to directly determine proliferation zones, the method of the invention works at the cellular level, by carrying out a cell count on small imagettes (which is therefore parallelizable for the first time) then by carrying out calculations of matrix cell grouping on the basis of the number of cells counted, which is faster and more efficient than current methods.

[0020] According to a particular characteristic, the determination step comprises, for a current tile of the set of tiles, a step of determining the data representative of a number of malignant cells of the current tile using a first neural network, previously trained to identify malignant cells.

[0021] According to a particular characteristic, the determination step comprises, for a current tile of the set of tiles, a step of determining the data representative of a total number of cells of the current tile using a second neural network, previously trained for the identification of cells.

[0022] Thus, more specialized cell search networks are available, which can be trained separately to increase the detection and counting rate of cells within the tissue. According to a particular characteristic, the first and second neural networks form a single neural network. Thus, in this situation, the speed of obtaining cell counts is increased, allowing the described method to be implemented in real or near real time.

[0023] According to a particular characteristic, the step of detecting the presence of malignant cell proliferation within said histopathological image comprises at least one detection iteration comprising the following steps: cutting a current matrix of positively marked cells into at least three resulting matrices of positively marked cells, each resulting matrix comprising an overlapping zone with another of the at least two remaining resulting matrices; identical cutting of a current matrix of the number of cells, delivering at least three resulting matrices of the number of cells; obtaining, for each pair of resulting matrices of positively marked cells and resulting matrices of the number of cells, a temporary proliferation index, which is recorded within a data structure;each pair of resulting matrices forming the current matrices of the next iteration of the detection step until a predetermined stopping parameter is reached.;

[0024] This methodology of grouping cell boxes together is effective for distributing calculations across multiple calculation units in parallel, in an increasing manner, in order to delimit proliferation zones precisely.

[0025] According to a particular characteristic, the detection step further comprises, after the detection iterations, a step of classifying, in descending order, the temporary proliferation indices so that the highest proliferation indices are associated with the elements of the resulting matrices which identify said least one data representative of malignant cell proliferation within the histopathological image.

[0026] This makes it possible to process the most promising clusters first if parallel computing resources are limited.

[0027] According to a particular characteristic, the step of detecting the presence of malignant cell proliferation within said histopathological image comprises at least one step of determining, from at least one element, of the first matrix Mp, having the highest value, at least one group of contiguous elements whose values ​​make it possible to calculate a temporary proliferation index of a value higher than those of other groups of contiguous elements, said group constituting a core of iterative research of a proliferation zone within said histopathological image.

[0028] Thus, this approach makes it possible to target several potential areas in parallel, depending on the threshold chosen for the selection of the highest values.

[0029] According to a particular characteristic, the detection step further comprises, after the step of determining said at least one core of elements, at least one iteration of a step of expanding said at least one core so as to maximize the temporary proliferation index obtained by selecting contiguous elements of said first matrix Mp and corresponding contiguous elements of the second matrix Mt.

[0030] According to a particular feature, the kernel expansion iterations are implemented using parallelizable computing means.

[0031] According to another aspect, the disclosure also relates to an electronic device for detecting the presence of at least one data item representative of malignant cell proliferation within a histopathological image.Such a device comprising modules for: cutting the image representative of the tissue section, delivering a set of tiles of predetermined size, each tile having a position within the image representative of the tissue section; determining, for each tile of the set of tiles, data representative of a number of malignant cells and data representative of a total number of cells, these data being inserted respectively into a first matrix Mp and into a second matrix Mt, each element of which is associated with a tile position within the image representative of the tissue section; detecting the presence of malignant cell proliferation within said histopathological image, said detection implementing calculations of grouping data from the first matrix Mp and the second matrix Mt, according to an algorithm for distributing the calculations in parallel.

[0032] According to another aspect, the invention also relates to a computer program capable of implementing the method described as well as to a data medium for recording this computer program.

[0033] The determination device has the architecture of a computer. It is equipped with one or more processors capable of executing all types of computer programs, from operating systems to application software, written in compiled or interpreted languages. The various components of the device are connected to each other by a communication bus. The device may optionally be equipped with a communication system to communicate via protocols such as Bluetooth, Ethernet or WiFi with other systems and to connect to mobile or non-mobile telecommunications networks. The device also includes memory components that will record the data and programs necessary for the operation of the device.The device is further modified so that it can perform determination operations on a large number of images and manage several thousand simultaneous operations per second, in particular by parallel implementation of calculations for grouping cell proliferation zones.

