Multichannel extended depth of field method for automated digital cytology

Abstract

This invention relates to a method (M) for generating a color-faithful EDF, i.e., an extended depth-of-field image, from a chromatic volume (1), wherein voxels with fixed indices from 1 to L are 2D images acquired using a microscope at different depths of focus in the z-direction. The method includes the following steps: generating a grayscale volume (2); applying a (M10) reversible color-to-grayscale transformation to the chromatic volume (1); applying a (M20) wavelet transform to the grayscale volume (2) to obtain a 3D wavelet coefficient matrix (3); selecting wavelet coefficients using a predefined coefficient selection rule; generating a (M40) 2D wavelet coefficient matrix WCM and a 2D coefficient map CM; and applying a (M40) wavelet coefficient matrix WCM to the 2D wavelet coefficient matrix WCM. M applies (M50) the inverse wavelet transform to obtain a 2D grayscale EDF image (4); generates a 2D color composite image CC; applies (M70) the inverse color-to-grayscale transform to the 2D grayscale EDF image (4) to obtain a 2D color EDF image (5); transforms (M80) the 2D color composite image CC and the 2D color EDF image (5) into a color space including at least one chromaticity component and at least one intensity component; and connects (M90) at least one chromaticity component of the 2D color composite image CC and at least one intensity component of the 2D color EDF image (5) to obtain a color-faithful extended depth-of-field image.

Description

Technical Field

[0001] This invention relates to the field of image processing of biological sample images. In particular, this invention relates to the field of image analysis of urine samples in the context of automated digital cytology. Background Technology

[0002] One limitation of traditional optical microscopy is the impossibility of obtaining a fully focused image of an object whose thickness exceeds the depth of focus along the optical axis. To overcome this problem, in digital cytology, a sequence of images is acquired as the object of interest moves along the optical axis (z-axis), resulting in a chromatic volume comprising the acquired image sequence; hence, it is often referred to as "image stacking," "z-stack," "3D image," or "3D image stacking." Inevitably, different parts or details of the object of interest are focused in different images of the chromatic volume.

[0003] To facilitate the analysis and storage of the acquired volume, a common solution is to fuse the acquired image sequence into a single synthetic image. For this purpose, the image containing the most information can be selected from all images of the volume (i.e., the image stack). In this "optimal focus" selection method, almost no computation is required. However, when the object of interest is deeper than the depth of field of the acquisition module, such as in the presence of cell clusters, the "optimal focus" method can lead to information loss.

[0004] The Extended Depth of Field (EDF) method is used to fuse z-stacked images into a single 2D image (EDF image), where all parts of the object of interest appear to be in focus, while avoiding information loss associated with the “best focus” method described earlier.

[0005] EDF methods are generally classified into three categories: spatial domain methods; transformation-based methods (such as wavelet transform (WT)-based methods); and deep learning-based methods.

[0006] A relatively easy-to-implement spatial domain approach involves selecting a focal region within each of the z-stacked images; then fusing the selected regions to obtain a single 2D image in which all the different parts of the object of interest are in focus.

[0007] Another example of the spatial domain approach is the smooth manifold extraction (SME) method described by a. Shihavuddin et al. in “Smooth 2D manifold extraction from 3D image stack” (2017).

[0008] A common drawback of spatial domain methods is the loss of information when thick or overlapping objects of interest (e.g., cell blocks) are present. Therefore, these methods are not suitable for transparent and translucent images (e.g., cellular images obtained through transmission microscopy).

[0009] Common wavelet transform-based methods include the Discrete Wavelet Transform (DWT) algorithm and the Stationary Wavelet Transform (SWT) algorithm. APBradley and PCCamford have demonstrated in "A One-pass Extended Depth of Field Algorithm Based on the Over-complete Discrete Wavelet Transform" (2004) that the latter has higher performance.

[0010] Traditionally, wavelet transform-based methods for computing EDF images of image stacks involve applying a wavelet transform algorithm to each image in the stack to obtain a wavelet transform stack; then, a coefficient selection rule is applied to select the most relevant wavelet coefficients from the wavelet transform stack. Traditionally, coefficient selection rules are well-defined for grayscale images, and several existing techniques compute stacked EDF images by applying wavelet transform to grayscale images (i.e., images containing only one channel). This technique can also be applied to color images, i.e., images containing more than one color channel. One possible approach to managing multi-channel images is to sequentially apply wavelet transform, a coefficient selection strategy, and inverse wavelet transform to each channel to obtain an EDF image for each channel. The resulting EDF images are then merged to obtain a single color EDF image. The main drawback of this approach is the generation of color saturation and / or artifacts. To overcome this problem, B. Forster et al. proposed a preprocessing step of color-to-grayscale transformation and a postprocessing step of color redistribution in "Extended Depth-of-Focus for multi-channel microscopy images: a complex wavelet approach" (2004). However, this color redistribution step can lead to color inaccuracies. Furthermore, methods based on complex wavelet transform (CWT) (such as those used by B. Forster et al.) often produce ringing artifacts near the edges of objects of interest, thus limiting the recovery of image details.

[0011] Therefore, a major challenge for the EDF method in the context of automated digital cytology is to accurately reproduce all the details of the object of interest while maintaining color fidelity.

[0012] Improved color fidelity is needed in several areas; for example, color is a crucial feature for bladder cancer detection in urinary cytopathology. Urinary cytopathology is the preferred non-invasive method for detecting bladder cancer; it involves analyzing microscopic slides containing urothelial cells to look for abnormalities. This method is satisfactory for detecting advanced cancers; however, it is time-consuming, expensive, and ineffective for detecting early-stage bladder cancer. Early detection is critical, aiming to increase patient survival while significantly reducing treatment costs, which are known to be much higher for advanced cancers. Automated solutions can accelerate the diagnostic process by reducing analysis time and aid in the management of early-stage cancers by improving the detection and quantification of abnormalities through the execution of segmentation algorithms.

[0013] The present invention aims to overcome the problems associated with generating EDF images from chromosomes in the field of image analysis of biological samples. Summary of the Invention

[0014] This invention relates to a computer-implemented method for generating color-faithful extended depth-of-field (EDF) images from a chromatograph of a biological sample having dimensions (N, M, L) and values ​​I (n, m, l), wherein voxels (n, m, l) of fixed l from 1 to L are images acquired at different depths of focus in the z-direction using a microscope or scanner, the method comprising the following steps:

[0015] a) Receive a chromaticity and generate a grayscale volume by applying a reversible color-to-grayscale transformation to the chromaticity;

[0016] b) Apply wavelet transform to each set of N×M voxels of a gray volume with the same l index and (n, m) indices from (1, 1) to (N, M) to obtain a 3D wavelet coefficient matrix, where the value of each voxel with (n, m, l) index comprises the set of wavelet coefficients.

[0017] c) For each set of L voxels of a 3D wavelet coefficient matrix with the same (n, m) index and l index from 1 to L, select a set of wavelet coefficients using a predefined coefficient selection rule.

