Automated method for evaluating the quality of LD CytoMatrix 20 slide scan images

Abstract

This invention provides an automated method for evaluating the image quality of LD CytoMatrix 20 slide scans. The algorithm obtains digital images of cervical cells by scanning the entire slide, calculates the information content of each image, and filters out images with acceptable information content. Then, it calculates the sum of the G, R, and B values ​​of the top five images based on their information entropy. The staining condition of the image is determined according to the RGB threshold of a standard image. All images with an information content greater than 2.0 are then input into a three-class focusing model to determine if the image is in focus. Finally, the staining and focusing results are combined to determine the overall image quality of the sample, and a quality score is assigned to each sample image. The LD CytoMatrix 20 slide scan image quality automated evaluation method can efficiently and accurately check the quality of cervical cytology slide images, reducing human error and improving the speed and accuracy of pathological diagnosis. This method is not only applicable to cervical cytology but also widely used in other types of cytopathology and histopathology research.

Description

Technical Field

[0001] This invention relates to the field of image quality assessment, and in particular to an automated method for assessing the quality of LD CytoMatrix 20 slide scan images. Background Technology

[0002] In the field of pathology, especially in liquid-based cytology (LBC), a branch of cellular pathology, the preparation and analysis of cell samples are crucial for disease diagnosis. LBC technology involves collecting cell samples in a liquid fixative, followed by slide preparation, staining, and other steps to create glass slides for microscopic examination. One of the key aspects of this process is ensuring that the prepared slide images are of sufficiently high quality so that pathologists can accurately identify and analyze cell morphology and structure.

[0003] Traditional assessment of pathological slide image quality relies on manual visual examination, which is not only time-consuming and labor-intensive, but also easily affected by subjective factors, resulting in limited consistency and accuracy of assessment results.

[0004] In recent years, with the development of digital pathology, the widespread adoption of slide scanning technology, and the advancements in artificial intelligence (AI) and machine learning algorithms, automated image quality assessment has become possible, greatly improving the efficiency and reliability of pathological image analysis. However, deep learning models are often referred to as "black boxes" because their decision-making processes are difficult to understand.

[0005] The design and training of deep learning models typically involve multiple complex algorithmic steps, including preprocessing, feature extraction, model selection, and optimization. The efficiency of these steps directly impacts the overall model's processing speed and the quality of its output. Unfortunately, existing deep learning frameworks and toolchains often fail to provide sufficiently intuitive and efficient solutions, resulting in lengthy model development cycles and output structures that may not meet the needs of clinical practice. Summary of the Invention

[0006] The main objective of this invention is to provide an automated method for evaluating the quality of LD CytoMatrix 20 slide scan images, which solves the problem that traditional pathological slide image quality evaluation relies on manual visual inspection, which is not only time-consuming and labor-intensive, but also easily affected by subjective factors.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: an automated method for evaluating the quality of LD CytoMatrix 20 slide scan images.

[0008] S1. First, digital images of cervical cells are obtained through full slide scanning;

[0009] S2. Input the 1024-sized image into the information entropy function and calculate the information content of each image;

[0010] If the information content is greater than 2.0, the image information content is considered acceptable.

[0011] Otherwise, it is considered a blank image or an image with little information;

[0012] The information entropy of all images will be recorded and sorted;

[0013] S3. Based on the top five images in terms of information entropy in step S2, calculate the sum of red, green, and blue pixel values ​​of these images, and compare them with the RGB threshold of the standard image to determine the coloring quality of the image, and record the determination result.

[0014] S4. Input all images with an information entropy greater than 2.0 into a three-class focusing model. This model can distinguish whether an image is well focused, out of focus, or focused on a layer of impurities.

[0015] S5. Finally, combining the staining and focusing results from steps S3 and S4, the image quality of the entire sample is assessed. If both staining and focusing are good, the sample is deemed acceptable and given a corresponding image quality score.

