General IVD detection method and system based on deep learning

Abstract

The invention belongs to the field of IVD image processing, and discloses a general IVD detection method and system based on deep learning. The application process of the visual algorithm comprises the following steps of: performing algorithm self-learning parameter configuration, image preprocessing, an algorithm core part and an algorithm post-processing part, and realizing optimal setting of hardware configuration and algorithm output through parameter self-learning. Meanwhile, various IVD detection scenes including tube judgment, sampling needle detection, blood bag detection and the like are integrated, multifunctional detection is achieved through one set of detection system, meanwhile, combined arrangement of algorithm processing can be achieved, and accurate and clear detection results can be automatically output.

Description

Technical Field

[0001] The present invention relates to the field of IVD image recognition, and in particular to a general IVD detection method and system based on deep learning. Background Art

[0002] In vitro diagnosis, or IVD (In Vitro Diagnosis), refers to a diagnostic method that obtains clinical diagnostic information by testing samples such as human body fluids, cells and tissues in vitro, and then determines the disease or body function. It plays an important role in disease prevention, diagnosis, and treatment. Currently, more than 80% of clinical disease diagnosis can be completed by IVD. It includes sample pre-treatment, multi-row rapid sample injection, multi-turn turntable sample barcode high-speed reading, etc., which are applied to automated test lines, test tube sorting, blood bag management, coagulation, immunity, urine, biochemistry, luminescence platforms, etc.

[0003] The IVD vision industry plays an important role in the field of medical diagnosis. Traditional hand detection methods are labor-intensive, time-consuming, and prone to human errors. Existing OCR recognition is not very accurate in special IVD scenarios. Some machine vision-based solutions have more interference information and less feature information when performing full-image classification, and the model size is large and time-consuming. In addition, the detection methods of existing technologies are limited to a single scene, and switching scenes requires different equipment and algorithms. Summary of the invention

[0004] The purpose of the present invention is to provide a universal IVD detection method and system based on deep learning.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The general IVD detection method based on deep learning includes the following steps:

[0007] S1: Configure self-learning model and parameters;

[0008] S2: Execute hardware parameter self-learning;

[0009] S3: Execute algorithm parameter self-learning;

[0010] S4: input the image to be detected;

[0011] S5: Select the corresponding IVD scenario algorithm and perform the test;

[0012] S6: Execute algorithm post-processing and output the detection results.

[0013] In step S2, the hardware parameters are self-learned, including autofocus and automatic adjustment of exposure parameters.

[0014] In step S3: the algorithm's parameter self-learning is specifically as follows:

[0015] S31: Select an IVD scene and complete the autofocus of the device through autofocus and autoexposure algorithms;

[0016] S32: Setting the scoring algorithm and preset parameters corresponding to the IVD scenario, randomly selecting a group of parameters each time, and obtaining the score of the group of parameters through the scoring algorithm, and completing the parameter self-learning when the score reaches the preset score threshold; otherwise, the parameter self-learning fails;

[0017] S33: Select the next IVD scenario and repeat steps S31-S32 until all IVD scenario algorithm parameters are self-learned.

[0018] In step S4: preprocessing is performed on the input detection image.

[0019] In step S4: the preprocessing includes image filtering, morphology, image enhancement, image transformation, histogram statistics, image resampling, etc.

[0020] In step S5: the IVD scenarios include: horizontal tube judgment, top-down tube judgment, sample needle detection, blood bag OCR detection, blood quality detection, etc.

[0021] In step S5: In step S5, the scene can be selected manually or the current scene to be detected can be determined by the self-learning model based on the input image.

[0022] In step S5: algorithm post-processing includes combining, sorting, summarizing or discarding the detection results output by the corresponding scene algorithm.

[0023] A universal IVD detection system includes an image acquisition device, a processor and a memory, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0024] Compared with the prior art, the invention has the beneficial effect of realizing configuration parameters and optimal settings of hardware configuration and algorithm output through self-learning of a model based on a neural network. At the same time, it integrates a variety of IVD detection scenarios including tube judgment, sample needle detection, blood bag detection, etc., realizing a set of detection systems to achieve multifunctional detection, and can also realize the combination and arrangement of algorithm processing, and automatically output accurate and clear detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 FIG. 4 is a general IVD flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] Example 1

[0028] like Figure 1 As shown, the general IVD detection method based on deep learning includes the following steps:

[0029] S1: Configure self-learning model and parameters;

[0030] S2: Execute hardware parameter self-learning;

[0031] S3: Execute algorithm parameter self-learning;

[0032] S4: input the image to be detected;

[0033] S5: Select the corresponding IVD scenario algorithm and perform the test;

[0034] S6: Execute algorithm post-processing and output the detection results.

