A Method and System for Statistical Analysis of Lysosomal Capacity Based on Morphological Analysis

Abstract

This application relates to the field of lysosome volume statistical analysis technology, and discloses a method and system for lysosome volume statistical analysis based on morphological analysis. The method includes: preparing cell samples into cell sections; using an electron microscope to photograph the cell sections at preset shooting intervals to generate a first image; classifying and identifying lysosomes from the first image, and counting the first number of all lysosomes; measuring the size of any lysosome in the first image set based on morphological analysis, outputting an estimated morphology, and calculating the first volume of the lysosome; calculating the second number and second volume sum corresponding to a new cell section, summing all volumes to generate a volume change sequence, and summing all numbers to generate a number change sequence; constructing an analysis model, and outputting the biological analysis results of lysosomes in the cell sample based on the number change sequence and volume change sequence. This application improves the accuracy and comprehensiveness of lysosome volume statistical analysis.

Description

Technical Field

[0001] This application relates to the field of lysosomal volume statistical analysis technology, and in particular to a lysosomal volume statistical analysis method and system based on morphological analysis. Background Technology

[0002] Lysosomes are important organelles in eukaryotic cells, responsible for breaking down and digesting intracellular substances, and playing a crucial role in maintaining cellular homeostasis and physiological function. However, current methods for statistically analyzing lysosomal volume still have many limitations. While traditional chemical analysis methods can measure enzyme activity or substrate content within lysosomes, they cannot directly reflect lysosomal morphology and quantity information. Existing morphological observation methods, such as optical microscopy, are limited by resolution and cannot clearly display the fine structure of lysosomes. Therefore, developing a method that can intuitively and accurately statistically measure lysosomal volume is of great significance.

[0003] A similar prior art is disclosed in Chinese patent application CN113935983A, which presents a fully automated quantitative analysis method for perinuclear lysosome distribution. The method includes: identifying and extracting an endoplasmic reticulum (ER) region from an endoplasmic reticulum (ER) image, using the ER region to approximate a single-cell region; segmenting the nucleus region in a nucleus image to obtain all nucleus regions; extracting all lysosome regions from a lysosome image; determining the nucleus and lysosome regions within each single-cell region; setting a corresponding perinuclear region based on the actual morphology of the single cell; and determining the lysosome distribution within the perinuclear region of the single cell based on the location and radius of the lysosomes within that region. This invention, based on deep learning technology, designs an automated analysis process, achieving precise quantitative analysis of perinuclear lysosome distribution, solving the current problem of manual calculation, and enabling high-throughput big data analysis. Another Chinese patent application, CN119831987A, discloses a method and system for cell classification and functional assessment based on nucleolar morphology. The method includes: acquiring microscopic image data of a biological sample; acquiring a first region of interest, a second region of interest, and a third region of interest; extracting nucleolar feature data; performing a first classification of cells in the biological sample; performing a second classification of C-type cells in the biological sample; calculating the number and / or proportion of A-type and B-type cells in the biological sample, thereby assessing the functional state of the corresponding tissue or organ. This invention can accurately and sensitively reflect and assess the functional state of the corresponding tissue or organ of a biological sample, and is particularly suitable for biological samples with high heterogeneity and lacking specific biomarkers.

[0004] The main shortcomings of existing technologies are that they only identify the distribution of lysosomes in the perinuclear region of single cells. In cases of dense cell density, cells may overlap, which increases the difficulty of identifying single-cell regions. Using only a single method to assess the functional status of the tissue or organ corresponding to the biological sample is not feasible. In practice, it is necessary to perform overlapping image analysis on cell samples under different conditions to provide an important experimental means to reveal the biological functions and regulatory mechanisms of lysosomes. Summary of the Invention

[0005] This application provides a lysosomal volume statistical analysis method and system based on morphological analysis, which can improve the accuracy and comprehensiveness of lysosomal volume statistical analysis.

