Cell relocation method and device under multiple different objectives and application thereof

Abstract

The application provides a cell relocation method and device under multiple different objective lenses and application thereof, and comprises the following steps: S000, single-channel image acquisition is performed on a slide under a first set magnification objective lens, and initial screening is performed on target cells in the image to obtain the positions of the target cells; S100, based on the positions of the target cells, the target cells in the image are relocated through advantage feature point matching of a feature descriptor under a second set magnification objective lens which is larger than the first set magnification; and S200, after relocation, multi-channel pictures are acquired under the second set magnification objective lens to analyze the target cells. The application can more accurately and quickly position the target cells.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image recognition, in particular to a cell relocation method and device under multiple different objectives and application thereof. BACKGROUND

[0002] In non-blood source cell typing detection, the proportion of non-target cells is small, and the proportion of white blood cells is large; after grouping by the product, the non-blood source cells are analyzed for fluorescence signal atlas, so a more automated way is needed to position the target cells for analysis. In the fluorescence scene, the cells are sparse and similar in shape; people generally use multiple references to accurately locate a position, and microscopes also use multiple references to locate the position of an object. The microscopic field is larger than the macroscopic field, and the time-consuming and error-prone search method for the cell area to be relocated is used.

[0003] Currently, there is an accurate positioning of target cells based on the collected pictures, and then multiple channel cell images are collected at different focal planes of the target cells for cell hybrid signal analysis, but there are the following technical problems:

[0004] 1. Morphological difference: under high magnification, the morphological details of cells are more, and the collected pictures may not be consistent.

[0005] 2. Scale difference caused by focusing: the motor is affected by movement, and the focusing position of each field of view is different, which may cause the relative position of the target cells in the field of view to be different, and the relative position that has been preliminarily determined may not be completely the same ratio through calculation.

[0006] 3. Exposure difference: due to impurities in the slide, too large field of view or camera imaging itself, the focusing result is affected, there are multiple target cells in a field of view, which causes the exposure of some target cells under 20X objective to be very different from that under 40X, and a simple way may cause some cells to be unable to be positioned.

[0007] 4. The density of cells in the slide is not the same, different cells may be very similar or the same cells in shape under low magnification, and the shape of the same cells under high magnification is not the same, and the target cell template matching processing may select multiple cells, select wrong cells or select no cells. The relocation of the selected cells needs to be accurately positioned.

[0008] Therefore, a cell relocation method and device under multiple different objectives and application thereof are needed to solve the above problems. SUMMARY

[0009] The embodiments of the present application provide a cell relocation method and device under multiple different objectives and application thereof, which solve the problems of the prior art.

[0010] The core technology of the present application is to reposition the target cell in the image through the matching of the advantage feature points of the feature descriptor.

[0011] In a first aspect, the present application provides a method for repositioning cells photographed under multiple different objectives, which comprises the following steps:

[0012] S000, under a first set magnification objective, single-channel image acquisition is performed on a slide, and the target cells in the image are grouped and preliminarily screened to obtain the position of the target cells;

[0013] S100, based on the position of the target cells, the target cells in the image are repositioned under a second set magnification objective which is larger than the first set magnification objective through the matching of the advantage feature points of the feature descriptor;

[0014] S200, after repositioning, multi-channel pictures are acquired under the second set magnification objective for analysis of the target cells.

[0015] Further, before repositioning in the step S100, the following steps are further included:

[0016] S110, the image acquired under the second set magnification objective is scaled to the same scale under the first set magnification to obtain a scaled image;

[0017] S120, the region of interest of the target cells in the single-channel image acquired under the first set magnification objective is expanded so that the region to be repositioned is substituted into more environmental reference information of the region of interest;

[0018] S130, the target cells in the scaled image are repositioned through the matching of the advantage feature points of the feature descriptor.

[0019] Further, in the step S100, the specific steps of repositioning include:

[0020] S131, the feature information of the scaled image and the single-channel image after the expansion of the region of interest is calculated respectively;

[0021] S132, the preliminary matching key points of the scaled image and the single-channel image are obtained through the key point feature descriptor;

[0022] S133, the samples of the matching key points are screened to exclude the mis-matching key point pairs;

[0023] S134, the points that cannot be processed by the above steps are processed, and the region containing more reference information is selected as a template image to calculate the coordinates of each point of the scaled image corresponding to the image under the second set magnification.