[0018] d) Generation:

[0019] -2D coefficient map CM, where the value of each pixel with (n, m) index CM(n, m) is the l index of the voxel of the 3D wavelet coefficient matrix (3) which includes the set of wavelet coefficients selected by the coefficient selection rule;

[0020] - A 2D wavelet coefficient matrix WCM, where the value of each pixel with an index of (n, m) is WCM(n, m), which is a set of wavelet coefficients selected from the 3D wavelet coefficient matrix according to the coefficient selection rules;

[0021] e) Apply the inverse wavelet transform to the 2D wavelet coefficient matrix WCM to obtain a 2D grayscale EDF image;

[0022] f) Generate a 2D color composite image CC, where the value of each pixel CC(n,m) with index (n,m) is the value of the voxel I(n,m,l) of the chromatic volume l index equal to the value of the 2D coefficient map CM CM(n,m).

[0023] g) Apply the inverse of the reversible color-to-grayscale transformation to the 2D grayscale EDF image to obtain a 2D color EDF image;

[0024] h) Convert the 2D color composite image (CC) and the 2D color EDF image to a color space that includes at least one chromaticity component and at least one intensity component;

[0025] i) Connect at least one chromaticity component of a 2D color composite image CC and at least one intensity component of a 2D color EDF image to obtain a color-faithful extended depth-of-field (EDF) image.

[0026] Advantageously, the method of the present invention allows for accurate color reproduction and restoration of fine details in EDF images obtained from chromatic bodies. In particular, the method significantly reduces color inaccuracies associated with prior art color redistribution strategies without compromising the level of detail.

[0027] Fine-grained restoration of details is fundamental to achieving accurate segmentation, and, in the context of automated digital cytology, precise segmentation of the cytoplasm and nucleus is crucial:

[0028] (i) Improve the detection of abnormal cells; (ii) Simplify cell counting and

[0029] (iii) Calculate biomarkers with high precision that are relevant to clinical information, such as the nucleus / cytoplasm ratio.

[0030] In this method, the set of wavelet coefficients may include at least four wavelet coefficients, and the subset of wavelet coefficients selected using predefined coefficient selection rules may include at least one wavelet coefficient.

[0031] The number of wavelet coefficients in the 3D wavelet coefficient matrix depends on the decomposition level of the wavelet transform. In one embodiment, the wavelet transform includes multiple decomposition levels applied sequentially. In another embodiment, the wavelet transform includes only one level of decomposition, thereby obtaining a color-faithful extended depth-of-field (EDF) image with fine detail restoration in fewer computational steps. In this embodiment, the set of wavelet coefficients includes four wavelet coefficients, and a selected subset of wavelet coefficients includes at least one and no more than four coefficients.

[0032] In one embodiment, in the selection step, a first subset of wavelet coefficients is selected using a first predefined coefficient selection rule, and a second subset of wavelet coefficients is selected using a second predefined coefficient selection rule. In this embodiment, the steps of generating the 2D wavelet coefficient matrix WCM and the 2D coefficient graph CM include:

[0033] - Generate the first 2D coefficient map;

[0034] - Generate a second 2D coefficient map;

[0035] - Combine the first and second 2D coefficient maps to obtain a 2D coefficient map CM.

[0036] - Generate a 2D wavelet coefficient matrix WCM, where the value of each pixel WCM(n,m) is the value of a voxel with an index of (n,m) in the 3D wavelet coefficient matrix and an index of l that is equal to the value of the obtained 2D coefficient map CM(n,m).

[0037] The values ​​of voxels in the 3D wavelet coefficient matrix with l indices equal to the values ​​of the 2D coefficient map CM can include the complete set of wavelet coefficients.

[0038] In the first 2D coefficient map, the value of each pixel with an index of (n, m) is the l-index of a voxel of the 3D wavelet coefficient matrix, which includes a subset of wavelet coefficients selected by the first coefficient selection rule.

[0039] In the second 2D coefficient map, the value of each pixel with an index of (n, m) is the l-index of a voxel of the 3D wavelet coefficient matrix, which includes a subset of wavelet coefficients selected by the second coefficient selection rule.

[0040] In the 2D wavelet coefficient matrix WCM, the set of wavelet coefficients in pixels with (n, m) indices is the set of coefficients in the 3D wavelet coefficient matrix that have the same (n, m) index and whose 2D coefficient map values ​​are used as index l. In this particular embodiment, the 2D coefficient map is obtained by combining the first and second coefficient maps.

[0041] In this embodiment, a first 2D wavelet coefficient matrix WCM is also generated in the step of generating the first 2D coefficient map. Similarly, a second 2D wavelet coefficient matrix WCM is also generated in the step of generating the second 2D coefficient map. However, unlike the first and second 2D coefficient maps, the first and second 2D wavelet coefficient matrices do not need to be combined. In fact, in this particular embodiment, the 2D wavelet coefficient matrix WCM is generated based on the first and second 2D coefficient maps.

[0042] This embodiment allows for the selection of detail information using a first coefficient selection rule and the acquisition of denoising effect using a second coefficient selection rule. The step of combining the first and second 2D coefficient maps allows for the acquisition of a 2D coefficient map CM, from which a wavelet coefficient matrix WCM is then calculated. Advantageously, the combination step allows for the acquisition of both the wavelet coefficient matrix WCM and the 2D coefficient map CM, which possess the detail information obtained by means of the first coefficient selection rule and the denoising effect obtained by means of the second coefficient selection rule.

[0043] According to one embodiment, the step of combining 2D coefficient maps includes:

[0044] - Filter the first 2D coefficient map and the second 2D coefficient map, preferably using a median filter;

[0045] -Average the first and second 2D coefficient graphs;

[0046] - Round the average 2D coefficient plot to obtain a 2D coefficient plot CM.

[0047] This embodiment allows for the avoidance of the interpolation step. Values ​​obtained in the rounded 2D coefficient map CM are used to select wavelet coefficients from a 3D wavelet coefficient matrix having an index equal to the obtained values. A wavelet coefficient matrix WCM is then constructed based on the selected coefficients.

[0048] In one embodiment, the color-faithful extended depth-of-field (EDF) image obtained using the present invention is converted to a different color space, preferably a color space of the chromaticity. This allows for simplification of other steps in image analysis, such as segmentation, by selecting a color space in which the algorithm used in other steps of image analysis works best.

[0049] In one embodiment, the reversible color-to-grayscale transformation is principal component analysis (PCA). PCA has the advantages of being a standard, easily implemented dimensionality reduction technique, and it is also reversible.

[0050] According to one embodiment, in the step of applying wavelet transform, stationary wavelet transform (SWT) is applied. In the presence of overlapping or partially overlapping objects, SWT exhibits better performance metrics compared to other methods (such as CWT-EDF and the "best focus" method). Furthermore, in the presence of color images, SWT-EDF achieves the best color fidelity compared to other wavelet-based methods, followed by color reconstruction strategies.

[0051] In one embodiment, the chromatic volume is a biomedical image, and the method further includes a step of segmenting a color-faithful extended depth-of-field (EDF) image. Segmentation performance is improved when applied to images with finely reproduced details and enhanced color fidelity.

[0052] The present invention also relates to a system for analyzing biological samples, the system having a processor and a computing module, the processor and computing module being configured to:

[0053] - Receive the chromaticity and generate a grayscale volume by applying a reversible color-to-grayscale transformation to the chromaticity.