[0034] The data carriers may be any entity or device capable of storing the programs. For example, the carriers may comprise a storage means, such as a ROM, for example a CD-ROM or a microelectronic circuit ROM, or a magnetic recording means such as a hard disk, or more often a Flash memory. On the other hand, the carriers may be transmissible media such as an electrical or optical signal, which may be conveyed via an electrical or optical cable, by radio or by other means. The programs according to the invention may in particular be downloaded from a network such as the Internet. Alternatively, the information carrier may be an integrated circuit in which the program is incorporated, the circuit being adapted to execute or to be used in the execution of the method in question.

[0035] Description of the drawings

[0036] Other characteristics and advantages of the invention will appear more clearly on reading the following description of a particular embodiment, given as a simple illustrative and non-limiting example, and the appended drawings, among which:

[0037] [Fig. 1] represents the main steps of the detection process;

[0038] [Fig. 2] illustrates the cutting and summing phases of a first example of realization;

[0039] [Fig. 3] illustrates the core determination and expansion phases of a second exemplary embodiment;

[0040] [Fig. 4] illustrates a device for implementing the detection method.

[0041] Description of an embodiment As explained previously, an object of the disclosure is to make it possible to obtain data representative of a proliferation index which is faster and more efficient than current techniques. Figure 1 illustrates the steps of the detection method which is the subject of the disclosure.

[0042] Such a method comprises: the cutting (S01) of the image representative of the IRCT tissue section, delivering a set of ETU tiles of predetermined size, each tile having a position within the image representative of the tissue section; the determination (S02) (e.g. counting), for each tile of the set of tiles, of data representative of a number of malignant cells and of data representative of a total number of cells, these data being inserted respectively into a first matrix Mp and into a second matrix Mt of which each element is associated with a tile position within the image representative of the tissue section; the detection (S03) of the presence of malignant cell proliferation within said histopathological image, said detection implementing calculations of groupings of data from the first matrix Mp and the second matrix Mt, according to an algorithm for distributing the calculations in parallel.

[0043] Thus, a process is implemented in several successive phases. The second of these phases consists of counting the positively marked cells and counting the negatively marked cells. This counting is carried out for a set of tiles that are derived from a division of the original image. The tiles are of fixed size, and their extraction is carried out without overlap (this means that there is no overlapping area, or sliding area on the tiles). Each tile is supplied to a detector, this detector being configured to deliver the total number of positively marked cells and the total number of negatively marked cells within a tile.Thus, at the end of this first phase, a first matrix of the number of positively marked cells (called the first matrix Mp) and a matrix comprising the number of negatively marked cells (or the total number of cells depending on the embodiments, called the second matrix Mt) are associated with the input histopathological image.

[0044] Each element (cell) of these two matrices Mp and Mt corresponds to a tile of the input histopathological image.

[0045] For example, purely as an illustration of the technique used, an input histopathological image could be of size 4096*8192 pixels. Each tile is of fixed size, for example 128*128 pixels. The first matrix Mp and the second matrix Mt then both have a size of 32*64 cells (elements) respectively and the values ​​of these elements correspond to cell numbers.

[0046] The last phase consists, from the first matrix Mp and the second matrix Mt, in identifying one or more proliferation zones (hotspots) by performing groupings of the cells (elements, boxes) of these matrices. It is important to note that these groupings are performed on the basis of the matrix data, and not on the basis of the images or tiles, as is the case in prior art techniques. This aspect of the proposed methodology is important because it allows for efficient and rapid processing of the input histopathological image, avoiding excessive resource consumption.

[0047] At least two different methods are proposed for performing this detection of proliferation zones (hotspots) from the first matrix Mp and the second matrix Mt. In the following, in connection with an exemplary embodiment, a first detection method is presented, then in connection with a second exemplary embodiment, a second detection method is presented.