[0024] Further, in the step S133, the specific steps of screening are:

[0025] S1331, normalize the coordinate values of the matched key point pairs and store as index pairs;

[0026] S1332, map the cropped scaled image and the single channel image after expanding the region of interest into a set grid according to the scale of each image to generate key point clustering phenomenon;

[0027] S1333, convert the two-dimensional coordinate values of the set grid into one-dimensional index values to construct an index relationship R graph and map the index pairs into R;

[0028] Wherein, the abscissa x of R is defined as the source matching subscript index, and the ordinate y is the target matching index;

[0029] S1334, select the maximum value of each row of R in the index relationship R graph as the correct matching value, and the rest as the false matching;

[0030] S1335, map the key points of the single channel image after expanding the region of interest to the set network, then add a set value to the mapped abscissa and ordinate respectively, then combine them, and repeat steps S1333-S1334 to get the final result. The index subscript corresponding to the final result is the correctly matched key point;

[0031] S1336, only take the feature points in the cell region to be relocated in the single channel image under the first set magnification as the coordinate values in the new coordinates after relocation.

[0032] Further, in step S134, the specific steps of processing are:

[0033] S1341, downsample the scaled image and the single channel image under the first set magnification simultaneously;

[0034] S1342, block the scaled image after downsampling multiple times, and calculate the center point of the block appearing in the downsampled single channel image;

[0035] S1343, cluster the center points to select the largest class as the new center point, and calculate the offset of the target cell in the downsampled single channel image from the new center point;

[0036] S1344, calculate the corresponding coordinates of the target cell in the image under the second set magnification through the offset.

[0037] Further, the specific steps of processing further include:

[0038] S1345, enlarge the region of interest as a template picture, and collect the average gray scale and standard deviation of the region where the template picture is located to perform gray scale statistics on the template picture, and proportionally increase the gray scale value;

[0039] S1346, according to the field of view range of the target cell in the second set magnification after repositioning, take the region above the center of the field of view as a search area image;

[0040] S1347, perform template matching after keeping the gray scale values of the search area image and the template picture at the same level;

[0041] S1348, select a point with a set matching threshold as a matching position, and calculate the new coordinate value of the image under the second set magnification objective; if the set matching threshold is not met, the region containing the center point is taken as the matching position, and the new coordinate value of the image under the second set magnification objective is calculated.

[0042] Further, in the S131 step, the feature information of the scaled image and the single-channel image after the region of interest is enlarged is calculated respectively by the ORB algorithm; in the S132 step, the key point feature descriptor obtains the preliminary matching key point by the nearest neighbor method.

[0043] In a second aspect, the present application provides a cell repositioning device for multiple different objectives, comprising:

[0044] The acquisition module is configured to acquire a single-channel image of the slide under a first set magnification objective, and perform grouping and preliminary screening on the target cells in the image to obtain the positions of the target cells.

[0045] The repositioning module is configured to reposition the target cells in the image under a second set magnification objective greater than the first set magnification objective based on the positions of the target cells by matching the advantage feature points of the feature descriptors.

[0046] The analysis module is configured to acquire a multi-channel image under the second set magnification objective after repositioning to analyze the target cells.

[0047] In a third aspect, the present application provides an electronic device, comprising a memory and a processor, the memory stores a computer program, and the processor is configured to run the computer program to execute the cell repositioning method under multiple different objectives described above.

[0048] In a fourth aspect, the present application provides a readable storage medium, the readable storage medium stores a computer program, the computer program comprises program code for controlling a process to execute the process, and the process comprises the cell repositioning method under multiple different objectives described above.