[0054] - Apply wavelet transform to each set of N×M voxels of a grayscale volume with the same l index and (n, m) indices from (1, 1) to (N, M) to obtain a 3D wavelet coefficient matrix, where the value of each voxel with (n, m, l) index comprises the set of wavelet coefficients.

[0055] - For each set of L voxels in a 3D wavelet coefficient matrix with the same (n, m) index and l index from 1 to L, select a set of wavelet coefficients using a predefined coefficient selection rule.

[0056] - Generate a 2D wavelet coefficient matrix WCM, where the value of each pixel with an index of (n, m) is WCM(n, m), which is a set of wavelet coefficients selected from the 3D wavelet coefficient matrix according to the coefficient selection rules;

[0057] - Generate a 2D coefficient map CM, where the value of each pixel with an index of (n, m) CM(n, m) is the l index of the voxel of the 3D wavelet coefficient matrix (3), which includes the set of wavelet coefficients selected by the coefficient selection rule;

[0058] - Apply the inverse wavelet transform to the 2D wavelet coefficient matrix WCM to obtain a 2D grayscale EDF image;

[0059] - Generate a 2D color composite image CC, where the value of each pixel CC(n,m) with an index of (n,m) is the value of the voxel I(n,m,l) with an index of l that is equal to the value CM(n,m) of the 2D coefficient map CM;

[0060] - Apply an inverse reversible color-to-grayscale transformation to the 2D grayscale EDF image to obtain a 2D color EDF image;

[0061] - Convert 2D color composite images (CC) and 2D color EDF images to a color space that includes at least one chromaticity component and at least one intensity component;

[0062] - Connect at least one chromaticity component of a 2D color composite image (CC) and at least one intensity component of a 2D color EDF image to obtain a color-faithful extended depth-of-field (EDF) image.

[0063] According to one embodiment, the system for analyzing biological samples according to the present invention is an automated digital cytology system for analyzing urine samples. In this embodiment, the system further includes an acquisition module configured to acquire chromatographs comprising at least two color images.

[0064] According to one embodiment, the system of the present invention is a bladder cancer detection system.

[0065] The present invention also relates to a computer program product for analyzing biological samples, the computer program product comprising instructions that, when executed by a computer, cause the computer to perform the steps of the method according to the present invention.

[0066] The present invention also relates to a computer-readable storage medium including instructions that, when executed by a computer, cause the computer to perform the steps of the method according to any of the above embodiments.

[0067] definition

[0068] In this invention, the following terms have the following meanings:

[0069] - "Stacking" refers to a volume containing at least two 2D images acquired by a microscope at different focal depths along the optical axis of the microscope.

[0070] - "EDF image" refers to a 2D image obtained from a stack of 2D images (i.e., from volume) using an extended depth-of-field method.

[0071] - "Color space" refers to the mathematical representation of color in a coordinate system. Each color is defined by a vector, and the projection of this vector onto the axes of the coordinate system are the different components of the color in the color space.

[0072] - "Chromaticity component" refers to the projection of a color vector in a color space onto an axis that contains chromaticity values ​​but not luminance values.

[0073] A chromaticity component can contain some or all of the chromaticity information of a color vector.

[0074] - An "intensity component" refers to the projection of a color vector in a color space onto an axis that contains luminance values ​​but not chromaticity information. An intensity component can contain some or all of the luminance information of a color vector.

[0075] - "Chi-square distance" refers to the distance between two standardized color histograms.

[0076] "Color fidelity" refers to accurately reproducing the colors of a volume from a 2D image obtained from the volume, that is, minimizing the distance metric between the distribution of chromaticity components in the obtained 2D image and the original 3D stack. Preferably, the distance metric is the chi-square distance (χ) between the color histogram of the 2D image and the color histogram of the volume.

[0077] - The term "processor" should not be interpreted as limited to hardware capable of executing software, but rather refers generally to a processing device, such as a computer, microprocessor, integrated circuit, or programmable logic device (PLD). A processor may also include one or more graphics processing units (GPUs), whether for computer graphics and image processing or other functions. Furthermore, instructions and / or data capable of performing related and / or resulting functions can be stored on any processor-readable medium, such as integrated circuits, hard disks, CDs (compressed discs), optical discs (e.g., DVDs (Digital Versatile Discs), RAM (Random Access Memory), or ROM (Read-Only Memory). Instructions may be specifically stored in hardware, software, firmware, or any combination thereof. Attached Figure Description

[0078] The following detailed description will be better understood when read in conjunction with the accompanying drawings. For illustrative purposes, the invention is shown in preferred embodiments. However, it should be understood that this application is not limited to the precise color spaces, reversible transformations, objects of interest, and aspects shown in the figures. Therefore, it should be understood that the inclusion of reference numerals following features mentioned in the appended claims is solely for the purpose of enhancing the comprehensibility of the claims and in no way limits the scope of the claims.

[0079] Figure 1 This is a flowchart illustrating a first non-limiting example of the method of the present invention.

[0080] Figure 2 The images shown represent non-limiting examples of images obtained at different steps according to the method of the present invention. Figure 2 'a' is an example of an image belonging to the initial 3D stack. Figure 2 b indicates from Figure 2 The grayscale EDF image obtained from the color volume to which the image in 'a' belongs. Figure 2 c represents a color EDF image obtained using color allocation techniques based on existing technologies. Figure 2 d represents a color EDF image obtained using the color reconstruction technique according to the present invention.

[0081] Figure 3 This is a flowchart illustrating a second non-limiting example of the method of the present invention.

[0082] Figure 4This is a flowchart schematically illustrating the main steps of the method of the present invention.

[0083] Figure 5 This is a block diagram schematically representing a specific mode of a system for generating color-faithful extended depth-of-field EDF images from a chromatic volume, according to the present disclosure.

[0084] While various embodiments have been described and illustrated, the specific implementation should not be construed as limited thereto. Those skilled in the art will be able to make various modifications to the embodiments without departing from the true spirit and scope of this disclosure as defined in the claims. Detailed Implementation

[0085] This specification illustrates the principles of this disclosure. Therefore, it will be appreciated that those skilled in the art will be able to design various arrangements that, while not expressly described or shown herein, embody the principles of this disclosure and are included within its scope.

[0086] All examples and conditional language described herein are intended for educational purposes to help readers understand the principles of this disclosure and the concepts contributed by the inventors to further developments in the field, and should be understood as not being limited to these specifically enumerated examples and conditions.

[0087] Furthermore, all statements herein describing the principles, aspects, and embodiments of this disclosure, as well as specific examples thereof, are intended to include their structural and functional equivalents. Moreover, such equivalents are contemplated to include both currently known equivalents and those developed in the future (i.e., any elements developed that perform the same function, regardless of their structure).

[0088] This invention relates to a computer-implemented method for generating extended depth-of-field (EDF) images from a stack of 2D microscopic images acquired at different depths of focus in the z-direction. The EDF images obtained using this method are characterized by improved color fidelity and finer detail recovery. The volume in this method is a color volume.