[0035] In step S2, the hardware parameters are self-learned, including autofocus and automatic adjustment of exposure parameters.

[0036] In step S3: the algorithm's parameter self-learning is specifically as follows:

[0037] S31: Select an IVD scene and complete the autofocus of the device through autofocus and autoexposure algorithms;

[0038] S32: Setting the scoring algorithm and preset parameters corresponding to the IVD scenario, randomly selecting a group of parameters each time, and obtaining the score of the group of parameters through the scoring algorithm, and completing the parameter self-learning when the score reaches the preset score threshold; otherwise, the parameter self-learning fails;

[0039] S33: Select the next IVD scenario and repeat steps S31-S32 until all IVD scenario algorithm parameters are self-learned.

[0040] In step S4: preprocessing is performed on the input detection image.

[0041] In step S4: the preprocessing includes image filtering, morphology, image enhancement, image transformation, histogram statistics, image resampling, etc.

[0042] In step S5: the IVD scenarios include: horizontal tube judgment, top-down tube judgment, sample needle detection, blood bag OCR detection, blood quality detection, etc.

[0043] In step S5: In step S5, the scene can be selected manually or the current scene to be detected can be determined by the self-learning model based on the input image.

[0044] In step S5: algorithm post-processing includes combining, sorting, summarizing or discarding the detection results output by the corresponding scene algorithm.

[0045] Example 2

[0046] A universal IVD detection system includes an image acquisition device, a processor and a memory, wherein the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0047] like Figure 1 As shown, the general IVD detection method based on deep learning includes the following steps:

[0048] S1: Configure self-learning model and parameters;

[0049] S2: Execute hardware parameter self-learning;

[0050] S3: Execute algorithm parameter self-learning;

[0051] S4: input the image to be detected;

[0052] S5: Select the corresponding IVD scenario algorithm and perform the test;

[0053] S6: Execute algorithm post-processing and output the detection results.

[0054] In step S2, the hardware parameters are self-learned, including autofocus and automatic adjustment of exposure parameters.

[0055] In step S3: the algorithm's parameter self-learning is specifically as follows:

[0056] S31: Select an IVD scene and complete the autofocus of the device through autofocus and autoexposure algorithms;

[0057] S32: Setting the scoring algorithm and preset parameters corresponding to the IVD scenario, randomly selecting a group of parameters each time, and obtaining the score of the group of parameters through the scoring algorithm, and completing the parameter self-learning when the score reaches the preset score threshold; otherwise, the parameter self-learning fails;

[0058] S33: Select the next IVD scenario and repeat steps S31-S32 until all IVD scenario algorithm parameters are self-learned.

[0059] In step S4: preprocessing is performed on the input detection image.

[0060] In step S4: the preprocessing includes image filtering, morphology, image enhancement, image transformation, histogram statistics, image resampling, etc.

[0061] In step S5: the IVD scenarios include: horizontal tube judgment, top-down tube judgment, sample needle detection, blood bag OCR detection, blood quality detection, etc.

[0062] In step S5: In step S5, the scene can be selected manually or the current scene to be detected can be determined by the self-learning model based on the input image.

[0063] In step S5: algorithm post-processing includes combining, sorting, summarizing or discarding the detection results output by the corresponding scene algorithm.

[0064] Embodiment 3:

[0065] The application process of a visual algorithm for an IVD scenario includes: algorithm self-learning configuration parameters, image preprocessing, algorithm core part and algorithm post-processing part.

[0066] The self-learning configuration parameters include two parts: one is the self-learning of hardware parameters, including autofocus, automatic adjustment of exposure parameters, etc.; the other is the self-learning of algorithm parameters. The specific process is as follows:

[0067] For the selected IVD scene, the autofocus and auto-exposure algorithms are first used to complete the autofocus of the device to ensure clear imaging and accurate exposure.