[0016] In the preferred scheme, the method for determining the image staining quality in step S3 is as follows:

[0017] A1. Use a pre-trained convolutional neural network (CNN) to extract a high-dimensional feature vector F = f(I) for each image, where I represents the input image and f is the feature extraction function of the CNN.

[0018] A2. Considering the diversity of color spaces, the RGB space is converted into HSV and Lab spaces, and features F are extracted respectively. HSV and F Lab Then, these features are combined using a weighted fusion strategy to obtain the comprehensive feature F. combined :

[0019] F combined =α·F + β·F HSV +γ·F Lab (1);

[0020] Where α, β, γ are weight parameters, and α+β+γ=1 is determined through cross-validation optimization.

[0021] A3. Determine the optimal staining threshold T using dynamic programming. color Based on the statistical characteristics of the current batch of image features, the following formula is used:

[0022] T color =μ+k·σ (2);

[0023] Where μ is the average value of the composite feature vector, σ is its standard deviation, and k is a coefficient adjusted based on historical data to balance sensitivity and specificity.

[0024] A4. To determine staining quality, compare the comprehensive characteristics with the threshold T. color Staining quality is classified. Staining S is defined. color The score is expressed as:

[0025]

[0026] Where λ is the slope parameter of the Sigmoid function, adjusting the steepness of the classification curve. color A value ≥0.7 indicates good staining quality, while a value lower than 0.7 indicates poor staining quality.

[0027] In the preferred scheme, the focusing model classification method used in step S4 is as follows:

[0028] B1. Extracting the focusing features F of an image using a ResNet residual network. focus ;

[0029] B2, Combining staining quality characteristics F combined and focusing feature F focus We construct a multi-task learning model to simultaneously optimize the prediction of coloring and focusing states;

[0030] B3. Calculate the probability distribution P of the focusing state using the softmax function. focus P focus Represented as:

[0031] P focus =softmax(W·[F combined ,F focus ] T +b)(4);

[0032] Where W is the weight matrix, b is the bias term, [F combined ,F focus ] represents the concatenated vector of features, and the softmax function ensures that the sum of probabilities is 1;

[0033] B4. Select the category with the highest probability as the final focusing state. If the probability of the focusing category exceeds 0.8, it is judged as good focusing; if the probability of focusing on the impurity layer is the highest and exceeds 0.6, it is judged as poor focusing.

[0034] In the preferred scheme, the method for assigning the corresponding image quality score in step S5 is as follows:

[0035] C1. Coloring quality is quantified by the color consistency index Qc. The Qc index is based on the difference between the sum of the RGB values ​​of the image and the sum of the RGB values ​​of the standard image. Qc is expressed as:

[0036]

[0037] Where Rs,Gs,Bs are the RGB average values ​​of the standard image, while R,G,B are the RGB average values ​​of the current image, and σ c It is the standard deviation of staining consistency, used to control the sensitivity of the formula;

[0038] C2, Focused state via the focus index Q f To evaluate, Q f The exponent is calculated using the Laplacian edge strength of the image and the image entropy, Q. f Represented as:

[0039]

[0040] in, H(I) is the Laplacian operator result of image I at point (x,y), and H(I) is the information entropy of image I.

[0041] C3, Overall Staining Quality Q c and focused state Q f Give the image quality score Q. t The quality score Q is calculated using a weighted average formula. t Q t Represented as:

[0042] Q t =w c ·Q c +w f ·Q f (7);

[0043] Among them, w c and w f These are the weighting factors for staining quality and focusing state, respectively, satisfying w c +w f =1;

[0044] The weighting factors can be adjusted based on the experimental results to optimize the overall score;

[0045] C4. Set the qualified threshold T q If the overall score Q t Greater than or equal to T q If the image quality is satisfactory, it is considered acceptable; otherwise, it is deemed unacceptable.