[0089] according to Figure 4The method used in this invention includes a receiving step of receiving body 1. According to one embodiment, body 1 is a stack of 2D color images obtained by moving a microscope slide including the object of interest along the z-direction parallel to the optical axis of the brightfield microscope. Body 1 has dimensions (N, M, L), and each voxel has a value I (n, m, l). The set of voxels of body 1 with a fixed l (included between l and l) is a 2D image acquired using a microscope at different depths of focus in the z-direction. The L dimension is equal to the number of images acquired. The l index of the voxel is the depth of focus of the initial image including the voxel that has been acquired. In this invention, the L dimension of body 1 is equal to or greater than 2, such that body 1 is a stack of 2D images including at least two 2D images acquired at different depths of focus.

[0090] The body 1 according to the invention can be represented in any color space. In one embodiment, the initial color body (i.e., a stack of 2D images) is the RGB body.

[0091] The invention will be better understood when read in conjunction with the accompanying drawings. For illustrative purposes, method M is shown in a preferred embodiment. However, it should be understood that this application is not limited to this particular embodiment, and the drawings are not intended to limit the scope of the claims to the embodiments described therein. Therefore, it should be understood that the inclusion of reference numerals following features mentioned in the appended claims is solely for the purpose of improving the comprehensibility of the claims and is in no way a limitation on the scope of the claims.

[0092] Furthermore, those skilled in the art will understand that any flowchart, etc., represents various processes that can be substantially represented in a computer-readable medium and thus executed by a computer or processor, whether or not such computer or processor is explicitly shown.

[0093] like Figure 1 As shown, the method M described herein may include the step of generating a grayscale body 2 from the chromatic body 1 by applying a reversible color-to-grayscale transformation of M10 to the chromatic body 1.

[0094] exist Figure 1 In the diagram, color volume 1 is schematically represented as a cube.

[0095] Method M may optionally include a step of preprocessing the received chroma body 1 before generating the grayscale body 2. The preprocessing step may specifically include a transformation of the chroma body 1 (e.g., decompression to an appropriate decompression standard).

[0096] In a favorable configuration, the preprocessing steps may include a normalization stage. This can improve the efficiency of downstream processing of volume 1. Such normalization can be particularly useful when processing multiple chromosomes 1 originating from different sources (including potentially different imaging systems) using the methods of this disclosure.

[0097] In order to convert the grayscale EDF image 4 obtained after applying the M50 wavelet inverse transform back to a color image, the color-to-grayscale transformation needs to be a reversible transformation.

[0098] In one embodiment, the reversible transformation from color to grayscale can be a principal component analysis (PCA) transformation; in this particular embodiment, its inverse transformation is an inverse PCA that allows the reconstruction of the original color variable from the principal components.

[0099] Since reversible color-to-grayscale transformations result in the loss of color information, a color reconstruction strategy must be followed. The steps that allow for color reconstruction according to this method are described in detail below.

[0100] The method M of the present invention includes a step M20 of applying a wavelet transform to a grayscale volume 2. The application of the wavelet transform allows for the acquisition of a 3D wavelet coefficient matrix 3 with (N, M, L) dimensions, where the value of each voxel with an index of (n, m, l) comprises a set of wavelet coefficients. The set of coefficients includes at least four wavelet coefficients. The term "3D" refers to a three-dimensional matrix. The wavelet coefficient matrix 3 is a 3D matrix because it has dimensions (N, M, L).

[0101] In this invention, the color volume 1, gray volume 2, and 3D wavelet coefficient matrix 3 are 3D matrices with dimensions (N, M, L); while the wavelet coefficient matrix WCM, coefficient image CM, grayscale EDF image 6, color composite image CC, color EDF image 7, and color faithful extended depth of field (EDF) image are 2D (i.e., two-dimensional matrices with dimensions (N, M)).

[0102] Step M20, which applies wavelet transforms to grayscale volume 2, is performed in parallel for each set of N×M voxels, where the voxels have the same l index and (n, m) indices from (1, 1) to (N, M). Therefore, for a grayscale volume 2 with (N, M, L) dimensions, L wavelet transforms are applied in step M20.

[0103] Preferably, the wavelet transform applied in step M20 is the stationary wavelet transform (SWT) because SWT-EDF achieves the best color fidelity among wavelet-based methods with reconstructed colors. This is particularly advantageous when the image obtained using this method M after concatenation step M90 is further segmented. In fact, according to the common segmentation evaluation metric "Intersection over Union" (IoU), SWT-EDF outperforms CWT-EDF and the "best focus" method.

[0104] Wavelet transform decomposes a signal into wavelet coefficients, which include approximation coefficients and detail coefficients. In the context of an image, wavelet transform is referred to as decomposing the original image into sub-images of wavelet coefficients. This decomposition is reversible, allowing the original image to be reconstructed from the wavelet coefficients by applying the inverse wavelet transform. Specifically, the wavelet transform of an image is computed by applying a first decomposition to each row of pixels in the image, and then applying a second decomposition to each column of the result of the first decomposition. The first-level wavelet decomposition of the image is the result of the first and second decompositions (i.e., row decomposition and column decomposition). The first-level wavelet decomposition of the image produces four wavelet coefficient matrices (sub-images). The pixel values ​​of the sub-images include wavelet coefficients. However, several levels of subsequent decomposition can be performed. At each decomposition level, four sub-images are obtained: three “detail” sub-images that include details of the original image and a fourth “approximation” sub-image that includes approximations of the original image. The pixel values ​​of the sub-images include wavelet coefficients. In particular, the pixel values ​​of the three “detail” sub-images include: horizontal detail coefficients in the first sub-image; vertical detail coefficients in the second sub-image; and diagonal detail coefficients in the third sub-image. The horizontal, vertical, and diagonal detail coefficients are functions of the horizontal, vertical, and diagonal details of the original image, respectively. The pixel values ​​of the fourth approximation sub-image include approximation coefficients. These approximation coefficients are functions of an approximation of the original image. In practice, at each decomposition level, the image is filtered with a high-pass filter to generate detail coefficients and with a low-pass filter to generate approximation coefficients. At each subsequent decomposition level, such filtering is performed on the image obtained in the previous decomposition steps. After one or more decomposition levels, the original image can be reconstructed by applying an inverse wavelet transform to the decomposition.

[0105] In the method M according to the invention, the wavelet transform applied in step M20 includes a first-order decomposition, thus decomposing each image of grayscale volume 2 into four sub-images including wavelet coefficients. An image of grayscale volume 2 refers to a set of N×M voxels of grayscale volume 2 with the same l index; therefore, the l-th image of grayscale volume 2 is a set of voxels with fixed l and (n, m) indices from (1, 1) to (N, M).

[0106] This method M further includes a selection step M30, wherein for each set of L voxels of the 3D wavelet coefficient matrix 3 having the same (n, m) index and l indices from 1 to L, this step is configured to select a set of wavelet coefficients using a predefined coefficient selection rule. The set of wavelet coefficients includes at least one detail coefficient or approximation coefficient from the sub-images defined above.

[0107] In one embodiment, the entire set of wavelet coefficients is used in a further step of the invention.

[0108] In one embodiment, a subset of wavelet coefficients is selected from the full set of wavelet coefficients, and this subset is further used to compute a 2D coefficient map CM. In this particular embodiment, several 2D coefficient maps can be computed; specifically, each of the 2D coefficient maps is computed based on different subsets of wavelet coefficients; subsequently, a single 2D coefficient map is obtained by combining the multiple 2D coefficient maps.