[0006] In a first aspect, this application provides a lysosomal volume statistical analysis method based on morphological analysis, the lysosomal volume statistical analysis method based on morphological analysis comprising:

[0007] Cell samples are prepared into cell slices, and the cell slices are photographed using an electron microscope at preset shooting intervals to generate first images. All the first images are summarized based on the preset shooting intervals to generate a first image set.

[0008] The first image is classified and identified to identify lysosomes, and the first number of all the lysosomes is counted. The size of any lysosome in the first image set is measured based on morphological analysis, the estimated morphology is output, and the first volume of the lysosome is calculated. Based on the first number, the first volumes corresponding to the lysosomes in the cell slice are summarized to generate the first volume sum.

[0009] Switch the cell sample preparation method to generate multiple new cell slices, calculate the second quantity and second volume of the new cell slices, summarize all volumes to generate a volume change sequence, and summarize all quantities to generate a quantity change sequence. The volume sum includes the first volume sum and the second volume sum, and the quantity includes the first quantity and the second quantity.

[0010] An analytical model is constructed, which outputs biological analysis results of lysosomes in the cell sample based on the quantity change sequence and the volume change sequence.

[0011] In conjunction with the first aspect, before classifying and recognizing the first image, the first image set is preprocessed, including:

[0012] Two adjacent first images are extracted from the first image set and set as the target image and the reference image, respectively. The recognition positions of the target image and the reference image are identified based on the image segmentation algorithm.

[0013] The identification location in the reference image is mapped to the target image, and a mapping function is generated. The size of the retrieval region is set, and the target image is searched with the identification location as the center and the size of the retrieval region as the size, generating multiple retrieval regions.

[0014] Obtain the first pixel value of the vertex position of any of the search regions, compare the second pixel value of the recognition position of the reference image with all the first pixel values, if the second pixel value belongs to the interval corresponding to the largest first pixel value and the smallest first pixel value, then the blank pixel is 0, if the second pixel value is less than the smallest first pixel value, then the blank pixel is the difference between the second pixel value and the smallest first pixel value, if the second pixel value is greater than the largest first pixel value, then the blank pixel is the difference between the second pixel value and the largest first pixel value;

[0015] Create a blank image with the same size as the reference image, and assign the blank pixels to the corresponding recognition positions in the blank image;

[0016] Repeat this step until all the identified locations in the reference image have been preprocessed, and then set the assigned blank image as the new first image.

[0017] In conjunction with the first aspect, all the new first images form a new first image set based on the preset shooting interval, and lysosomes are identified by classifying the new first images, including:

[0018] Obtain the organelle region and background region from the new first image;

[0019] Extract multiple feature quantities for any of the organelle regions, calculate the average value of all the organelle regions on the feature quantities, set extraction conditions based on the average value, and extract feature objects from the organelle regions based on the extraction conditions;

[0020] A multi-level classification method is set up, and multiple sub-classification models are constructed based on the multi-level classification method. A new first image is extracted, and a category label is set based on the type of the organelle region. The image is then input into any of the sub-classification models. The sub-classification models output the classification accuracy based on the category label corresponding to the feature object. The sub-classification model with the highest classification accuracy is set as the classification recognition model.

[0021] The remaining new first image is input into the classification and recognition model, which outputs the category label corresponding to the feature object. The continuous pixel region contained in the feature object with the category label of lysosome is set as the lysosome region, and the number of the lysosome regions is set as the first number.

[0022] In conjunction with the first aspect, the output estimation form includes:

[0023] The arrangement image of the lysosomes is acquired based on the preset shooting interval, the average length value is set based on the region diameter of the lysosomes, and the standard length value is set based on the average length value.

[0024] After initializing each pixel value in the arranged image, the peak pixel in the arranged image is output based on the filter, and the peak pixel is corresponding to the first center position of the lysosome.

[0025] Using each of the first center positions as the center and the standard length value as the range radius, based on the range radius, pixel positions that are offset in the Z direction relative to the first center positions are extracted in the arranged image, the background pixel value of the arranged image is obtained, and the pixel positions with pixel values ​​greater than the background pixel value are selected and set as the second center positions;

[0026] For any second center location, assign an identification label that is the same as the first center location; sort all the first center locations and all the second center locations based on distance to generate a location list.

