Bone marrow smear quality grading and control system based on multi-dimensional image quality evaluation
Patent Information
- Application Number
- CN202611451734.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-18
- Publication Date
- 2026-10-23
AI Technical Summary
[0003]当前行业内的常规骨髓涂片图像质控方案，大多采用整幅图像全局清晰度评估的粗放模式，仅能通过单幅图像的平均梯度值判定整张涂片是否合格，无法对不合格的模糊涂片进行原因溯源，例如缺乏涂片模糊根因的自动分级识别能力，无法区分设备自动对焦抖动、运动平台偏移引发的动态失焦模糊与制片环节盖玻片移位、细胞层局部堆积引发的稳定制片缺陷模糊，所有模糊的样本涂片均被统一退回重新制作，耗时耗力
1.本发明利用多维度图像在将骨髓涂片进行子区划分后，对每个所划分后的子区进行清晰度评估，筛选出模糊子区，对模糊子区进行空间维度上的稳定分析，完成对模糊子区的主导原因进行分级识别操作，能精准区分是设备自动对焦抖动、运动平台偏移引发的动态模糊，属于成像环节的设备责任；还是制片过程中盖玻片移位、细胞层局部堆积引发的稳定模糊，属于制片环节的样本责任，从而对造成骨髓涂片质量模糊的原因进行分级，进而制定针对性的处理策略。
Smart Images
Figure CN122967054A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image data processing and microscopic imaging quality control technology, specifically to a bone marrow smear quality grading and control system based on multi-dimensional image quality assessment. Background Technology
[0002] Bone marrow cytology is a core technical means for the diagnosis, treatment evaluation, and prognosis of hematological diseases, and its diagnostic accuracy highly depends on the quality of the microscopic images acquired by the digital slide reading system. With the continuous advancement of the full digital transformation of pathological diagnosis, the traditional manual microscopic examination mode is gradually being replaced by fully automated digital bone marrow smear scanning systems. A single device can scan hundreds of smear samples per day, which increases the requirements for automated control of image quality.
[0003] Current industry standard bone marrow smear image quality control solutions mostly adopt a crude approach of global image sharpness assessment. This approach can only determine whether the entire smear is qualified based on the average gradient value of a single image, and cannot trace the root cause of unqualified blurry smears. For example, it lacks the ability to automatically classify and identify the root cause of smear blur, and cannot distinguish between dynamic defocus blur caused by equipment autofocus jitter and motion platform offset and stable slide preparation defects blur caused by coverslip displacement and local accumulation of cell layers during the slide preparation process. All blurry sample smears are uniformly returned for reprocessing, which is time-consuming and labor-intensive.
[0004] Furthermore, in bone marrow smears prepared using proper techniques, the cell layer thickness is not completely uniform. Local areas exhibit steep, abrupt changes in thickness due to cell accumulation, much larger than typical, flatter areas. When the scanning platform traverses these high-span areas at a fixed speed, the autofocus module's response bandwidth cannot match the rate of thickness abrupt changes, easily leading to focus lag and causing localized defocusing and blurring. Simultaneously, it's impossible to dynamically adjust the scanning speed for specific areas during the scan; the only way to mitigate the defocusing problem is to drastically reduce the overall scanning speed. Ultimately, this doubles the scanning time for a single smear, significantly reducing the equipment's daily scanning capacity.
[0005] To this end, the present invention provides a bone marrow smear quality grading and control system based on multi-dimensional image quality assessment. Summary of the Invention
[0006] The purpose of this invention is to provide a bone marrow smear quality grading and control system based on multi-dimensional image quality assessment, so as to solve the above-mentioned background problems.
[0007] The objective of this invention can be achieved through the following technical solutions: A bone marrow smear quality grading and control system based on multi-dimensional image quality assessment includes: Image quality assessment module: After dividing the bone marrow smear into sub-regions using multi-dimensional images, the clarity of each sub-region is assessed, and blurry sub-regions are filtered out. Fuzzy hierarchical recognition module: Performs spatial stability analysis on fuzzy sub-regions and completes hierarchical recognition of the dominant causes of fuzzy sub-regions; Scan response analysis module: When the main cause of the blurred sub-region is identified as autofocus change, the thickness change on the bone marrow smear is analyzed to identify the high-span sub-region and compare it with the blurred sub-region in spatial dimension to determine whether the current scan speed is not responding in time. Scanning dynamic control module: If the current scanning speed response is not timely, the scanning speed in the high-span area will be dynamically adjusted and controlled based on the thickness changes in the high-span area.