[0109] The coefficient selection rule applied in step M30 can be any coefficient selection rule known to those skilled in the art. In particular, the coefficient selection rule is applicable to grayscale images or grayscale volumes. In the case of multi-channel images, applying the coefficient selection rule to each color channel and merging the results can produce false colors; therefore, it is necessary to convert the initial color image to a grayscale image, then calculate the 2D grayscale EDF image, and finally perform color reconstruction.

[0110] In wavelet coefficient selection step M30, one or more coefficient selection rules can be applied.

[0111] The method M of the present invention may further include the step of generating M40:

[0112] -2D wavelet coefficient matrix WCM, where the value of each pixel with index (n, m) WCM(n, m) is a set of wavelet coefficients selected from 3D wavelet coefficient matrix 3 according to the coefficient selection rules;

[0113] -2D coefficient map CM, where the value of each pixel with (n, m) index CM(n, m) is the l index of the voxel of the 3D wavelet coefficient matrix (3) which includes the set of wavelet coefficients selected by the coefficient selection rule;

[0114] For example, the value of pixel CM(2,3) includes the l index of the voxels of the 3D wavelet coefficient matrix 3, which is selected from all voxels with an n index equal to 2, an m index equal to 3, and l indices from 1 to L.

[0115] According to one embodiment, in selection step M30, a subset of wavelet coefficients using the first predefined coefficient selection rule is selected, and a subset of wavelet coefficients using the second predefined coefficient selection rule is selected. In this embodiment, the steps of generating the M402D wavelet coefficient matrix WCM and the 2D coefficient map CM include multiple steps: M41, M42, M43, and M44.

[0116] Step M41 includes generating a first 2D wavelet coefficient matrix and a first 2D coefficient map from a subset of the first wavelet coefficients. In this embodiment, step M41 is followed by step M42, which generates a second 2D wavelet coefficient matrix and a second 2D coefficient map from a second subset of the wavelet coefficients. Finally, step M43 combines the first and second 2D coefficient maps to obtain a final 2D coefficient map CM. Based on the final 2D coefficient map CM, a 2D wavelet coefficient matrix WCM is generated in step M44. The value WCM(n, m) of each pixel in the 2D wavelet coefficient matrix WCM is a set of coefficients of a 3D wavelet coefficient matrix having an index of (n, m) and an index l equal to the value CM(n, m) of the final 2D coefficient map CM.

[0117] Step M43, which combines the first and second 2D coefficient maps to obtain a 2D coefficient map CM, may include multiple steps. Specifically, it may include:

[0118] - First, filter the first 2D coefficient map and the second 2D coefficient map, prioritizing the use of a median filter;

[0119] - Secondly, average the first and second 2D coefficient graphs;

[0120] Finally, the average 2D coefficient map is rounded to obtain a 2D coefficient map CM.

[0121] In one embodiment, the median filter is a 3x3 median filter.

[0122] In one embodiment, first and second subsets of wavelet coefficients are selected using first and second coefficient selection rules, respectively, generating two 2D wavelet coefficient matrices and two 2D coefficient maps. In the first 2D wavelet coefficient matrix, the value of each pixel with an index of (n, m) represents the first subset of wavelet coefficients selected from the 3D wavelet coefficient matrix 3 using the first coefficient selection rule. In the second 2D wavelet coefficient matrix, the value of each pixel with an index of (n, m) represents the second subset of wavelet coefficients selected from the 3D wavelet coefficient matrix 3 using the second coefficient selection rule. In the first 2D coefficient map, the value of each pixel with an index of (n, m) is the l-index of a voxel in the 3D wavelet coefficient matrix 3, which includes the first subset of wavelet coefficients selected by the first coefficient selection rule. In the second 2D coefficient map, the value of each pixel with an index of (n, m) is the l-index of a voxel in the 3D wavelet coefficient matrix 3, which includes the second subset of wavelet coefficients selected by the second coefficient selection rule.

[0123] According to a specific embodiment, a first selection rule is applied to select the most relevant coefficients from a subset of coefficients including detail coefficients, and a second selection rule is applied to select the most relevant coefficients from a subset of coefficients including approximate coefficients. Detail coefficients include horizontal detail coefficients, vertical detail coefficients, and diagonal detail coefficients. In this specific embodiment, the first coefficient selection rule includes:

[0124] - For each voxel of a 3D wavelet coefficient matrix with (n, m, l) indices, calculate the sum of the absolute values ​​of the horizontal detail coefficients, vertical detail coefficients, and diagonal detail coefficients with (n, m, 1) indices;

[0125] - Select the voxel with the largest sum of absolute values ​​of horizontal detail coefficients, vertical detail coefficients, and diagonal detail coefficients from all voxels of the 3D wavelet coefficient matrix with (n, m) index and l index from 1 to L.

[0126] In this embodiment, the second coefficient selection rule includes:

[0127] - For each voxel of a 3D wavelet coefficient matrix with (n, m, l) indices, compute the variance of the Laplace operator with (n, m, l) indices for the approximate coefficients.

[0128] - Select the voxel with the largest absolute value of the variance of the Laplace operator, including approximation coefficients, from all voxels of the 3D wavelet coefficient matrix with (n, m) index and l index from 1 to L.

[0129] In this embodiment, the first 2D wavelet coefficient matrix includes information about image details, while the second 2D wavelet coefficient matrix is ​​a smooth representation of the image resulting from the denoising effect of the low-pass filter of the wavelet transform.

[0130] According to one embodiment, the method includes the step of applying the inverse wavelet transform (M50) to the 2D wavelet coefficient matrix WCM to obtain a 2D grayscale EDF image 4.

[0131] In one embodiment, the method includes the step of generating a (M60) 2D color composite image CC, wherein the value CC(n,m) of each pixel having an index of (n,m) is the value I(n,m,l) of a voxel of chromosome I, the voxel I(n,m,l) of which has an index l equal to the value CM(n,m) of the 2D coefficient map CM.

[0132] In one embodiment, the method also includes step M70 of applying an inverse transform of a reversible color-to-grayscale transform to the 2D grayscale EDF image 4 to obtain a 2D color EDF image 5.

[0133] In one embodiment, the method also includes the step M80 of converting the 2D color composite image CC and the 2D color EDF image 5 to a color space including at least one chromaticity component and at least one intensity component;

[0134] The term "color space" refers to the mathematical representation of colors in a coordinate system, where each color is defined by a vector whose projection onto the axes of the coordinate system is the different components of the color in the color space.

[0135] Most color spaces use three components to represent color. Traditionally, biomedical images are acquired by devices such as cameras and scanners, whose sensors include three channels: red (R), green (G), and blue (B), thus generating images in the RGB color space. In the RGB color space, the R, G, and B components are highly correlated; furthermore, luminance information is unavailable. Therefore, for some applications, such as in step M80, a conversion from the RGB color space to another color space with a different coordinate axis system is suitable. Each coordinate axis of the color space includes color information, which may be a luminance value, luminance, or a function of luminance; or optionally, chromaticity (a combination of chromaticity and luminance values) or a function of chromaticity. Preferably, in the color space, each coordinate axis includes chromaticity information, i.e., chromaticity or a function of chromaticity, or intensity information, i.e., luminance value or luminance or a function of luminance; and none of the coordinate axes includes a combination of said chromaticity and intensity information.