[0049] The main contribution and innovation of the present application are as follows: 1. Compared with the prior art, the present application can ignore more factors such as gray level, whether the cell is at the edge or corner, by matching the advantage feature points of the feature descriptor to reposition the target cell in the image, so as to significantly improve the repositioning accuracy;

[0050] 2. Compared with the prior art (simply performing window template matching on the cell needing repositioning in a small area, selecting the coordinate with the highest similarity as the repositioned coordinate, and selecting the coordinate with the highest similarity as the repositioned coordinate), the present application can re-match the points that cannot extract feature points, mismatch or have too large deviation center points, thereby greatly improving the repositioning accuracy. Higher accuracy of repositioning can reduce manual positioning through various microscopic reference information, and when cells are automatically scanned, the repositioned cells can be focused, collected and analyzed, which can not only exclude the influence of environmental impurities on focusing, but also reduce the analysis of all cells in the entire field of view to improve the analysis efficiency.

[0051] The details of one or more embodiments of the present application are presented in the following drawings and description to make other features, objects and advantages of the present application more apparent. BRIEF DESCRIPTION OF DRAWINGS

[0052] The drawings described herein are intended to provide further understanding of the present application, and constitute a part of the present application. The illustrative embodiments of the present application and their description serve to explain the present application, and do not constitute an improper limitation of the present application. In the drawings:

[0053] Figure 1 is a flowchart of a cell repositioning method under multiple different objectives according to an embodiment of the present application;

[0054] Figure 2 is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0055] The exemplary embodiments will be described in detail herein with reference to the drawings. Unless otherwise defined, the same numbers in different drawings indicate the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with one or more embodiments of the present specification. Instead, they are merely examples of devices and methods consistent with some aspects of one or more embodiments of the present specification, as detailed in the appended claims.

[0056] It should be noted that the steps of the corresponding method are not necessarily performed in the order shown and described in the present specification in other embodiments. In some other embodiments, the steps included in the method thereof can be more or less than described in the present specification. In addition, a single step described in the present specification can be divided into multiple steps for description in other embodiments, and multiple steps described in the present specification can be combined into a single step for description in other embodiments.

[0057] 20X objective lens collects images for initial screening to select target cells, and the target cells are analyzed under 40X objective lens positioning cell position under 40X objective lens after providing preliminary reference under 20X. Since the relative position changes each time, the pictures collected by 20X to 40X and 40X to 40X objective lenses need to be repositioned. A relatively simple and effective way is to simply perform window template matching on the cells that need to be repositioned in a small area, and select the highest similarity coordinates as the repositioned coordinates. This method simply does not meet the needs in the fluorescence scene. Moreover, the image acquisition and motor cannot be moved to the ideal position completely accurately, and a small coordinate difference under fluorescence microscopy will cause the cell to be detected to be different from the target cell.

[0058] Based on this, the present application is based on the advantage of feature point matching of feature descriptor for repositioning the target cell in the image.

[0059] Embodiment one

[0060] The present application aims to provide a cell repositioning method under multiple different objective lenses, specifically referring to Figure 1 , the method comprises the following steps:

[0061] S000, under a first set magnification objective lens, single-channel image acquisition is performed on the slide, and the target cells in the image are grouped and initially screened to obtain the position of the target cells;

[0062] In the present embodiment, single-channel image acquisition is performed on the slide under 20X objective lens, and the target cells are grouped and initially screened by algorithm.

[0063] The grouping and initial screening is prior art, and the cell extraction is mainly reflected in the segmentation of the background and the foreground, and the purpose is to select the cells according to the gray value, the circularity, the aspect ratio and other values through a certain threshold according to the needs of the present application; and the selected cells are grouped according to the calculated radius size. The grouping method makes it easier to view and manually intervene on the software interface. The present application does not process the situation of cell adhesion at present, so it is excluded during preprocessing. Therefore, the grouping and initial screening and the segmentation of the background and the foreground are prior art and are not the technical points to be protected by the present application, and therefore will not be described here.

[0064] S100, based on the location of the target cell, the target cell in the image is repositioned by the dominant feature point matching of the feature descriptor under the second set magnification objective larger than the first set magnification;

[0065] In this embodiment, the selected target is repositioned by collecting pictures under the 40X objective.