[0046] This invention provides an automated method for evaluating the quality of LD CytoMatrix 20 slide scan images. This algorithm integrates deep learning with traditional statistical methods, increasing the model's complexity and adaptability, and aims to more accurately assess the quality of cervical cell slide images. Through feature extraction, fusion, and adaptive threshold calculation, it can more comprehensively consider the staining quality of the image; by utilizing deep learning for focus state classification and combining it with multi-task learning, the accuracy of focus determination is improved. These steps work together to provide a more robust algorithmic framework for the quality assessment of cervical cell slide images.

[0047] The LD CytoMatrix 20 slide image quality automated assessment method can efficiently and accurately check the quality of cervical cytology slide images, reducing human error and improving the speed and accuracy of pathological diagnosis. This method is not only applicable to cervical cytology, but also widely used in other types of cytopathology and histopathology research. Attached Figure Description

[0048] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0049] Figure 1 This is a flowchart of the algorithm of this invention. Detailed Implementation

[0050] Example 1

[0051] like Figure 1 As shown, an automated method for evaluating the quality of LD CytoMatrix 20 slide scan images is presented.

[0052] S1. First, digital images of cervical cells are obtained through full slide scanning;

[0053] S2. Input the 1024-sized image into the information entropy function and calculate the information content of each image;

[0054] If the information content is greater than 2.0, the image information content is considered acceptable.

[0055] Otherwise, it is considered a blank image or an image with little information;

[0056] The information entropy of all images will be recorded and sorted.

[0057] S3. Based on the top five images in terms of information entropy in step S2, calculate the sum of red, green, and blue pixel values ​​of these images, and compare them with the RGB threshold of the standard image to determine the coloring quality of the image, and record the determination result.

[0058] S4. Input all images with an information entropy greater than 2.0 into a three-class focusing model. This model can distinguish whether an image is well focused, out of focus, or focused on a layer of impurities.

[0059] S5. Finally, combining the staining and focusing results from steps S3 and S4, the image quality of the entire sample is assessed. If both staining and focusing are good, the sample is deemed acceptable and given a corresponding image quality score.

[0060] Example 2

[0061] Further explanation in conjunction with Example 1, such as Figure 1 As shown, the method for determining the image staining quality in step S3 is as follows:

[0062] A1. Use a pre-trained convolutional neural network (CNN) to extract the high-dimensional feature vector F = f(I) for each image, where I represents the input image and f is the feature extraction function of the CNN. The CNN can automatically learn and capture the hierarchical features in the image, which is crucial for subsequent staining quality analysis.

[0063] A2. Considering the diversity of color spaces, the RGB space is converted into HSV and Lab spaces, and features F are extracted respectively. HSV and F Lab Then, these features are combined using a weighted fusion strategy to obtain the comprehensive feature F. combined :

[0064] F combined =α·F + β·F HSV +γ·F Lab (1);

[0065] Here, α, β, and γ are weight parameters that satisfy α + β + γ = 1 and are determined through cross-validation optimization. These parameters are optimized through cross-validation to achieve the best fusion effect.

[0066] A3. Determine the optimal staining threshold T using dynamic programming. color Based on the statistical characteristics of the current batch of image features, the following formula is used:

[0067] T color =μ+k·σ (2);

[0068] Where μ is the average value of the composite feature vector, σ is its standard deviation, and k is a coefficient adjusted based on historical data to balance sensitivity and specificity.

[0069] A4. To determine staining quality, compare the comprehensive characteristics with the threshold T. color Staining quality is classified. Staining S is defined. colorThe score is expressed as:

[0070]

[0071] Where λ is the slope parameter of the Sigmoid function, adjusting the steepness of the classification curve. color A value ≥0.7 indicates good staining quality, while a value lower than 0.7 indicates poor staining quality.

[0072] This algorithm uses deep learning technology for feature extraction and fusion, combined with adaptive threshold decision-making, to more accurately evaluate the staining quality of images, thereby improving the overall accuracy of sample image quality evaluation.