[0136] In contrast to RGB, color spaces such as YUV, YES, YT1T2, CIE L*a*b*, YcbCr, CIE L*U*V*, and HSV include at least one coordinate axis that includes only intensity information and not chromaticity information, and at least one coordinate axis that includes only chromaticity information and not intensity information. In other words, each of these color spaces includes at least two axes where intensity and chromaticity information are separated.

[0137] More specifically, YUV, YES, YT1T2, CIE, L*a*b*, YcbCr, and CIE L*U*V* include one luminance component and two chrominance components.

[0138] Conversely, the HSV color space includes two intensity components, namely hue and saturation, and one chromaticity component, namely brightness / value.

[0139] In the example above, the color space of conversion step M80 includes three components. However, in alternative embodiments, four-dimensional or higher-dimensional color spaces can be used. These color spaces are commonly used in printing and graphic design.

[0140] According to one embodiment, the conversion step M80 is configured to convert the 2D color composite image CC and the 2D color EDF image 5 to the YUV color space, where the luminance component Y is the intensity component, and U and V are the chrominance components.

[0141] According to another embodiment, the color space of the conversion step M80 is one of the following color spaces: CIE L*a*b*, YcbCr, CIE L*U*V*, HSV.

[0142] In one embodiment, the method further includes step M90 of concatenating at least one chromaticity component of the 2D color composite image CC and at least one intensity component of the 2D color EDF image 5 to obtain a color-faithful extended depth-of-field (EDF) image.

[0143] Therefore, in the above example (conversion M80 of converting the 2D color composite image CC and the 2D color EDF image 5 to the YUV color space), the connection step M90 may include:

[0144] - Connect the U component of the 2D color composite image CC, the V component of the 2D color composite image CC, and the Y component of the 2D color EDF image 5.

[0145] Because the chromaticity information of the 2D color composite image CC is retrieved from the initial chromaticity volume 1 in step M60, the 2D color composite image CC has improved color fidelity compared to volume 1. Since the detail information in the 2D color EDF image 5 is maximized through the coefficient selection rule, the 2D color EDF image 5 improves the recovery of fine details from chromaticity volume 1. Connecting the intensity component of the 2D color EDF image 5 and the chromaticity component of the 2D color composite image CC advantageously allows for the combination of fine detail recovery from chromaticity volume 1 and improved color fidelity to chromaticity volume 1. Therefore, method M allows for obtaining a “color-faithful” extended depth-of-field (EDF) image with fine detail recovery from the initial volume 1.

[0146] exist Figure 1 In the specific embodiment shown, the color space has one intensity component and two chromaticity components, and in the concatenation step M90, one intensity component of the color EDF image 7 and two chromaticity components of the color composite image CC are concatenated. Therefore, in this specific embodiment, the color-faithful extended depth-of-field (EDF) image has one intensity component and two chromaticity components.

[0147] The color-faithful extended depth-of-field (EDF) image obtained in connection step M90 can be further converted to any color space. Optionally, it is converted to the color space of the initial chromaticity 1. This could be the RGB color space. In another embodiment of the invention, the obtained color-faithful extended depth-of-field (EDF) image is converted to a color space different from the color space of the initial chromaticity 1.

[0148] As described above, the color space of conversion step M80 can be a four-dimensional or higher-dimensional color space. In this case, after concatenation step M90, the color-faithful extended depth-of-field (EDF) image can be converted to a color space with a lower number of components. This embodiment is particularly advantageous because the four and higher components of such a color space typically include redundant information, as it can be derived from the first three components.

[0149] According to one embodiment, the color-faithful extended depth-of-field (EDF) image obtained in the connection step M90 is further segmented. Preferably, the U-net segmentation model is used to perform the segmentation step.

[0150] Figure 1 Non-limiting examples of steps M10 to M90 of the method according to the invention are shown. Figure 1 The selection step M30 is not shown in the text. Various modifications to this step can be implemented in steps M10 to M70 without deviating from the results of obtaining the 2D wavelet coefficient matrix WCM, the 2D coefficient map CM, the 2D color composite image CC, and the 2D color EDF image 5.

[0151] The method M of the present invention allows a 2D image with improved color fidelity and fine detail recovery to be obtained from the chromatic volume 1 by concatenating the intensity component of the 2D color EDF image 5 and the chromaticity component of the color composite image 2D color composite image CC in the color space.

[0152] One advantage of this invention is in the combination Figure 2 You will gain a better understanding while reading. Figure 2 In the example shown, the object of interest is cell 6, some of which form cell clusters 7. Specifically, Figure 2 a, 2b, and 2d represent non-limiting examples of images obtained before or after performing different steps of the method M of the present invention. Figure 2 c and 2d respectively represent: after the color redistribution strategy according to existing technology, from Figure 2 The color image obtained from the image in b; and the color redistribution strategy according to the invention from Figure 2 The color image obtained from the image in b.

[0153] More specifically, Figure 2Image 'a' is an example belonging to the initial chromaticity 1, including focused object 8 and out-of-focus object 9. In chromaticity 1 of a biological sample, typically one color is dominant. Figure 2 In the image in image 'a', the primary color of all objects is blue, except for element 10, whose primary color is red. Figure 2 In the color volume 1 of the image in 'a', blue is the primary color and red is the rare color. Figure 2 All objects in the images represented in b to 2d are in focus. Figure 2 b is an example of a 2D color EDF image 5 obtained as the output of applying the M70 reversible color-to-grayscale inverse transform step to the 2D grayscale EDF image 4. Specifically, Figure 2 The image in b is obtained after inverse PCA. After inverse PCA, the red color of element 10 disappears. Figure 2 c indicates from Figure 2 The image in b yielded a 2D color EDF image 5, in which a color allocation technique according to the prior art was applied. The red element of element 10, following the color allocation strategy of the prior art, was partially recovered. Figure 2 d represents the 2D color EDF image 5 obtained by method M of the present invention. Specifically, Figure 2 The image in d was obtained after converting the 2D color EDF image 5 and the 2D color composite image CC in the YUV color space, and then concatenating the luminance component Y of the 2D color EDF image 5 and the U and V chrominance components of the 2D color composite image CC. Figure 2 The image in c is the opposite, in Figure 2 In the image in d, the red of element 10 is fully restored, and element 10 has the same... Figure 2 The images in 'a' have the same color.

[0154] The color fidelity obtained using existing techniques and the color fidelity obtained using this method are compared by using the chi-square distance (χ). Specifically, the chi-square distance (χ) is calculated. Figure 2 The RGB normalized color histogram of each image represented in the figure is... Figure 2 The chi-square distance (χ) between the color histograms of the images in a. Figure 2 The image in d and Figure 2 The chi-square distance (χ) between the images in a is less than Figure 2 Images in c Figure 2 The chi-square distance (χ) between the images in a is obtained using this method. These results demonstrate that M... Figure 2 The image in d is advantageously superior to that obtained using existing methods. Figure 2The image in c exhibits better color fidelity. Specifically, this invention ensures accurate reproduction of elements with rare colors, i.e., elements whose chromaticity components are not the dominant chromaticity components of the original chromaticity body 1. The main problems observed in the 2D color EDF image 5 are: lack of color accuracy and the presence of artifacts. Compared to other wavelet-based methods and color reconstruction strategies, this method achieves the best color fidelity. For example, Figure 2 The examples in the paper demonstrate that, compared to existing color reconstruction methods, this method M reproduces elements with rare colors much better.