[0066] S110, the 40X collected image (denoted as S) is scaled to the same scale under 20X (denoted as S40X). The ratio of the magnifying lens can be simply taken as the scaling value, and the slight scale difference caused by the incomplete consistency of focusing is ignored.

[0067] S120, after the processing of S110, the roi (Region of Interest) area of the target cell in the 20X image is expanded to bring the repositioning area into more environmental reference information of the roi area. The image after expanding the roi is denoted as Sroi.

[0068] Among them, the feature point refers to the point where the image gray value changes sharply or the point with large curvature on the image edge (i.e. the intersection of two edges) in image processing. It is often used in artificial intelligence description.

[0069] The key point contains information other than the coordinate value of the feature point (such as the diameter, direction, category to which the feature point belongs, and the strength of the feature point).

[0070] The key point is a kind of feature point. In this paper, opencv is used to process the image, and the KeyPoint and the feature descriptor corresponding to the KeyPoint are extracted by ORB calculation.

[0071] The purpose of this step is to increase the environmental information of feature matching and prevent the non-uniqueness of repositioning caused by the similarity between cells. At the same time, this step is equivalent to the cropping process.

[0072] S130, since more information such as gray scale and whether the cell is on the edge or corner can be ignored by feature point matching, the feature point matching of the feature descriptor is preferentially used for cell repositioning. The specific steps are as follows:

[0073] S131, calculate the feature information of the image: compared with the advantages of various feature extraction, the ORB algorithm is selected to calculate the feature information of Sroi and S40X.

[0074] S132, acquire matched feature points: and get through feature k-neighbor calculation, key point feature descriptor gets the initial matching key point through the nearest neighbor method; key point screening is crucial, and the method of simply filtering matching points through feature point distance cannot guarantee that each matched key point is in correct relative position;

[0075] The purpose of this step is to ensure that the relative positions of the matched feature points remain correct.

[0076] S133, sample screening of feature key points: the matched key points have a large number, are scattered, and there are many false matches, so false key point pairs need to be excluded; the specific screening steps are as follows:

[0077] S1331, normalize the coordinate values (x, y) of the matched key points to the [0, 1] interval, and store the index pair of the matched feature points;

[0078] S1332, map the image key point coordinates of the cropped Sroi and S40X to a 20*20 grid according to the scale of each image, to generate key point aggregation phenomenon;

[0079] Among them, the aggregation phenomenon of key points is formed by high-resolution scale transformation to low-resolution (for example, 0-19 originally contains 20 numbers, after mapping y=x / 5, the remapped y corresponds to 4 numbers in the original set to form aggregation). Since the similarity of cells under low magnification is very high, only considering the roi region may lead to non-unique matching region and fail to achieve repositioning effect; the present application reduces the left upper corner coordinates in the x and y directions by 300 pixels in the roi region, and increases the left lower corner coordinates by 300 pixels, while ensuring that the Sroi does not exceed the image boundary, and in an ideal case, the cropped region is increased by 600 pixels in width and height compared to the roi region. This processing method ensures sufficient reference information in the feature matching process, and reduces the calculation of part of the meaningless feature points to improve the code running speed.

[0080] In this way, the relative positional relationship between points is kept unchanged; after mapping to the grid, the size of the key point index relationship graph is reduced, thereby reducing the calculation amount of key point screening; and the formed aggregation is more easy to exclude the outlier false matching key point pairs.

[0081] S1333, convert the 20*20 grid two-dimensional coordinate values into lower 1*400 index values, and construct a 400*400 source-target index relationship R graph (wherein the horizontal coordinate x of R is defined as the matching lower index of the source, and the vertical coordinate y is defined as the matching index of the target), and the index pair of the matched Sroi and S40X key points is mapped into R.