[0073] Example 3

[0074] Further explanation in conjunction with Example 1, such as Figure 1 As shown, the focusing model classification method used in step S4 is as follows:

[0075] B1. Extracting the focusing features F of an image using a ResNet residual network. focus ;

[0076] B2, Combining staining quality characteristics F combined and focusing feature F focus We construct a multi-task learning model to simultaneously optimize the prediction of coloring and focusing states;

[0077] B3. Calculate the probability distribution P of the focusing state using the softmax function. focus P focus Represented as:

[0078] P focus =softmax(W·[F combined ,F focus ] T +b)(4);

[0079] Where W is the weight matrix, b is the bias term, [F combined ,F focus ] represents the concatenated vector of features, and the softmax function ensures that the sum of probabilities is 1;

[0080] B4. Select the category with the highest probability as the final focusing state. If the probability of the focusing category exceeds 0.8, it is judged as good focusing; if the probability of focusing on the impurity layer is the highest and exceeds 0.6, it is judged as poor focusing.

[0081] This algorithm combines the advantages of deep learning with statistical methods, improving the model's complexity and flexibility, thereby enabling more accurate assessment of the focus quality of cervical cell slide images. Through deep feature extraction and a multi-task learning model, the algorithm not only considers the staining quality of the image but also specifically optimizes the prediction of the focus state, ensuring the comprehensiveness and accuracy of cervical cell slide image quality assessment.

[0082] Example 4

[0083] Further explanation in conjunction with Example 1, such as Figure 1 As shown, the method for assigning the corresponding image quality score in step S5 is as follows:

[0084] C1. Coloring quality is quantified by the color consistency index Qc. The Qc index is based on the difference between the sum of the RGB values ​​of the image and the sum of the RGB values ​​of the standard image. Qc is expressed as:

[0085]

[0086] Where Rs,Gs,Bs are the RGB average values ​​of the standard image, while R,G,B are the RGB average values ​​of the current image, and σ c It is the standard deviation of staining consistency, used to control the sensitivity of the formula;

[0087] C2, Focused state via the focus index Q f To evaluate, Q f The exponent is calculated using the Laplacian edge strength of the image and the image entropy, Q. f Represented as:

[0088]

[0089] in, H(I) is the Laplacian operator result of image I at point (x,y), and H(I) is the information entropy of image I.

[0090] C3, Overall Staining Quality Q c and focused state Q f Give the image quality score Q. t The quality score Q is calculated using a weighted average formula. t Q t Represented as:

[0091] Q t =w c ·Q c +w f ·Q f (7);

[0092] Among them, w c and w fThese are the weighting factors for staining quality and focusing state, respectively, satisfying w c +w f =1;

[0093] The weighting factors can be adjusted based on the experimental results to optimize the overall score;

[0094] C4. Set the qualified threshold T q If the overall score Q t Greater than or equal to T q If the image quality is satisfactory, it is considered acceptable; otherwise, it is deemed unacceptable.

[0095] Perform steps C1 and C2 for each image, and calculate Q respectively. c and Q f .

[0096] Calculate the overall quality score Q using the formula in step C3. t .

[0097] Using the decision rule in step C4, based on Q t and T q Determine whether the image is acceptable.

[0098] The function of formula Qc is to quantify the coloring quality by calculating the similarity between the image and the standard image in the color space, so as to ensure that the image is colored evenly and close to the expected standard.

[0099] Formula Q f Its function is to quantify the focus state of an image by measuring its edge sharpness and information entropy, ensuring that the image is sharp and contains sufficient information.

[0100] Formula Q t Its function is to obtain a comprehensive quality score by weighted fusion of staining quality and focus status, providing a quantitative measure of the overall image quality.

[0101] The above algorithms can be used to systematically evaluate the quality of cervical cell slide scan images, providing strong support for pathological diagnosis.