[0048] Thus, unlike the techniques presented in relation to the prior art, which are based firstly on the detection of regions of interest (on the basis of histopathological images) and then on the calculation of the proliferation index within these regions of interest, the invention makes it possible to obtain this information in a much simpler and faster manner by transforming the initial problem of recognition (identification) of the regions of interest (Roi) into a problem of counting cells, which is much quicker to process subsequently, this counting making it possible both to identify the hotspot(s) and to determine the proliferation index.

[0049] Counting negatively labeled cells and positively labeled cells

[0050] In the following, we present a cell counting technique that can be implemented for a common histopathological image. This technique is based on the implementation of one or more neural networks, as detailed later. Prior to the implementation of cell counting within the current histopathological image, it is recalled that all the images (i.e. the current histopathological image, but also the histopathological images that were used for training the neural network(s)) have been preprocessed in order to improve the quality of the images, the accuracy of the segmentation and to facilitate subsequent analysis, to prepare the images in such a way as to minimize variations, to highlight the relevant characteristics (in particular staining) and to facilitate an accurate and reproducible analysis.

[0051] As indicated, in at least one exemplary embodiment, the cell counting is carried out by implementing at least one neural network. This neural network is trained, according to the present invention, by providing it with a set of images, of predetermined size, which have been manually marked (annotated). Unlike existing techniques, however, the manual marking was carried out at the level of the cell itself (and not at the level of regions of interest). More particularly, the manual marking was carried out by practitioners who identified, within the images presented to them, the positively marked cells, the negatively marked cells, and in a variant, the lymphocytes. These training images preferably have a resolution of 0.25 pm (corresponding to a zoom of approximately x40) to allow precise detection of individual cells and their nuclear characteristics.To train the cell detection model, annotations for each image were prepared, indicating the regions of the image that contain Ki67-positive and negative cells, as well as the regions of the image that contain lymphocytes. These annotations are provided at the pixel level, meaning they specify which parts of the image belong to each category. These annotations serve as a “ground truth” for the model during training, allowing it to learn to correctly segment Ki67-positive / negative cells as well as lymphocytes. The inclusion of lymphocytes allows the algorithm to distinguish them from tumor cells and thus reduce any segmentation bias.

[0052] According to this, the training is supervised and carried out offline. The images used for training are normalized and the information on the presence of positively labeled cells, all cells and lymphocytes is provided to the neural network to enable learning. At the end of the learning, one or more neural networks are thus available which are able to carry out cell identification as explained above.

[0053] According to the present invention, in at least one exemplary embodiment, after this learning, when the models are stabilized, the neural network(s) can be modified, to add an output layer dedicated to counting cells of the different categories (positively marked, negatively marked and lymphocytes). In another exemplary embodiment, a counting module is additionally implemented by taking as input the classification data from the neural network(s).

[0054] In any case, in possession of the neural network(s) trained as explained above, the current histopathological image is used to obtain both the first matrix Mp and the second matrix Mt. To do this, the following method is implemented, it comprises: a step of segmentation of the tissue of the current histopathological image, making it possible to isolate the tissue from the rest of the image (in particular from the background and artifacts); depending on the situations, the segmentation may have been carried out upstream or may not be necessary; a tiling step in which the area of ​​the segmented histopathological image containing the tissue is cut into tiles, for example of size 256x256 pixels, the coordinates of each tile in the (segmented) histopathological image being preserved, so as to be able to associate the quantities of cells revealed subsequently with a location in the original image;for each tile, a prediction step, by means of the neural network(s) described above, estimating the total number of cells (or the number of negative cells) and the number of positively marked cells, these numbers being recorded, in the cell (element) of the corresponding matrix, at the location relative to this tile.;

[0055] Depending on the operational implementation conditions, the prediction step can be implemented in series or in parallel, in particular when several parallelizable computing resources can be implemented on the detection device.

[0056] In all cases, at the end of this step, we have, for the segmented histopathological image, the first matrix Mp and the second matrix Mt.

[0057] Depending on the embodiments, two different neural networks can be used: for example, a network that is responsible for detecting positively marked cells, for each tile, and a network that is responsible for detecting the total number of cells (or a network that is responsible for detecting the number of negatively marked cells). This type of configuration makes it possible to further specialize the neural networks, and therefore to obtain, under certain conditions, more satisfactory results for the detection of certain cells, depending for example on the tissues from which the histopathological images are extracted.