[0082] Wherein, the row and column of two-dimensional coordinates are multiplied by 20 respectively, and then they are added up, that is, the corresponding one-dimensional index value can be obtained. The index relationship R graph is a two-dimensional array, wherein each element (i, j) represents whether there is a matching relationship between the source index i and the target index j. If there is a matching relationship, the value of the (i, j) element is added by 1, otherwise it is 0. Since there are 400 source indexes and 400 target indexes. After the knn key point matching, the matching key points also have the problem of scattered matching. The image is scaled into a 20*20 grid from a higher resolution, and the original different coordinate points may fall on the same point after scaling, but the original relative relationship is maintained, and the matching feature points are from accurate to generalization. Since the coordinates of the source 20*20 key points are (1, 2), which match the target image 20*20 point (4, 5), they have a corresponding relationship in the index relationship graph (22, 85), and this index pair is added by 1. Assuming that the key points extracted from the source may match any key points in the target image, and in fact, the scattered matching will exist. This phenomenon forms a 400*400 index relationship graph.

[0083] S1334, since the matching points have been calculated, the mismatched points are likely to be some outliers, and the correct matching is approximately a classic experiment in probability--needle experiment; That is, the maximum value of each row of R in step iii is most likely to be the correct matching value, and other matching pairs are considered to be mismatched;

[0084] S1335, S40X is the image region size intercepted by the searched object, which must be larger than Sroi; and the 20*20 grid formed by Sroi has a certain flexible variable range, avoiding the omission of some correctly matched key points, so the image Sroi key points are mapped to the 20*20 grid, and then the mapped coordinate values x and y are added respectively{{0, 0.5}, {0.5, 0}, {0.5, 0.5}} and combined, and the final result corresponding to the index subscript of the key points is obtained by repeating the steps of S1333~S1334.

[0085] Wherein, since the true coordinate value is normalized, that is, the value range of the coordinate value x, y is changed to [0, 1], and the lower integer mapping to the integer space of [0, 20] is 1, and it itself may be mapped to 2. In order to eliminate the influence of such cases, a certain offset value is added to the coordinate value before the lower integer is taken; and the index position after the flexible processing of the coordinate value is closer to the actual value. In the formed index relationship graph, the index pair with the maximum value in each row is taken as the matching key point pair; the distance and other filtered matching key points obtained by the feature descriptor knn have a large scattered matching, which may interfere with the repositioning.

[0086] S1336, only take the key points in the 20X cell area to be relocated, and the distance calculated by the knn descriptor of the taken key points is less than a certain threshold, and then take the feature matching point with the minimum distance calculated by the knn descriptor from the remaining key points as, calculate the new coordinates of the field of view under the new magnification.

[0087] Wherein, this step is the last step of key point relocation after screening key points, and a more accurate judgment is made on the matched key points. When there is no matched key point in the roi, the matched key point outside the roi has little reference significance.

[0088] S134, points that cannot extract feature points, mismatch or have large deviation center points are processed in S134 step, and a larger area is selected as a template image to contain more reference information; the specific steps are as follows:

[0089] S1341, downsampling: 20X and S40X are simultaneously downsampled to obtain 20X' and S40X', the background is not sensitive to the downsampling, which eliminates the influence of the background on the matching and maintains the complete reference information of the image, and the image processing speed is improved;

[0090] S1342, after downsampling, S40X' is divided into 4 blocks and 16 blocks in turn (the advantage of this block division method is that the center point of the image is easily determined, which is beneficial to the coordinate calculation of the original image in the new image, so that the relative position of the new coordinate is easily calculated), and the small blocks after the block division appear in 20X'. The center point is calculated by block division, the largest class is selected as the new center point by clustering the center points, and the offset of the target cell from the center point in 20X' is calculated, that is, the offset of the center point of S40X' under S40X' is calculated, so as to calculate the corresponding coordinates of the target cell under 40X;

[0091] S1343, as a supplementary scheme of S1342 step, for cells in the edge corners, the information is sparse, and the roi area has low gray scale. A certain range is expanded as a template picture T, the average gray scale and standard deviation of the T region are collected, and the gray scale of T is statistically improved in proportion;

[0092] According to the field of view range of the 40X target cell according to the relocation scheme, the area above the center of the field of view is taken as a search area image F, and the gray scale of the search image F and the template image T is kept to the same level before template matching;

[0093] Select a point with a certain matching threshold as a matching position to calculate the new coordinate value of the image under the 40X objective lens; if a certain matching threshold is not met, the center point and the region containing the center point are selected as F', and a point with a certain matching threshold is selected as a matching position.