[0155] Figure 3 Specific embodiments of the method of the present invention are shown.

[0156] More precisely, in Figure 3 In the example shown, the color space of conversion step M80 has three components. Therefore, the result of conversion step M80 is schematically represented as a stack of three prisms, each prism representing one component.

[0157] As described above, in the connection step M90, at least one intensity component of the color EDF image 7 and at least one chromaticity component of the color composite image CC are connected.

[0158] exist Figure 3 In the example, the intensity component is represented by the top and middle prisms of the pile, and the chromaticity component is represented by the bottom prism of the pile.

[0159] For example, the top prism could represent the hue (H) component, the middle prism could represent the saturation (S) component, and the bottom prism could represent the value (V) component. Therefore, in this example, the connection step M90 could include:

[0160] - Connect the H component of the 2D color composite image CC, the V component of the 2D color EDF image 5, and the S component of the 2D color EDF image 5.

[0161] exist Figure 3 In the example shown, the color-faithful extended depth-of-field (EDF) image has two intensity components and one chromaticity component.

[0162] The present invention also relates to a system 11 for analyzing chromosomes 1 obtained from biological samples.

[0163] Reference Figure 5 Describe System 11.

[0164] The system 11 includes at least one input 12 adapted to receive a chromatic body 1. The chromatic body can be stored in one or more local or remote databases 13. The one or more local or remote databases 13 can take the form of storage resources that can be obtained from any kind of suitable storage device, which can be RAM or EEPROM (Electrically Erasable Programmable Read-Only Memory), such as flash memory that may be in an SSD (Solid State Drive).

[0165] System 11 includes processor 14, which is preferably configured to perform the steps of method M according to any of the above embodiments.

[0166] Although the system 11 described herein is general and is provided with several functions that can be performed alternatively or in any cumulative manner, other implementations within the scope of this disclosure include systems that have only a portion of these functions.

[0167] Each of systems 11 is advantageously a device or physical part of a device designed, configured, and / or adapted to perform the said function and produce the said effect or result. In alternative embodiments, whether grouped in the same machine or in different, possibly remote, machines, any one of systems 11 is embodied as a group of devices or physical parts of a device. System 11 may, for example, have functionality distributed on a cloud infrastructure and may be offered to users as a cloud-based service, or have remote functionality accessible via an API.

[0168] In the following text, modules will be understood as functional entities rather than physically distinct material components. Therefore, they can be combined within the same tangible and concrete component or distributed across several such components. Furthermore, each of these modules may share itself among at least two physical components. Moreover, modules can be implemented in hardware, software, firmware, or any hybrid form thereof. They are preferably embodied within at least one processor of system 11.

[0169] According to one embodiment, the system 11 further includes a visualization module for displaying the volume 1, the grayscale volume 2, the 2D color composite image CC, the 2D grayscale EDF image 4, the 2D color EDF image 5, and the color faithful extended depth of field (EDF) image.

[0170] The visualization module can also display the 3D wavelet coefficient matrix 3, the 2D wavelet coefficient matrix WCM, and the 2D coefficient graph CM.

[0171] The visualization module can be directly connected to the GPU. In one variant, the visualization module is located external to system 11 and connected to system 11 via cable or wirelessly for transmitting display signals. In this case, system 11 includes an interface for transmission or a connection suitable for transmitting display signals to an external display device (e.g., an LCD or plasma screen or a video projector).

[0172] Figure 5 The system 11 shown interacts with a user interface 15, through which the user can input and retrieve information. The user interface 15 includes any device suitable for inputting or retrieving data, information, or instructions, particularly visual, tactile, and / or audio devices, which may include any one or more of the following devices well known to those skilled in the art: screen, keyboard, trackball, touchpad, touch screen, speaker, voice recognition system.

[0173] In one embodiment, the biological sample is a urine sample. According to this embodiment, system 11 may include an automated digital cytology system configured to collect volume 1, which includes at least two 2D color images.

[0174] The present invention also relates to a computer program product for generating a color-faithful extended depth-of-field (EDF) image from a volume 1 obtained from a microscope, the computer program product including instructions that, when executed by a computer, cause the computer to perform the steps of the method according to any of the embodiments described above.

[0175] A computer program product for performing the methods described above can be written as a computer program, code segment, instruction, or any combination thereof, for individually or collectively instructing or configuring a processor or computer to operate as a machine or special-purpose computer to perform operations performed by hardware components. In one example, the computer program product includes machine code that is directly executed by the processor or computer, such as machine code generated by a compiler. In another example, the computer program product includes high-level code that is executed by the processor or computer using an interpreter. Those skilled in the art can readily write instructions or software based on the block diagrams and flowcharts shown in the accompanying drawings and the corresponding descriptions in the specification, which disclose algorithms for performing the operations described above.

[0176] The present invention further relates to a computer-readable storage medium including instructions that, when executed by a computer, cause the computer to perform the steps of the method according to any of the above embodiments.

[0177] According to one embodiment, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0178] Computer programs implementing the methods of this embodiment are typically distributed to users on distributed computer-readable storage media, such as, but not limited to, SD cards, external storage devices, microchips, flash memory devices, portable hard disk drives, and software websites. The computer program can be copied from the distributed media to a hard disk or similar intermediate storage media.

[0179] A computer program can execute by loading computer instructions from its distributed medium or intermediate storage medium into the computer's execution memory, configuring the computer to operate according to the method of the invention. All these operations are well known to those skilled in the art of computer systems.

[0180] Instructions or software for controlling a processor or computer to implement the aforementioned hardware components and perform the aforementioned methods, along with any associated data, data files, and data structures, are recorded, stored, or fixed in or on one or more non-transitory computer-readable storage media. Examples of non-transitory computer-readable storage media include read-only memory (ROM), random access memory (RAM), flash memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state drive, and any device known to those skilled in the art capable of storing instructions or software and any associated data, data files, and data structures in a non-transitory manner, and providing instructions or software and any associated data, data files, and data structures to a processor or computer, enabling the processor or computer to execute instructions. In one example, instructions or software, along with any associated data, data files, and data structures, are distributed across a network-coupled computer system, enabling processors or computers to store, access, and execute instructions and software, along with any associated data, data files, and data structures, in a distributed manner.