[0008] The preferred method for filtering fuzzy sub-regions is as follows: The bone marrow smear image is divided into multiple quality assessment sub-regions, and the gray-level gradient magnitude of each pixel in the quality assessment sub-region is calculated using the calculation principle of the Sobel operator to obtain the single-pixel gradient magnitude. The average gradient magnitude of each pixel in each quality assessment sub-region is calculated to obtain the sharpness assessment value. If the sharpness assessment value is less than the sharpness assessment threshold, it is marked as a blurry sub-region.
[0009] Preferably, the spatial stability analysis of the fuzzy sub-region is performed as follows: The bone marrow smear images were transformed into two-dimensional spatial coordinates to construct a coordinate system for the smear images; Extract the center point of each quality assessment sub-region and use it as the coordinate point of the quality assessment sub-region on the smear image coordinate system, i.e., the sub-region coordinate point; At each monitoring time point, the coordinate points of the fuzzy sub-region are obtained and sorted according to the time sequence of the obtained coordinates to construct the fuzzy sub-region coordinate sequence. Based on the obtained time sequence, the coordinates of adjacent monitoring time points within the fuzzy sub-region coordinate sequence are substituted into the Euclidean distance formula for calculation. The result is then calculated as a ratio of the coordinates to the perimeter of the bone marrow smear image edge contour to obtain the fuzzy position stability value.
[0010] Preferably, the hierarchical recognition operation is as follows: At each monitoring time point, the area of the fuzzy sub-region is obtained and sorted according to the time sequence of the obtained data to construct a sequence of fuzzy sub-region areas. Based on the obtained time sequence, the areas of the fuzzy sub-regions at adjacent monitoring time points within the fuzzy sub-region area sequence are substituted into the Euclidean distance formula for calculation and processing. The ratio of this ratio to the area of the bone marrow smear image is then used as the stable value for the fuzzy range. The fuzzy location stability value and the fuzzy range stability value are summed to obtain the fuzzy cause identification value; If the value of the cause of ambiguity is greater than the threshold for the cause of ambiguity, then the dominant cause is identified as autofocus change, i.e., the device-dominant cause. If the fuzzy cause identification value is less than or equal to the fuzzy cause identification threshold, the dominant cause is identified as coverslip displacement, which is the dominant cause of the slide preparation itself.
[0011] Preferably, the thickness changes on the bone marrow smear are analyzed as follows: Each quality assessment sub-region on the bone marrow smear is extracted, and the gray-level gradient amplitude is calculated using the Sobel operator as a characterization of the thickness change rate, thus obtaining the single-pixel thickness amplitude.
[0012] Preferably, the high-span sub-region is determined as follows: The average value of the thickness amplitude of a single pixel is summed to obtain the thickness variation analysis value. If the thickness variation analysis value is greater than the thickness variation analysis threshold, it belongs to the high span sub-region.
[0013] Preferably, the determination of whether the current scanning speed is unresponsive is as follows: The coordinates of the high-span sub-region and the blurred sub-region on the smear image coordinate system are extracted separately. If the coordinates of the high-span sub-region and the blurred sub-region on the smear image coordinate system coincide, an autofocus transformation phenomenon occurs, indicating that the current scanning speed response is not timely.
[0014] Preferably, the thickness variation in the high-span region is analyzed as follows: Set a scanning monitoring cycle, divide the set scanning monitoring cycle into several scanning monitoring points, and extract the single pixel thickness amplitude of each pixel in the blurred sub-region under each scanning monitoring point.
[0015] Preferably, based on the thickness variation within the high-span region, the scanning speed analysis for the high-span region is performed as follows: The difference between the single-pixel thickness amplitude of adjacent scanning monitoring points is calculated, and the ratio is calculated with the interval time of adjacent scanning monitoring points to obtain the thickness change value of neighboring pixels. The cell scanning speed is calculated by multiplying the thickness change of each neighboring pixel by the current scanning speed.