[0094] This step is used as a final supplement for the case where the matching is successful, but the imaging quality is poor due to image reasons, and the matching coordinate value is taken as the matching result under the current threshold.

[0095] S200, after repositioning, a multi-channel image is collected under a second set magnification objective lens for analysis of the target cell.

[0096] In this embodiment, the fluorescence signal spectrum analysis of the multi-channel image collected under the 40X objective lens is performed. Thus, there is no problem that the desired cell to be detected is not the same cell as the target cell due to the small coordinate difference in the fluorescence microscope caused by the image collection and the motor that cannot move to the ideal position accurately. Moreover, the more detailed extraction of the instrument can achieve unattended operation; meanwhile, in the case where the approximate position can be located by the scanning coordinate transformation of the repositioning of the slide for manual review, the position of the target cell provides great convenience.

[0097] In this embodiment, the image is processed to the same scale in step S110;

[0098] In step S120, the feature point matching is performed, the image gray difference is ignored, the 40X image coordinates are repositioned by the mesh screening of the matching key points, and the relative coordinates of the machine are used for repositioning.

[0099] In step S134, due to the gray difference of the target cell morphology during the image collection process, more reference information is compared for repositioning, and the block voting method is used to solve the problem; in step S1343, the target is in the image, and the difference between the source and the target is large in the gray correction to the same gray level, and the size of the searched box is adjusted to reduce the gray and morphological changes of the target cell, so as to solve the repositioning problem.

[0100] Thus, the key point matching, the image block, the center point voting, and the supplementary matching of the source and target images in the specified small area are combined. The higher accuracy of the repositioning can reduce the manual positioning of the reference information under various microscopic conditions, and the cells can be focused, collected, and analyzed in the automatic scanning, so that the influence of the environmental impurities on the focusing can be excluded, and the analysis efficiency can be improved by analyzing all the cells in the entire field of view.

[0101] Embodiment Two

[0102] Based on the same concept, the application further provides a cell repositioning device for multiple different objective lenses, comprising:

[0103] The acquisition module is configured to collect a single-channel image of the slide under a first set magnification objective lens, and perform grouping and preliminary screening on the target cells in the image to obtain the position of the target cells.

[0104] a repositioning module, configured to reposition the target cell in the image under a second set magnification objective larger than the first set magnification, based on the position of the target cell, by matching the feature descriptor with the dominant feature point;

[0105] an analysis module, configured to collect multi-channel pictures under the second set magnification objective after the repositioning, to analyze the target cell.

[0106] Embodiment Three

[0107] The embodiment also provides an electronic device, referring to Figure 2 , comprising a memory 404 and a processor 402, the memory 404 storing a computer program, and the processor 402 being configured to run the computer program to perform the steps in any of the above method embodiments.

[0108] Specifically, the processor 402 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement one or more embodiments of the application.

[0109] The memory 404 can include mass storage for data or instructions. By way of example, and not limitation, the memory 404 can include a Hard Disk Drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, a USB drive, a magneto-optical disk, optical disks such as CD-ROMs or DVDs, magnetic tapes, or other types of storage media, which can be removable or non-removable (or fixed). The memory 404 can include a combination of two or more of these types of storage media. In suitable cases, the memory 404 can be internal or external to the data processing apparatus. In particular embodiments, the memory 404 is a Non-Volatile memory. In particular embodiments, the memory 404 includes Read-Only Memory (ROM) and Random-Access Memory (RAM). In suitable cases, the ROM can be a mask-programmed ROM, a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically EPROM (EEPROM), an Electrically Alterable ROM (EAROM), or a FLASH, or a combination of two or more of these. In suitable cases, the RAM can be a Static Random-Access Memory (SRAM) or a Dynamic Random-Access Memory (DRAM), which can be a Fast Page Mode Dynamic Random-Access Memory (FPMDRAM), an Extended Data Out Dynamic Random-Access Memory (EDODRAM), a Synchronous Dynamic Random-Access Memory (SDRAM), or the like.

[0110] The memory 404 can be used to store or buffer various data files needed for processing and / or communication, and possible computer program instructions executed by the processor 402.