[0027] Based on the location list, it is determined whether any center position coincides with the lysosome region in the first image. If so, the center position is deleted from the location list, wherein the center position includes the first center position and the second center position.

[0028] If the center position is the second center position, then determine whether there is a lysosomal region with the same identification tag in the adjacent pixel positions in the Z direction. If not, then delete the center position.

[0029] Extract the extended region corresponding to the center position from the arranged image, and generate a planar lysosome region by binarizing the pixel values ​​of the extended region.

[0030] Repeat this step until all remaining center positions in the position list have been processed to generate multiple planar lysosome regions. All planar lysosome regions are summarized based on the Z direction to generate a three-dimensional lysosome region, and the three-dimensional lysosome region is set as the estimated morphology.

[0031] In conjunction with the first aspect, after uniformly marking all the first images containing lysosomes in the X and Y directions, they are stacked and arranged along the Z direction according to the time sequence of the preset shooting interval to form the arranged images.

[0032] In conjunction with the first aspect, the product of the average length value and the preset coefficient is set as the expansion distance, and the expansion area is generated by expanding outwards from the center position in the X and Y directions with the expansion distance as the center.

[0033] In conjunction with the first aspect, the calculation of the first volume of the lysosome includes:

[0034] Obtain the number of pixels and the area of ​​pixels in any of the planar lysosome regions, and set the product of the pixel area and the number of pixels as the first area of ​​the planar lysosome region;

[0035] The distance interval in the Z direction is obtained by using an electron microscope, and the first volume is generated by calculating the numerical integral of the first area and the distance interval based on the trapezoidal rule.

[0036] In conjunction with the first aspect, the output of biological analysis results of lysosomes in the cell sample includes:

[0037] The analysis model includes a first model and a second model. The quantity change sequence and the volume change sequence are input into the first model, and the sliding window method is used for training. Based on the fabrication method, the lysosome prediction results are output.

[0038] A second model is constructed based on the lysosome prediction results. A biological problem is set, state variables and parameters are defined, change patterns and specific time points are obtained based on the lysosome prediction results, and the biological analysis results are output based on the change patterns and specific time points.

[0039] Secondly, this application provides a lysosomal volume statistical analysis system based on morphological analysis, the lysosomal volume statistical analysis system based on morphological analysis comprising:

[0040] The image acquisition module is used to prepare cell samples into cell slices, use an electron microscope to take pictures of the cell slices at a preset shooting interval to generate a first image, and summarize all the first images based on the preset shooting interval to generate a first image set;

[0041] The morphology analysis module is used to classify and identify lysosomes in the first image, count the first number of all the lysosomes, measure the size of any lysosome in the first image set based on morphology analysis, output the estimated morphology, calculate the first volume of the lysosome, and summarize the first volume corresponding to the lysosomes in the cell slice based on the first number to generate the first volume sum.

[0042] The summarization module is used to switch the cell sample preparation method, generate multiple new cell slices, calculate the second quantity and second volume sum corresponding to the new cell slices, summarize all volume sums to generate a volume change sequence, and summarize all quantities to generate a quantity change sequence. The volume sum includes the first volume sum and the second volume sum, and the quantity includes the first quantity and the second quantity.

[0043] The bioanalysis module is used to output the biological analysis results of lysosomes in the cell sample based on the quantity change sequence and the volume change sequence.