[0016] Preferably, the scanning speed is adjusted and controlled as follows: The maximum cell scanning speed is compared with the maximum scanning speed set in the scanning system. If the maximum cell scanning speed is greater than the maximum scanning speed set in the scanning system, a dynamic adjustment signal for the scanning speed is triggered, and the difference between the maximum cell scanning speed and the maximum scanning speed is calculated as the amount of scanning speed adjustment required.
[0017] The beneficial effects of this invention are as follows: 1. This invention utilizes multi-dimensional images to divide bone marrow smears into sub-regions, assesses the sharpness of each sub-region, filters out blurred sub-regions, performs spatial stability analysis on the blurred sub-regions, and completes the hierarchical identification of the dominant causes of the blurred sub-regions. It can accurately distinguish whether the blur is caused by dynamic blur due to equipment autofocus jitter or motion platform offset, which is the equipment responsibility in the imaging stage; or whether it is stable blur caused by coverslip displacement or local accumulation of cell layers during slide preparation, which is the sample responsibility in the slide preparation stage. Thus, the causes of blurred bone marrow smear quality can be classified, and targeted treatment strategies can be formulated.
[0018] 2. When the main cause of the blurred sub-region is identified as autofocus change, the present invention analyzes the thickness change on the bone marrow smear to determine the high-span sub-region and compares it with the blurred sub-region in spatial dimension to determine whether the current scanning speed is not responding in time. If the current scanning speed is not responding in time, the scanning speed in the high-span region is dynamically adjusted and controlled according to the thickness change in the high-span region. After identifying the device-driven autofocus change-type blur, the spatial overlap verification between the high-span sub-region and the blurred sub-region accurately distinguishes between the smear thickness change causing the focus response lag and the random defocusing device failure caused by the hardware failure of the focus module. Moreover, after determining that the focus response lag is caused by the smear thickness change, the scanning speed is dynamically controlled to ensure that there is sufficient time margin for the focus response in the region of thickness change. Attached Figure Description
[0019] The invention will now be further described with reference to the accompanying drawings.
[0020] Figure 1 This is a functional module diagram of the bone marrow smear quality grading and control system based on multi-dimensional image quality assessment, as described in this invention. Figure 2 This is an overall flowchart of the method within the bone marrow smear quality grading and control system based on multi-dimensional image quality assessment in this invention; Figure 3 This is a flowchart of the logic for determining the hierarchical classification of fuzzy causes in this invention. Detailed Implementation
[0021] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments. Example 1
[0022] like Figure 1 - Figure 3 As shown, this invention provides a bone marrow smear quality grading and control system based on multi-dimensional image quality assessment, comprising: Image quality assessment module: After dividing the bone marrow smear into sub-regions using multi-dimensional images, the clarity of each sub-region is assessed, and blurry sub-regions are filtered out. It should be noted that the multi-dimensional image includes at least two or more of the following: grayscale image, gradient magnitude image, local binary pattern (LBP) texture image, and frequency domain transformed image. Each dimension image is obtained by performing corresponding transformations on the original acquired bone marrow smear microscopic image. In some embodiments, the entire bone marrow smear image is divided into multiple rectangular sub-regions of the same size according to a preset size, such as 256×256 pixels or 512×512 pixels, as quality assessment sub-regions; Using the computational principle of the Sobel operator, the grayscale gradient magnitude of each pixel in the quality evaluation sub-region is calculated to obtain the single-pixel gradient magnitude. When the Sobel operator is applied to an image, it calculates the gray-level changes in the horizontal and vertical directions at each pixel location and then synthesizes a gradient magnitude, which reflects the intensity of the gray-level changes at the analyzed pixel, that is, the intensity of the edge or detail, i.e., the gradient magnitude of the pixel, or the gradient response value or edge intensity value. The average gradient magnitude of each pixel in each quality assessment sub-region is calculated to obtain the sharpness assessment value. It should be noted that the sharpness assessment value refers to the average edge detail intensity of the entire quality assessment sub-region on the bone marrow smear. Specifically, the larger the sharpness assessment value, the greater the average gray level variation of the pixels in the quality assessment sub-region, meaning that the detail information such as cell outlines and staining textures is clearer. Conversely, the smaller the sharpness assessment value, the smaller the average gray level variation of the pixels in the quality assessment sub-region, meaning that the detail information such as cell outlines and staining textures is less clear. If the sharpness assessment value is greater than or equal to the sharpness assessment threshold, it means that the average gray level of the pixels in the quality assessment sub-region changes significantly, that is, the sharpness of details such as cell outlines and staining textures is high, and the analyzed quality assessment sub-region is marked as a sharp sub-region. If the sharpness assessment value is less than the sharpness assessment threshold, it means that the average gray level of the pixels in the quality assessment sub-region is small, that is, the sharpness of details such as cell outlines and staining textures is low, and the analyzed quality assessment sub-region is marked as a blurry sub-region. It should be noted that the sharpness assessment threshold is a critical quantitative boundary value used to divide the sharp sub-regions and the blurry sub-regions. It is an engineering judgment benchmark obtained by those skilled in the art based on the actual business needs of bone marrow smear pathology reading, the inherent parameters of the image acquisition equipment, and statistical calibration of a large number of labeled samples. Specifically, those skilled in the art can collect qualified bone marrow smear images from the same type of equipment in batches, statistically analyze the distribution range of the clarity assessment values of all quality assessment sub-regions, remove low-value samples that deviate significantly from the normal gradient distribution, and take the lower boundary of the distribution range as the clarity assessment threshold, with a value range of 15 to 35.