[0068] For each specific IVD scenario, there is a corresponding scoring algorithm and preset parameters. Each time, a set of parameters is randomly selected and the score of the set of parameters is obtained through the scoring algorithm. When the score reaches a given score threshold, the parameter self-learning is completed. Otherwise, the output of this parameter self-learning failure.

[0069] The image preprocessing part includes image filtering, morphology, image enhancement, image transformation, histogram statistics, image resampling, etc.

[0070] IVD algorithms for various scenarios include:

[0071] 1. Horizontal tube judgment algorithm, including test tube presence detection, height detection, width detection, tube cap detection, tube cap color detection, local morphology detection and code scanning, and through the combination of the above result features, different task requirements can be achieved.

[0072] 2. Overlooking tube judgment algorithm, an algorithm for simultaneous detection of multiple rows of test tubes, including detection of test tube presence, detection of abnormal test tube placement, and detection of test tube caps.

[0073] 3. Sample needle detection algorithm: an algorithm for detecting the liquid level height of one or more sample needles, which completes related task requirements by detecting the height characteristics of the sample needles.

[0074] 4. Blood bag OCR detection: by identifying the blood type mark on the blood bag and the corresponding blood type OCR, the relevant task requirements can be met.

[0075] 5. Blood quality testing, including optimal test tube opening area, blood stratification, blood quality testing (including hemolysis H, jaundice I, lipemia L). Through the combination of the above steps, blood quality testing related tasks can be achieved.

[0076] Fourth, the algorithm post-processing part is to combine, organize, summarize or discard the output of each scene algorithm. For example, the output results that meet the given conditions are set as OK, those that do not meet the conditions are set as NG, or the algorithm task 1 and algorithm task 5 are combined and outputted.

[0077] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention in the form of a ring-shaped light source. Therefore, the embodiments should be considered exemplary and non-restrictive in every sense, and the scope of the invention is defined by the appended claims rather than the above description, and it is intended that all changes within the meaning and range of equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0078] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. General IVD detection method based on deep learning, It is characterized in that The following steps are involved: S1: Configure self-learning model and parameters; S2: Execute hardware parameter self-learning; S3: Execute algorithm parameter self-learning; S4: input the image to be detected; S5: Select the corresponding IVD scenario algorithm and perform the test; S6: Execute algorithm post-processing and output the detection results.

2. According to the universal IVD detection method based on deep learning according to claim 1, It is characterized in that In step S2, the hardware parameters are self-learned, including autofocus and automatic adjustment of exposure parameters.

3. According to the universal IVD detection method based on deep learning according to claim 1, It is characterized in that In step S3: the algorithm's parameter self-learning is specifically as follows: S31: Select an IVD scene and complete the autofocus of the device through autofocus and autoexposure algorithms; S32: Setting the scoring algorithm and preset parameters corresponding to the IVD scenario, randomly selecting a group of parameters each time, and obtaining the score of the group of parameters through the scoring algorithm, and completing the parameter self-learning when the score reaches the preset score threshold; otherwise, the parameter self-learning fails; S33: Select the next IVD scenario and repeat steps S31-S32 until all IVD scenario algorithm parameter self-learning is completed.

4. The universal IVD detection method based on deep learning according to claim 1, It is characterized in that In step S4: preprocessing is performed on the input detection image.

5. The universal IVD detection method based on deep learning according to claim 4, It is characterized in that In step S4: the preprocessing includes image filtering, morphology, image enhancement, image transformation, histogram statistics, image resampling, etc.

6. The universal IVD detection method based on deep learning according to claim 1, It is characterized in that In step S5: the IVD scenarios include: horizontal tube judgment, top-down tube judgment, sample needle detection, blood bag OCR detection, blood quality detection, etc.

7. The universal IVD detection method based on deep learning according to claim 6, It is characterized in that In step S5: In step S5, the scene can be selected manually or the current scene to be detected can be determined by the self-learning model based on the input image.

8. The universal IVD detection method based on deep learning according to claim 1, It is characterized in that In step S5: algorithm post-processing includes combining, sorting, summarizing or discarding the detection results output by the corresponding scene algorithm.

9. General IVD detection system, including image acquisition equipment, processor and memory, It is characterized in that The memory stores a computer program, and when the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images