[0058] Depending on the embodiments, several neural networks can be implemented to identify cells (in particular to detect positively labeled cells). In such a situation, each network is trained differently. The advantage of such a configuration, although more computationally intensive (for both training and inference), is that it allows the individual results (i.e., the number of cells detected) of each neural network to be averaged to ensure that the averaged results are even more representative of the actual number of cells in each tile.

[0059] Identification of hotspots and calculation of proliferation indices

[0060] The second step is to determine the locations of hotspots, if any, in the original image.

[0061] In a first exemplary embodiment, described in relation to FIG. 2, this step of determining proliferation zones of malignant cells is implemented by performing a plurality of divisions and summation calculations. More particularly, in this exemplary embodiment, the identification of the proliferation zones is implemented by performing increasingly fine divisions within the first matrix Mp and the second matrix Mt, each of these two matrices having identical starting sizes. In FIG. 2, a fictitious example of the matrix Mp is illustrated. The division of each of the matrices is carried out alternately horizontally and vertically. At each division, at least three resulting sub-matrices of level 0 are obtained (Mp01, Mp02, Mp03), each of the resulting matrices having an overlapping zone with the resulting matrix which precedes it.Then for each of the resulting level 0 matrices, a sum of the number of cells (a sum of the numbers entered in the boxes of the resulting matrix) is carried out (a sum for the resulting level 0 matrix of the positively marked cells and a sum for the resulting level 0 matrix of the negatively marked cells). A temporary proliferation index is then obtained for each "binomial" of resulting level 0 matrix. The "binomials" of resulting level 0 matrices are then classified in the decreasing order of their temporary proliferation indices.Then, for each “binomial” of resulting matrices of level 0, either in descending order or in parallel (depending on the operational implementation conditions, in particular when parallel computing means are available, such as a graphics card for example), a division (for example horizontal if the first division was vertical, and vice versa) is carried out according to the same logic (Mp011, Mp012, Mp013): at least three resulting sub-matrices of level 1 are obtained (for the current resulting matrix of level 0 of the positively marked cells and for the current resulting matrix of level 0 of the negatively marked cells), each of the resulting matrices of level 1 having an overlap zone with the resulting matrix of level 1 which precedes it in the division, and this for each resulting matrix of level 0.A temporary proliferation index is calculated for each of the "bins" of the resulting level 1 matrices and these bins are ranked in descending order of their temporary proliferation indices. The procedure is then continued, as for the previous iterations, by alternating horizontal and vertical divisions. In this method, it should be noted the importance of the overlapping zones in the division, which ensure that the proliferation zones are effectively detected. If this overlap were not carried out, proliferation zones might not be detected because they could be found straddling two cut matrices.The successively obtained temporary proliferation indices are stored in memory so as to obtain a series of localizable temporary proliferation indices (i.e., associable with a plurality of elements of the first matrix Mp and second matrix Mt (and therefore ultimately, with a zone of the image). At the end, the highest temporary proliferation indices (or those located above a predetermined ceiling value) are those which are stored as representative of the candidate zones of proliferation of the malignant cells. These indices can then be compared to thresholds to determine the degree of proliferation of the malignant cells, and possibly the cancerous phases associated therewith, these comparison operations not being the subject of the technique described here, with the exception of the comparison to thresholds which makes it possible to discriminate the zones of interest from the other zones.Taking the purely illustrative example of the first matrix Mp and the second matrix Mt, which both have a size of 32*64 cells (elements), a first horizontal division is carried out, delivering three resulting matrices of level 0 of sizes 32*32 cells (elements). In this example, there is an overlap of 16 columns between the matrices. For each of the two groups of resulting matrices of level 0 (there is a group of three resulting matrices of level 0 for the positively marked cells and a group of three resulting matrices of level 0 for the negatively marked cells), a summation of the values ​​present in the boxes of these matrices is carried out and a temporary proliferation index is calculated according to these sums for each pair of resulting matrices of level 0.At the next iteration, the binomials are split again: for example, for a current level 0 binomial (i.e. the resulting level 0 matrices of size 32*32), we obtain two groups of three resulting level 1 matrices of size 16*32 and the previous procedure is repeated, to calculate temporary proliferation indices, until no lower level splitting is possible. The temporary proliferation indices are kept at each splitting and they allow to determine which are the highest temporary proliferation indices among all the calculated indices and to keep only the three indices which are the highest (and which correspond to three different regions of the original image). The advantage of this procedure is that even in the absence of parallel computing means, it allows to progress quickly towards the determination of the proliferation index.Given such resources, this technique has made it possible to progress very quickly, even for very large images.