[0111] The processor 402 can implement the cell relocation method under multiple different objectives by reading and executing the computer program instructions stored in the memory 404.

[0112] Optionally, the electronic device can further include a transmission device 406 connected to the processor 402 and an input / output device 408 connected to the processor 402.

[0113] The transmission device 406 can be used to receive or send data via a network. The network can include a wired or wireless network provided by a communication provider of the electronic device. In one example, the transmission device includes a network adapter (NIC) that can be connected to other network devices through a base station to communicate with the Internet. In one example, the transmission device 406 can be a radio frequency (RF) module for communicating with the Internet through wireless means.

[0114] The input / output device 408 is used to input or output information. In the present embodiment, the input information can be a single-channel image under a 20X objective, and the output information can be a fluorescence signal spectrum analysis result.

[0115] Embodiment Four

[0116] The present embodiment also provides a readable storage medium having a computer program stored therein, the computer program including program codes for controlling a process to execute the process, the process including the cell relocation method under multiple different objectives according to the first embodiment.

[0117] It should be noted that the specific examples in the present embodiment can refer to the examples described in the above embodiments and optional implementation manners, which will not be described herein again.

[0118] In general, the various embodiments can be implemented in hardware or special purpose circuits, software, logic or any combination thereof. Some aspects of the application can be implemented in hardware, while other aspects can be implemented in firmware or software which can be executed by a controller, microprocessor or other computing device, but the application is not limited thereto. While various aspects of the application can be illustrated and described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, techniques or methods described herein can be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controler or other computing devices, or some combination thereof.

[0119] Embodiments of the application can be implemented by computer software executable by a data processor of the mobile device such as in the processor entity, or by hardware, or by a combination of software and hardware. Computer software or program, also called program product, including software routines, applets and / or macros, can be stored in any apparatus-readable data storage medium and they comprise program instructions to perform particular tasks. The program product can include one or more computer-executable components such as by the program instructions of one or more applications, or portions thereof. The one or more computer-executable components can be one or more of initial applications, utility programs, background services, system services, program libraries, and so on.The one or more computer-executable components can be stored on the physical media as physical media computer instructions, or loaded into memory from another computer readable medium. The one or more computer-executable components can include, but are not limited to, software for implementation of the embodiments. The program instructions can be implemented in a high level procedural or object oriented programming language, or in assembly or machine language. The software can be

[0120] It should be apparent to those skilled in the art that the above-described embodiments of the present application can be modified in various ways without departing from the scope of the present application. Accordingly, the scope of the present application should be determined by the following claims.

[0121] The above embodiments are only some of the several embodiments of the present application, and the description is specific and detailed, but it should not be understood as limiting the scope of the present application. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the scope of the present application should be subject to the appended claims.

Claims

1. A method for cell relocalization using multiple images under different objectives, for non-blood-derived cell typing detection, characterized in that... Includes the following steps: S000: Under the first set magnification objective lens, a single-channel image is acquired on the slide, and the target cells in the image are grouped and pre-screened to obtain the location of the target cells; S100. Based on the location of the target cells, the image acquired under the second set magnification objective lens, which is larger than the first set magnification, is scaled down to the same scale as the first set magnification to obtain a scaled image. Expand the region of interest (ROI) of the target cell in the single-channel image acquired under the first set magnification objective lens so that the region to be relocalized can incorporate more environmental reference information of the ROI. Target cells in the scaled image are relocated by matching the dominant feature points of the feature descriptor; After repositioning (S200), multi-channel images are acquired under the second set magnification objective lens for target cell analysis.

2. The method for cell repositioning under multiple different objectives as described in claim 1, characterized in that, In step S100, the specific steps for relocation include: S131. Calculate the feature information of the scaled image and the single-channel image after expanding the region of interest, respectively. S132. Obtain preliminary matching key points between the scaled image and the single-channel image through key point feature descriptors; S133. Filter the samples that match key points to exclude false key point pairs; S134. Process the points that cannot be processed in the above steps, and select an area containing more reference information as a template image to calculate the coordinates of each point in the scaled image corresponding to the image at the second set magnification.