[0181] Figure Labels

[0182] (1) Color body

[0183] (2) Grayscale

[0184] (3) 3D wavelet coefficient matrix

[0185] (4) 2D grayscale EDF image

[0186] (5) 2D color EDF image

[0187] (6) Cells

[0188] (7) Cell clusters

[0189] (8) Focusing on the object

[0190] (9) Out-of-focus objects

[0191] (10) Red elements

[0192] (11) System

[0193] (12) Input

[0194] (13) Database

[0195] (14) Processor

[0196] (15) User Interface

[0197] (16) Output

Claims

1. A computer-implemented method (M) for generating a color-faithful extended depth-of-field (EDF) image from a chromatograph (1) of a biological sample having dimensions (N, M, L) and values ​​I (n, m, l), wherein voxels (n, m, l) of fixed l from 1 to L are images acquired using a microscope at different depths of focus in the z-direction, the method comprising the steps of: a) Receive the chromatic body (1) and generate a grayscale body (2) by applying a reversible color-to-grayscale transformation (M10) to the chromatic body (1); b) Apply (M20) wavelet transform to each set of N×M voxels of gray volume (2) with the same l index and (n, m) indices from (1, 1) to (N, M) to obtain a 3D wavelet coefficient matrix (3), where the value of each voxel with (n, m, l) index comprises the set of wavelet coefficients. c) For each set of L voxels in a 3D wavelet coefficient matrix (3) with the same (n, m) index and l index from 1 to L, select a set of (M30) wavelet coefficients using a predefined coefficient selection rule. d) Generation (M40): -2D wavelet coefficient matrix WCM, where the value of each pixel with index (n, m) WCM(n, m) is a set of wavelet coefficients selected from the 3D wavelet coefficient matrix (3) according to the coefficient selection rules; -2D coefficient map CM, where the value of each pixel with (n, m) index CM(n, m) is the l index of the voxel of the 3D wavelet coefficient matrix (3) which includes the set of wavelet coefficients selected by the coefficient selection rule; e) Apply the inverse wavelet transform (M50) to the 2D wavelet coefficient matrix WCM to obtain a 2D grayscale EDF image (4). f) Generate a (M60) 2D color composite image CC, where the value of each pixel CC(n,m) with an index of (n,m) is the value of the voxel I(n,m,l) of the chromatic volume (1) whose l index is equal to the value of the 2D coefficient map CM (n,m). g) Apply the inverse transformation of the reversible color-to-grayscale transformation (M70) to the 2D grayscale EDF image (4) to obtain a 2D color EDF image (5); h) Convert the 2D color composite image CC and the 2D color EDF image (5) (M80) to a color space that includes at least one chromaticity component and at least one intensity component; i) Connect (M90) the at least one chromaticity component of the 2D color composite image CC and the at least one intensity component of the 2D color EDF image (5) to obtain a color-faithful extended depth of field (EDF) image.

2. The method according to claim 1, wherein, In selecting a set of wavelet coefficients (M30) using the predefined coefficient selection rule, a first subset of the wavelet coefficients is selected using the first predefined coefficient selection rule, and a second subset of the wavelet coefficients is selected using the second predefined coefficient selection rule; and The steps for generating the (M40) 2D wavelet coefficient matrix WCM and the 2D coefficient graph CM include: - Generate a first 2D coefficient map (M41) from the first subset of the wavelet coefficients; - Generate a (M42) second 2D coefficient map from the second subset of the wavelet coefficients; - Combine (M43) the first and second 2D coefficient maps to obtain a single 2D coefficient map CM. - Generate a (M44) 2D wavelet coefficient matrix WCM, where the value of each pixel WCM(n,m) is the value of a voxel with an index of (n,m) in the 3D wavelet coefficient matrix and an index equal to the value of the obtained 2D coefficient map CM (n,m).

3. The method of claim 2, wherein the step (M43) of combining the first and second 2D coefficient maps comprises: - Filter the first 2D coefficient map and the second 2D coefficient map; -Average the first and second 2D coefficient plots; - Round the average 2D coefficient plot to obtain a 2D coefficient plot CM.

4. The method according to claim 1, wherein, The set of wavelet coefficients includes at least four wavelet coefficients, and a subset of wavelet coefficients includes at least one wavelet coefficient.

5. The method according to claim 1, wherein, The color-faithful extended depth-of-field (EDF) image is converted to a different color space.

6. The method according to claim 1, wherein, The reversible color-to-grayscale transformation is principal component analysis (PCA).

7. The method according to claim 1, wherein, The wavelet transform used in the application step (M20) is the stationary wavelet transform (SWT).

8. The method according to claim 1, wherein, The chromosome (1) is a biomedical image, and the method further includes the step of segmenting the color-faithful extended depth-of-field (EDF) image.

9. The method according to claim 3, wherein, The first and second 2D coefficient maps are filtered using a median filter.

10. The method according to claim 5, wherein, The color-faithful extended depth-of-field EDF image is converted to the color space of the color volume (1).

11. A system (11) for analyzing biological samples, said system (11) comprising: Suitable for receiving at least one input (12) of a color body (1); At least one processor (14) is configured to: - A grayscale volume (2) is generated by applying a reversible color-to-grayscale transformation (M10) to the color volume (1); - Apply (M20) wavelet transform to each set of N×M voxels of gray volume (2) with the same l index and (n, m) indices from (1, 1) to (N, M) to obtain a 3D wavelet coefficient matrix (3), where the value of each voxel with (n, m, l) index comprises the set of wavelet coefficients; - For each set of L voxels in a 3D wavelet coefficient matrix (3) with the same (n, m) index and l index from 1 to L, select a set of (M30) wavelet coefficients using a predefined coefficient selection rule; -Generate (M40): The o2D wavelet coefficient matrix WCM, where the value of each pixel with an index of (n, m) WCM(n, m) is a set of wavelet coefficients selected from the 3D wavelet coefficient matrix (3) according to the coefficient selection rules; o2D coefficient map CM, where the value of each pixel with (n, m) index CM(n, m) is the l index of the voxel of the 3D wavelet coefficient matrix (3) which includes the set of wavelet coefficients selected by the coefficient selection rule; - Apply the inverse wavelet transform (M50) to the 2D wavelet coefficient matrix WCM to obtain a 2D grayscale EDF image (4). - Generate a (M60) 2D color composite image CC, where the value of each pixel CC(n,m) with an index of (n,m) is the value of the voxel I(n,m,l) of the chromatic volume (1) whose l index is equal to the value of the 2D coefficient map CM (n,m). - Apply the inverse transformation of the reversible color-to-grayscale transformation (M70) to the 2D grayscale EDF image (4) to obtain a 2D color EDF image (5); - Convert the 2D color composite image CC and the 2D color EDF image (5) (M80) to a color space that includes at least one chromaticity component and at least one intensity component; - Connect at least one chromaticity component of the 2D color composite image CC and at least one intensity component of the 2D color EDF image (5) (M90) to obtain a color-faithful extended depth-of-field (EDF) image; and At least one output (16) is adapted to provide a color-faithful extended depth-of-field (EDF) image.

12. The system (11) according to claim 11, wherein, The biological sample is a urine sample, and the system includes an automated digital cytology system configured to acquire a chromatograph (1) comprising at least two 2D color images.

13. A bladder cancer detection device, characterized in that... The device includes the system according to claim 11.

14. A computer program product for analyzing biological samples, the computer program product comprising instructions that, when executed by a computer, cause the computer to perform the steps of the method according to claim 1.

15. A non-transitory computer-readable storage medium comprising instructions that, when executed by a computer, cause the computer to perform the steps of the method according to claim 1.

Images