[0023] Fuzzy hierarchical recognition module: Performs spatial stability analysis on fuzzy sub-regions and completes hierarchical recognition of the dominant causes of fuzzy sub-regions; In some embodiments, the bone marrow smear image is transformed into two-dimensional spatial coordinates, wherein the center point on the bone marrow smear image is extracted as the origin of spatial coordinates to construct a smear image coordinate system; Extract the center point of each quality assessment sub-region and use it as the coordinate point of the quality assessment sub-region on the smear image coordinate system, i.e., the sub-region coordinate point; Since the fuzzy sub-regions are identified by analyzing each quality assessment sub-region, the center point of the fuzzy sub-region is also the coordinate point of the fuzzy sub-region on the smear image coordinates, i.e., the sub-region coordinate point. Set a sub-area location monitoring cycle and divide the sub-area location monitoring cycle into several monitoring time points, wherein the interval between adjacent monitoring time points is equal; Preferably, the total duration of the sub-region location monitoring cycle is the same as the total duration of the whole bone marrow smear scan, with a value range of 30 seconds to 300 seconds, and the interval duration Δt_m between adjacent monitoring time points ranges from 0.1 seconds to 2 seconds, preferably 0.5 seconds.
[0024] At each monitoring time point, the coordinate points of the fuzzy sub-region are obtained and sorted according to the time sequence of the obtained coordinates to construct the fuzzy sub-region coordinate sequence. Based on the obtained time sequence, the coordinates of adjacent monitoring time points within the fuzzy sub-region coordinate sequence are substituted into the Euclidean distance formula for calculation and processing. The ratio of this value to the perimeter of the bone marrow smear image edge contour is then calculated to obtain the fuzzy position stability value. At each monitoring time point, the area of the fuzzy sub-region is obtained and sorted according to the time sequence of the obtained data to construct a sequence of fuzzy sub-region areas. Based on the obtained time sequence, the areas of the fuzzy sub-regions at adjacent monitoring time points within the fuzzy sub-region area sequence are substituted into the Euclidean distance formula for calculation and processing. The ratio of this ratio to the area of the bone marrow smear image is then used as the stable value for the fuzzy range. The fuzzy location stability value and the fuzzy range stability value are summed to obtain the fuzzy cause identification value; Understandably, the fuzzy cause identification value refers to a comprehensive quantitative index designed for the spatial stability analysis and hierarchical identification of dominant causes in the fuzzy sub-regions of bone marrow smears. It reflects the stability of the spatial location and coverage of the fuzzy sub-region over time. Specifically, the larger the fuzzy cause identification value, the greater the fluctuation of the spatial location and coverage of the fuzzy sub-region over time; the smaller the fuzzy cause identification value, the smaller the fluctuation of the spatial location and coverage of the fuzzy sub-region over time. If the blur cause identification value is greater than the blur cause identification threshold, it indicates that the spatial location and coverage of the blurry sub-region fluctuate greatly over time. In this case, the dominant cause is identified as autofocus change, i.e., the device-dominant cause. If the fuzzy cause identification value is less than or equal to the fuzzy cause identification threshold, it indicates that the spatial location and coverage of the fuzzy sub-region fluctuate little over time. The dominant cause is identified as coverslip displacement, which is the dominant cause of the slide fabrication itself. It should be noted that the blur cause identification value refers to the critical quantitative boundary value used to accurately distinguish between equipment-driven blur and slide-driven blur. It is a value that is used by medical image quality control technicians in this field to collect bone marrow smear samples from the same model of microscopic imaging equipment in batches, and to separately count the blur cause identification values of stable blur samples known to be caused by slide-making defects and dynamic blur samples known to be caused by equipment focusing jitter. The numerical distribution range of the two types of samples is obtained, and the midpoint of the two distribution ranges is taken as the blur cause identification threshold, with a value range of 0.3 to 0.7.