3. The method for cell repositioning under multiple different objectives as described in claim 2, characterized in that, In step S133, the specific steps for screening are as follows: S1331. Normalize the coordinate values ​​of the matched keypoint pairs and store them as index pairs; S1332. Map the cropped scaled image and the single-channel image after expanding the region of interest to a set grid according to the scale of their respective images to produce key point clustering. S1333. Convert the two-dimensional coordinate values ​​of the set grid into one-dimensional index values ​​to construct an index relationship R diagram and map the index pairs into R; Here, the x-coordinate of R is defined as the matching index of the source, and the y-coordinate is defined as the matching index of the target. S1334. Select the maximum value of each row in the index relationship R diagram as the correct match value, and the rest are incorrect matches; S1335. After mapping the key points of the single-channel image after expanding the region of interest to the set network, add the set values ​​to the mapped horizontal and vertical coordinates respectively and then combine them. Repeat steps S1333~1334 to obtain the final result. The index subscript corresponding to the final result is the key point that is correctly matched. S1336. Select only the key points in the cell region to be relocated in the single-channel image at the first set magnification, and the distance calculated by the knn descriptor of the selected key points is less than the set threshold. Then, select the feature matching point with the smallest distance calculated by the knn descriptor from the remaining key points, and use the feature matching point as the coordinate value in the new coordinates after relocation.

4. The method for cell repositioning under multiple different objectives as described in claim 2, characterized in that, In step S134, the specific processing steps are as follows: S1341. Simultaneously downsample the scaled image and the single-channel image at the first set magnification. S1342. Divide the downsampled scaled image into multiple blocks and calculate the center point of each block that appears in the downsampled single-channel image. S1343. Cluster the center points, select the largest cluster as the new center point, and calculate the offset of the target cell from the new center point in the downsampled single-channel image. S1344. Calculate the corresponding coordinates of the target cell in the image at the second set magnification using this offset.

5. The method for cell repositioning under multiple different objectives as described in claim 4, characterized in that, The specific steps of the process also include: S1345. Expand the region of interest as a template image, and collect the average gray level and standard deviation of the region where the template image is located to perform gray level statistics on the template image and increase the gray level value proportionally. S1346. Based on the field of view where the target cell is located in the second set magnification after repositioning, take the area slightly above the center of the field of view as the search area image. S1347. After keeping the grayscale values ​​of the search area image and the template image at the same level, perform template matching; S1348. Select a point with a set matching threshold as the matching position, and calculate the new coordinate value of the matching position as the image under the second set magnification objective lens; if the set matching threshold is not met, the area including the center point is used as the matching position, and the matching position is calculated as the new coordinate value of the matching position as the image under the second set magnification objective lens.

6. The method for cell repositioning under multiple different objectives as described in any one of claims 2-5, characterized in that, In step S131, the feature information of the scaled image and the single-channel image after expanding the region of interest is calculated using the ORB algorithm; in step S132, the key point feature descriptors are used to obtain preliminary matching key points through the nearest neighbor method.

7. A device for repositioning cells under multiple different objectives, characterized in that, include: The acquisition module is used to acquire a single-channel image of the slide under a first set magnification objective lens, and to perform preliminary screening of target cells in the image by grouping them to obtain the location of the target cells; The repositioning module is used to scale the image acquired under the second set magnification objective lens (which is larger than the first set magnification) to the same scale as the first set magnification, based on the position of the target cell, to obtain a scaled image. Expand the region of interest (ROI) of the target cell in the single-channel image acquired under the first set magnification objective lens so that the region to be relocalized can incorporate more environmental reference information of the ROI. Target cells in the scaled image are relocated by matching the dominant feature points of the feature descriptor; The analysis module is used to acquire multi-channel images under a second set magnification objective lens after repositioning for the analysis of target cells.

8. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to perform the cell repositioning method under multiple different objectives as described in any one of claims 1 to 6.

9. A readable storage medium, characterized in that, The readable storage medium stores a computer program, the computer program including program code for controlling a process to execute the process, the process including the cell repositioning method under multiple different objectives as described in any one of claims 1 to 6.

Images