[0102] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. An automated method for evaluating the quality of LD CytoMatrix 20 slide scan images, characterized by: S1. First, digital images of cervical cells are obtained through full slide scanning; S2. Input the 1024-sized image into the information entropy function and calculate the information content of each image; If the information content is greater than 2.0, the image information content is considered acceptable. Otherwise, it is considered a blank image or an image with little information; The information entropy of all images will be recorded and sorted; S3. Based on the top five images in terms of information entropy in step S2, calculate the sum of red, green, and blue pixel values ​​of these images, and compare them with the RGB threshold of the standard image to determine the coloring quality of the image, and record the determination result. S4. Input all images with an information entropy greater than 2.0 into a three-class focusing model. This model can distinguish whether an image is well focused, out of focus, or focused on a layer of impurities. S5. Finally, based on the staining and focusing results in steps S3 and S4, the image quality of the entire sample is determined. If both staining and focusing are performed well, the sample is deemed acceptable and given a corresponding image quality score.

2. The automated image quality assessment method for LD CytoMatrix 20 slide scanning according to claim 1, characterized in that: The method for determining the coloring quality of the image in step S3 is as follows: A1. Use a pre-trained convolutional neural network (CNN) to extract high-dimensional feature vectors for each image. ,in I The image represents the input. f It is the feature extraction function of CNN; A2. Considering the diversity of color spaces, the RGB space is converted into HSV and Lab spaces, and features are extracted respectively. and Then, these features are combined using a weighted fusion strategy to obtain the comprehensive features. : （1）； in, For the weight parameters, satisfying Determined through optimization using cross-validation; A3. Determine the optimal staining threshold using dynamic programming. Based on the statistical characteristics of the current batch of image features, the following formula is used: （2）； in, μ It is the average value of the combined feature vectors. σ It is its standard deviation, and k It is a coefficient adjusted based on historical data to balance sensitivity and specificity; A4. To determine staining quality, compare comprehensive characteristics with threshold values. To classify staining quality and define staining The score is expressed as: （3）； in, λ The slope parameter of the sigmoid function adjusts the steepness of the classification curve. A value ≥0.7 indicates good staining quality, while a value lower than 0.7 indicates poor staining quality.

3. The automated image quality assessment method for LD CytoMatrix 20 slide scanning according to claim 1, characterized in that: The focusing model classification method used in step S4 is as follows: B1. Extracting focus features of images using ResNet residual network. ; B2. Combining staining quality characteristics and focusing features We construct a multi-task learning model to simultaneously optimize the prediction of coloring and focusing states; B3. Calculate the probability distribution of the focusing state using the softmax function. , Represented as: （4）； Where W is the weight matrix and b is the bias term, The concatenated vector represents the features, and the softmax function ensures that the sum of the probabilities is 1. B4. Select the category with the highest probability as the final focusing state. If the probability of the focusing category exceeds 0.8, it is judged as good focusing; if the probability of focusing on the impurity layer is the highest and exceeds 0.6, it is judged as poor focusing.

4. The automated image quality assessment method for LD CytoMatrix 20 slide scanning according to claim 1, characterized in that: The method for assigning the corresponding image quality score in step S5 is as follows: C1. Dyeing quality is assessed through color consistency indicators. Qc To quantify, Qc The metric is based on the difference between the sum of the RGB values ​​of an image and the sum of the RGB values ​​of a standard image. Qc Represented as: （5）； in, Rs , Gs , Bs It is the average RGB value of a standard image, while R , G , B It is the average RGB value of the current image. It is the standard deviation of staining consistency, used to control the sensitivity of the formula; C2. Focused state via focus index To evaluate, The exponent is calculated using the Laplacian operator edge strength and the image entropy. Represented as: （6）； in, It is an image I At point ( x , y The Laplace operator result at ) H ( I () is an image I Information entropy; C3. Overall staining quality and focus state Provide an image quality score The quality score is calculated using a weighted average formula. , Represented as: （7）； in, and These are the weighting factors for staining quality and focusing state, respectively, satisfying... ; The weighting factors are adjusted based on the experimental results to optimize the overall score; C4. Set the acceptable threshold If the overall score Greater than or equal to If the image quality is satisfactory, it is considered acceptable; otherwise, it is deemed unacceptable.

Images