[0025] This embodiment specifically involves: after dividing the bone marrow smear into sub-regions using multi-dimensional images, evaluating the sharpness of each sub-region, filtering out blurred sub-regions, performing spatial stability analysis on the blurred sub-regions, and completing the hierarchical identification of the dominant causes of the blurred sub-regions. It can accurately distinguish whether the blur is caused by dynamic blur due to equipment autofocus jitter or motion platform offset, which is the equipment responsibility in the imaging process; or whether it is caused by stable blur due to coverslip displacement or local accumulation of cell layers during slide preparation, which is the sample responsibility in the slide preparation process. This allows for the hierarchical classification of the causes of blurred bone marrow smear quality, and the formulation of targeted treatment strategies. Example 2
[0026] like Figure 1 - Figure 3 As shown, this invention provides a bone marrow smear quality grading and control system based on multi-dimensional image quality assessment, comprising: Scan response analysis module: When the main cause of the blurred sub-region is identified as autofocus change, the thickness change on the bone marrow smear is analyzed to identify the high-span sub-region and compare it with the blurred sub-region in spatial dimension to determine whether the current scan speed is not responding in time. In some embodiments, each quality assessment sub-region on the bone marrow smear is extracted, and the Sobel operator is used to calculate the thickness of the pixels in each quality assessment sub-region to obtain the single pixel thickness amplitude. It should be noted that in bone marrow smear microscopy, there is a Lambert-Beer law relationship between the physical thickness of the cell layer and the intensity of transmitted light. In grayscale images, the pixel grayscale value is directly proportional to the intensity of transmitted light. Therefore, changes in physical thickness will lead to changes in grayscale value. There is an approximately linear relationship between the grayscale gradient amplitude |∇G| and the thickness change rate |∇d|. Based on this, this invention uses the grayscale gradient amplitude calculated by the Sobel operator as a characterization of the thickness change rate, which is called the single-pixel thickness amplitude. The average value of the thickness amplitude of a single pixel is summed to obtain the thickness variation analysis value. If the thickness change analysis value is greater than the thickness change analysis threshold, it indicates that the thickness change in the local area where the pixel is located within the analyzed quality assessment sub-region is relatively steep, belonging to a high-span sub-region. If the thickness variation analysis value is less than or equal to the thickness variation analysis threshold, it indicates that the thickness variation in the local area where the pixel is located within the analyzed quality assessment sub-region is relatively gentle, belonging to a low-span sub-region. It should be noted that the thickness variation analysis threshold is used to accurately divide the high-span sub-regions and low-span sub-regions, and to determine the steepness of the local smear thickness variation. Specifically, those skilled in the art calculate the corresponding thickness variation analysis value for each sub-region of all the qualified standard smears collected, and statistically analyze the numerical distribution histogram of the thickness variation analysis values of all sub-regions. After obtaining the normal distribution range of the thickness variation analysis values under the whole sample, the thickness variation analysis value distribution of two types of extreme samples is statistically analyzed: one type is a low-span sub-region sample known to have a smooth thickness change and a focus response that can keep up with the scanning speed, and the other type is a high-span sub-region sample known to have a steep thickness change and a lagging focus response. The intermediate critical boundary point between the two distribution ranges is found as the thickness variation analysis threshold. The coordinates of the high-span sub-region and the blurred sub-region on the coordinate system of the smear image were extracted separately, and the overlap comparison was performed as follows: If the coordinates of the high-span sub-region and the blurred sub-region coincide in the coordinate system of the smear image, it indicates that the thickness change in the high-span sub-region is relatively steep, and the scanning response is not timely when scanning at the current scanning speed, resulting in the phenomenon of autofocus switching. This indicates that the current scanning speed is not responding in time. If the coordinates of the high-span sub-region and the blurred sub-region do not coincide in the coordinate system of the smear image, it indicates that the thickness change in the high-span sub-region is relatively steep, and does not lead to the phenomenon of autofocus transformation.