[0062] In a second exemplary embodiment, described in relation to Figure 3, this step of determining malignant cell proliferation zones is implemented by performing a plurality of localized expansions. More particularly, in this embodiment, one or more zone expansions are performed around one or more elements of the two matrices (the first matrix Mp and the second matrix Mt). The starting point of this method is to identify, within one and / or the other of these matrices, a predetermined number of elements (for example 3), these elements being those whose values ​​(i.e. either the number of negatively marked cells or the number of positively marked cells) are the highest (or located above a predetermined ceiling). In other words, the maximum values ​​(or close to the maxima) are sought within the matrices, for example the 3 maximum values.According to the present technique, these maximum values ​​are searched so that a minimum predetermined distance is respected between the locations (within the matrix) of these values. For example, a distance between 5 and 20 can be imposed in the search. This distance depends essentially on the size of the matrix and therefore on the size of the original image before segmentation into tiles (in the example illustrated in Figure 3, this distance is not respected for greater readability of the figure). Thus, we avoid selecting matrix elements that are too close to each other, and which are potentially members of the same hotspot. The calculation of the distance separating two elements (cells) of the matrix is ​​performed by calculating the square root of the sum of the square of the index difference and the square of the ordinate difference of the two elements.Once the predetermined number of elements is obtained, an iterative expansion method is implemented around these elements. For example, for an original matrix Mp and an element mpij of this matrix, whose value is vpij (this value being one of the three maximum values ​​obtained), an expansion is performed on all or part of the elements whose index values ​​are adjacent to the element mpij (i.e. a kernel of all or part of the index values ​​i-1 , i+1 , j-1 , j+1). These expansion iterations are illustrated by dotted lines of successively smaller sizes to illustrate two iterations only, and only for a purely illustrative example of matrix Mp. The expansion can be performed element by element or by group of elements, as shown in Figure 3. In this example, the expansion is performed by group of 4 elements (group of which the element mpij is part).As in the previous implementation example, this expansion is carried out by implementing parallel calculation means when such means exist. Thus, the four possible expansions from the element mpij are calculated in parallel. As before, the expansions are carried out in "binomials": the same locations of the elements of the matrix of the positively marked cells and the negatively marked cells are used for the expansion. The expansion includes the calculation of the sum of the values ​​of the contiguous elements of the matrices: for example the sum of (vpij, vpi+ 1 j, vpij+1 , vpi+ 1 j+1 ) and includes the calculation of a temporary proliferation index.After calculating the (four in the case presented) expansions of the mpij element, the elements that give the highest temporary proliferation index value are retained and form a "kernel" of elements for the next expansion iteration, which is performed according to the same principles as those explained above, with expansion parameters that can be the same or different (for example, it is possible to perform the expansion by selecting either three additional matrix elements again that are adjacent to one of the four elements that form the kernel, or it is possible to select a complete group of four new matrix elements, or to mix these expansion calculations). Whatever the parameters retained, again, this expansion is preferably performed using parallel calculations when they are available.Furthermore, as in the previous example, the "cores" can be ranked in descending order of their temporary proliferation indices and also be subject to expansions, even if these are considered less promising. Finally, the expansion process is stopped when the temporary proliferation index successively decreases during a predetermined number of iterations: it is then considered that a temporary hotspot has been previously determined and the coordinates of the elements of the matrices used to determine this temporary hotspot are preserved as is the value of the temporary proliferation index assigned to it.

[0063] At the end of all the expansion phases, whether carried out in parallel or not, a plurality of temporary hotspots is obtained, and the temporary hotspots with the highest proliferation indices are retained.