[0027] Scanning dynamic control module: If the current scanning speed response is not timely, the scanning speed in the high-span area will be dynamically adjusted and controlled according to the thickness changes in the high-span area. In some embodiments, a scanning monitoring cycle is set, and the set scanning monitoring cycle is divided into several scanning monitoring points, wherein the interval between adjacent scanning monitoring points is equal. Preferably, the total duration of the scanning monitoring cycle ranges from 1 second to 10 seconds, and is preferably 3 seconds; Extract the single-pixel thickness amplitude of each pixel in the blurred sub-region under each scanning monitoring point; It should be noted that a single scan monitoring point acquires the thickness amplitude of a single pixel. The difference between the single-pixel thickness amplitude of adjacent scanning monitoring points is calculated, and the ratio is calculated with the interval time of adjacent scanning monitoring points to obtain the thickness change value of neighboring pixels. The cell scanning speed is calculated by multiplying the thickness change of each neighboring pixel by the current scanning speed; The scanning speeds of all cells are compared, and the maximum cell scanning speed is compared with the maximum scanning speed set in the scanning system as follows: If the maximum cell scanning speed is greater than the maximum scanning speed set in the scanning system, a dynamic adjustment signal for the scanning speed is triggered, and the difference between the maximum cell scanning speed and the maximum scanning speed is calculated as the amount of scanning speed adjustment required. If the maximum unit scanning speed is less than or equal to the maximum scanning speed set in the scanning system, the current scanning speed will remain unchanged, and thickness data will continue to be collected during the scanning monitoring cycle. When the scanning speed dynamic adjustment signal is triggered, the scanning speed will be dynamically controlled. It should be noted that the maximum scanning speed set within the scanning system is determined by the rated speed of the scanning platform motor and the exposure time of the image acquisition camera.
[0028] Specifically, in this embodiment: when the main cause of the blurred sub-region is identified as autofocus change, the thickness change on the bone marrow smear is analyzed to determine the high-span sub-region, and the spatial overlap of the high-span sub-region with the blurred sub-region is compared to determine whether the current scanning speed is not responding in time. If the current scanning speed is not responding in time, the scanning speed in the high-span region is dynamically adjusted and controlled based on the thickness change in the high-span region. After identifying the device-driven autofocus change-type blur, the spatial overlap verification between the high-span sub-region and the blurred sub-region accurately distinguishes between the smear thickness change causing the focus response lag and the random defocusing device failure caused by the hardware failure of the focus module. Moreover, after determining that the focus response lag is caused by the smear thickness change, the scanning speed is dynamically controlled to ensure sufficient time margin for the focus response in the region of thickness change.
[0029] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.
Claims
1. A bone marrow smear quality grading and control system based on multi-dimensional image quality assessment, characterized in that: Image quality assessment module: After dividing the bone marrow smear into sub-regions using multi-dimensional images, the clarity of each sub-region is assessed, and blurry sub-regions are filtered out. Fuzzy hierarchical recognition module: Performs spatial stability analysis on fuzzy sub-regions and completes hierarchical recognition of the dominant causes of fuzzy sub-regions; Scan response analysis module: When the main cause of the blurred sub-region is identified as autofocus change, the thickness change on the bone marrow smear is analyzed to identify the high-span sub-region and compare it with the blurred sub-region in spatial dimension to determine whether the current scan speed is not responding in time. Scanning dynamic control module: If the current scanning speed response is not timely, the scanning speed in the high-span area will be dynamically adjusted and controlled based on the thickness changes in the high-span area.
2. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 1, characterized in that: The fuzzy sub-regions are filtered as follows: The bone marrow smear image is divided into multiple quality assessment sub-regions, and the gray-level gradient magnitude of each pixel in the quality assessment sub-region is calculated using the calculation principle of the Sobel operator to obtain the single-pixel gradient magnitude. The average gradient magnitude of each pixel in each quality assessment sub-region is calculated to obtain the sharpness assessment value. If the sharpness assessment value is less than the sharpness assessment threshold, it is marked as a blurry sub-region.
3. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 1, characterized in that: The spatial stability analysis of the fuzzy subregion is as follows: The bone marrow smear images were transformed into two-dimensional spatial coordinates to construct a coordinate system for the smear images; Extract the center point of each quality assessment sub-region and use it as the coordinate point of the quality assessment sub-region on the smear image coordinate system, i.e., the sub-region coordinate point; At each monitoring time point, the coordinate points of the fuzzy sub-region are obtained and sorted according to the time sequence of the obtained coordinates to construct the fuzzy sub-region coordinate sequence. Based on the obtained time sequence, the coordinates of adjacent monitoring time points within the fuzzy sub-region coordinate sequence are substituted into the Euclidean distance formula for calculation. The result is then calculated as a ratio of the coordinates to the perimeter of the bone marrow smear image edge contour to obtain the fuzzy position stability value.
4. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 3, characterized in that: The hierarchical recognition process is as follows: At each monitoring time point, the area of the fuzzy sub-region is obtained and sorted according to the time sequence of the obtained data to construct a sequence of fuzzy sub-region areas. Based on the obtained time sequence, the areas of the fuzzy sub-regions at adjacent monitoring time points within the fuzzy sub-region area sequence are substituted into the Euclidean distance formula for calculation and processing. The ratio of this ratio to the area of the bone marrow smear image is then used as the stable value for the fuzzy range. The fuzzy location stability value and the fuzzy range stability value are summed to obtain the fuzzy cause identification value; If the value of the cause of ambiguity is greater than the threshold for the cause of ambiguity, then the dominant cause is identified as autofocus change, i.e., the device-dominant cause. If the fuzzy cause identification value is less than or equal to the fuzzy cause identification threshold, the dominant cause is identified as coverslip displacement, which is the dominant cause of the slide preparation itself.
5. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 1, characterized in that: The thickness variations in bone marrow smears were analyzed as follows: Each quality assessment sub-region on the bone marrow smear is extracted, and the gray-level gradient amplitude is calculated using the Sobel operator as a characterization of the thickness change rate to obtain the single-pixel thickness amplitude.
6. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 5, characterized in that: The high-span sub-regions are determined as follows: The average value of the thickness amplitude of a single pixel is summed to obtain the thickness variation analysis value. If the thickness variation analysis value is greater than the thickness variation analysis threshold, it belongs to the high span sub-region.
7. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 1, characterized in that: The determination of whether the current scan speed is unresponsive is as follows: The coordinates of the high-span sub-region and the blurred sub-region on the smear image coordinate system are extracted separately. If the coordinates of the high-span sub-region and the blurred sub-region on the smear image coordinate system coincide, an autofocus transformation phenomenon occurs, indicating that the current scanning speed response is not timely.
8. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 7, characterized in that: The thickness variation in the high-span region is analyzed as follows: Set a scanning monitoring cycle, divide the set scanning monitoring cycle into several scanning monitoring points, and extract the single pixel thickness amplitude of each pixel in the blurred sub-region under each scanning monitoring point.
9. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 8, characterized in that: Based on the thickness variation within the high-span region, the scanning speed analysis for the high-span region is as follows: The difference between the single-pixel thickness amplitude of adjacent scanning monitoring points is calculated, and the ratio is calculated with the interval time of adjacent scanning monitoring points to obtain the thickness change value of neighboring pixels. The cell scanning speed is calculated by multiplying the thickness change of each neighboring pixel by the current scanning speed.
10. The bone marrow smear quality grading and control system based on multi-dimensional image quality assessment according to claim 9, characterized in that: The scanning speed can be adjusted and controlled as follows: The maximum cell scanning speed is compared with the maximum scanning speed set in the scanning system. If the maximum cell scanning speed is greater than the maximum scanning speed set in the scanning system, a dynamic adjustment signal for the scanning speed is triggered, and the difference between the maximum cell scanning speed and the maximum scanning speed is calculated as the amount of scanning speed adjustment required.