[0064] As explained previously, these techniques demonstrate greater efficiency when the calculation steps are implemented in parallel, that is to say when the summation calculations are performed in parallel using suitable calculation means, such as parallelizable processors of graphics cards, the cores of which each receive, at each iteration, a resulting matrix (slicings) to calculate the sums of the cells present in these matrices. In the absence of such parallel calculation means, these are performed in decreasing order of the temporary proliferation indices, iteratively or recursively, so as to attempt to converge quickly towards one or more proliferation zones when they exist. Other parallel calculation techniques derivable from the characteristics relating to parallelizable matrix calculations as previously explained can of course be implemented without departing from the scope of the present invention.Furthermore, the two techniques presented above are of course combinable with each other: it is possible, for example, to use the first technique to perform divisions, then from the divisions performed, use the second technique to perform groupings by nucleus; the size of the nuclei is modified, but the technique for grouping the nuclei together is similar, it is also possible, without departing from the coverage offered by the present disclosure, to generate intermediate matrices at each iteration, these intermediate matrices representing the sum of certain cell values, and to work iteratively on these intermediate matrices in order to identify a proliferation zone. It will also be noted that, according to the present disclosure, it is possible to vary all of the parameters, whether they are division or grouping parameters, iteratively.This means, for example, that parallel computing methods (whether those exemplified previously or others) can be iteratively implemented by varying parameters (cutting size, kernel size, cluster size), until the proliferation zones with the highest proliferation indices are identified, which further increases the advantages of using parallel computing resources.

[0065] Other features and benefits

[0066] Optionally, the method may also comprise a step of displaying data representative of the number of positively marked cells, corresponding to each tile of the original histopathological image, since a matrix is ​​available in which each element (box, cell) corresponds to a specific tile of the original image. The value of each element of the matrix represents the number of positive cells detected in the corresponding tile. Such a display step may comprise: a data association step in which the matrix is ​​used to establish a spatial association between the original histopathological image and the number of positive cells in each tile: each row and each column of the matrix corresponds to a specific tile and the values ​​of the matrix provide quantitative information on the positivity of the cells in these tiles. a coordinate mapping step,in which the elements of the matrix are brought back to the coordinates of the original histopathological image: the row and column indices of each element correspond to the row and column indices of the original image, which makes it possible to establish a direct association between the number of positive cells and the location of the corresponding tiles of the original histopathological image; a step of displaying the counts of positive cells on the original image, for example in the form of annotations (such as the number of positively labeled cells) and / or in the form of a color map (on the annotation, or on the surface of the tile): for example,The use of a color scale where different colors represent different ranges of positive cell counts can provide a quick and intuitive overview of cell proliferation patterns in the original histopathological image.

[0067] 4, a simplified architecture of an electronic detection device capable of carrying out all or part of the processing as presented previously is presented. An electronic detection device comprises a first electronic module comprising a memory 41, a processing unit 42 equipped for example with a microprocessor, and controlled by a computer program 43. In at least one embodiment, the present technique is implemented in the form of a set of programs installed in part or in full on the electronic detection device. In at least one other embodiment, the present technique is implemented in the form of a dedicated component (CpX) capable of processing data from the processing units and installed in part or in full on the processing device.Furthermore, the device also comprises communication means (CIE) presented for example in the form of network components (Wi-Fi, 3G / 4G / 5G, wired, RFID / NFC, Bluetooth, BLE, LPWan, VLC, etc.) which allow the device to receive data (I) from entities connected to one or more communication networks and to transmit processed data (T) to such entities. Such a device further comprises parallelizable calculation means (MCP), which are used to carry out matrix calculations, as previously explained, to search for and detect, in the matrices Mp and Mt, the presence of one or more proliferation hotspots within the image representative of the tissue section.

[0068] Such a device comprises, depending on the embodiments, a module for dividing the image of the tissue section into a set of tiles of predetermined size; a module for determining the data representing the number of malignant cells and the data representing the total number of cells, these data being inserted respectively into the first matrix Mp and into the second matrix Mt, each element of which is associated with a tile position within the image representative of the tissue section; and a module for detecting the presence of malignant cell proliferation within said histopathological image which implements calculations of grouping data from the first matrix Mp and the second matrix Mt, according to an algorithm for distributing the calculations in parallel.

[0069] As explained previously, these modules are implemented through components, software or hardware.

Claims

CLAIMS 1. Method for detecting the presence of at least one data item representative of a malignant cell proliferation within a histopathological image, said method being implemented within an electronic processing device, the method comprising the following steps: cutting (S01) of the image representative of the tissue section, delivering a set of tiles of predetermined size, each tile having a position within the image representative of the tissue section; determining (S02), for each tile of the set of tiles, a data item representative of a number of malignant cells and a data item representative of a total number of cells, these data being inserted respectively into a first matrix Mp and into a second matrix Mt, each element of which is associated with a tile position within the image representative of the tissue section;detection (S03) of the presence of a malignant cell proliferation within said histopathological image, said detection implementing calculations of data groupings of the first matrix Mp and of the second matrix Mt, according to an algorithm for distributing the calculations in parallel, comprising at least one detection iteration comprising the following steps: cutting a current matrix of the positively marked cells into at least three resulting matrices of the positively marked cells, each resulting matrix comprising an overlap zone with another of the at least two remaining resulting matrices; identical cutting of a current matrix of the number of cells, delivering at least three resulting matrices of the number of cells;obtaining, for each pair of resulting matrices of positively marked cells and resulting matrices of the number of cells, a temporary proliferation index, which is recorded within a data structure; each pair of resulting matrices forming the current matrices of the following iteration of the detection step (S03) until a predetermined stopping parameter is reached.; 2. Method according to claim 1, characterized in that the determination step (S02) comprises, for a current tile of the set of tiles, a step of determining the data representative of a number of malignant cells (vp) of the current tile using a first neural network, previously trained to identify malignant cells.

3. Method according to one of claims 1 to 2, characterized in that the determination step (S02) comprises, for a current tile of the set of tiles, a step of determining the data representative of a total number of cells (vt) of the current tile using a second neural network, previously trained for the identification of cells.

4. Method according to claim 3, characterized in that the first and second neural networks form a single neural network.

5. Method according to claim 1, characterized in that the detection step further comprises, after the detection iterations, a step of classifying, in descending order, the temporary proliferation indices so that the highest proliferation indices are associated with the elements of the resulting matrices which identify said least one data representative of malignant cellular proliferation within the histopathological image.

6. Method according to any one of claims 1 to 4, characterized in that the step of detecting (S03) the presence of a malignant cell proliferation within said histopathological image comprises at least one step of determining, from at least one element, of the first matrix Mp, having the highest value, at least one group of contiguous elements whose values make it possible to calculate a temporary proliferation index of a value higher than those of other groups of contiguous elements, said group constituting a core of iterative research of a proliferation zone within said histopathological image.

7. Method according to claim 6, characterized in that the detection step further comprises, after the step of determining said at least one core of elements, at least one iteration of a step of expanding said at least one core so as to maximize the temporary proliferation index obtained by selecting contiguous elements of said first matrix Mp and corresponding contiguous elements of the second matrix Mt.

8. Method according to one of claims 6 and 7, characterized in that the kernel expansion iterations are implemented using parallelizable calculation means.

9. Electronic device for detecting the presence of at least one piece of data representative of malignant cell proliferation within a histopathological image, the device comprising modules for: cutting the image representative of the tissue section, delivering a set of tiles of predetermined size, each tile having a position within the image representative of the tissue section; determining, for each tile of the set of tiles, a piece of data representative of a number of malignant cells and a piece of data representative of a total number of cells, these data being inserted respectively into a first matrix Mp and into a second matrix Mt, each element of which is associated with a tile position within the image representative of the tissue section;detecting the presence of malignant cell proliferation within said histopathological image, said detection implementing calculations of data groupings of the first matrix Mp and the second matrix Mt, according to an algorithm for distributing the calculations in parallel, which iteratively implements the following means of: cutting a current matrix of positively marked cells into at least three resulting matrices of positively marked cells, each resulting matrix comprising an overlap zone with another of the at least two remaining resulting matrices; identical cutting of a current matrix of the number of cells, delivering at least three resulting matrices of the number of cells;obtaining, for each pair of resulting matrices of positively marked cells and resulting matrices of the number of cells, a temporary proliferation index, which is recorded within a data structure; each pair of resulting matrices forming the current matrices of the following iteration of the detection step (S03) until a predetermined stopping parameter is reached.; 10. Computer program comprising instructions for implementing the method according to one of claims 1 to 10, when said instructions are executed by a processor of a computer processing circuit.

11. Data carrier on which a computer program according to claim 10 is recorded.

Images