[0044] The technical solution provided in this application firstly preprocesses the first image set, including image segmentation, pixel value comparison, and blank image creation. Multiple sub-classification models are constructed based on a multi-level classification method. Feature extraction and classification are performed using features (such as geometric features, shape features, and size features), accurately identifying lysosomes. This effectively removes noise and unnecessary background information from the images, improving the accuracy of lysosome identification and ensuring the accuracy of lysosome quantity statistics. Then, based on the arranged images and the peak pixels output by the filter, the center position of the lysosome is determined. The three-dimensional morphology of the lysosome is determined through Z-direction offset and pixel value filtering. The first area of ​​the planar lysosome region is calculated using the trapezoidal rule, and the first volume of the lysosome is calculated by combining the distance interval in the Z-direction. This provides a more comprehensive description of the morphological and volume changes of the lysosome, offering a reliable data foundation for subsequent biological analysis, reducing calculation errors, and improving the reliability of the analysis results. Finally, an analytical model was constructed, including a first model and a second model. The sliding window method was used to train the sequence of changes in quantity and volume, and the lysosome prediction results were output. Biological questions were set, state variables and parameters were defined, and biological analysis results were output. This model can reveal the changing trends and patterns of lysosomes under different conditions, and achieve accurate statistical analysis of lysosome capacity and quantity. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a schematic diagram of an embodiment of the lysosome volume statistical analysis method based on morphological analysis in this application.

[0047] Figure 2 This is a schematic diagram of an embodiment of the output of the estimated morphology of lysosomes in this application;

[0048] Figure 3 This is a schematic diagram of an embodiment of the analytical model outputting biological analysis results in this application.

[0049] Figure 4 This is a schematic diagram of an embodiment of the lysosome volume statistical analysis system based on morphological analysis in this application. Detailed Implementation

[0050] This application provides a method and system for statistical analysis of lysosomal volume based on morphological analysis. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0051] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the lysosomal volume statistical analysis method based on morphological analysis in this application includes:

[0052] Step S101: Prepare cell slices from cell samples, take pictures of the cell slices using an electron microscope at preset shooting intervals to generate first images, and summarize all first images based on preset shooting intervals to generate a first image set.

[0053] It is understood that the executing entity of this application can be a lysosomal volume statistical analysis device based on morphological analysis, or it can be a terminal or a server; no specific limitation is made here. This application's embodiments use a server as an example for illustration.

[0054] Specifically, suitable cell samples are selected and subjected to fixation, dehydration, infiltration, and embedding processes to prepare cell sections suitable for electron microscopy observation. During preparation, conditions are strictly controlled to minimize the impact on lysosome morphology and quantity. The cell sections are observed using a high-resolution electron microscope. Magnification and focal length are adjusted to clearly display the morphology and structure of lysosomes. Images of the cell sections containing various organelles are captured intermittently, i.e., first images. These first images are two-dimensional images, and the shooting time is set by a preset shooting interval. Therefore, multiple first images are generated. All first images are sequentially summarized according to the shooting time corresponding to the preset shooting interval to generate a set of first images corresponding to the cell section, i.e., the first image set.

[0055] Step S102: Classify and identify lysosomes in the first image, count the first number of all lysosomes, measure the size of any lysosome in the first image set based on morphological analysis, output the estimated morphology, calculate the first volume of the lysosome, summarize the first volume corresponding to the lysosomes in the cell slice based on the first number, and generate the first volume sum.

[0056] Specifically, since the first image corresponding to a cell slice may contain various organelle structures, such as lysosomes, mitochondria, and centrosomes, it is necessary to classify and identify the lysosomes in the first image. The first quantity refers to the total number of lysosomes contained in the first image. Because the morphology of lysosomes in a cell slice is dynamic and may not be regular, the size of any lysosome can be measured through morphological analysis based on the first image set to obtain the three-dimensional shape of each lysosome, i.e., to estimate its morphology. The first volume refers to the three-dimensional volume value of the lysosome calculated from its estimated morphology. The sum of the first volumes refers to the sum of the first volumes of all lysosomes in the cell slice.

[0057] Step S103: Switch the cell sample preparation method to generate multiple new cell slices, calculate the second quantity and second volume of the new cell slices, summarize all volume sums to generate a volume change sequence, and summarize all quantities to generate a quantity change sequence. The volume sum includes the first volume sum and the second volume sum, and the quantity includes the first quantity and the second quantity.

[0058] Specifically, switching the cell sample preparation method allows for statistical analysis of lysosomes from multiple cell slices or cell samples under different conditions, facilitating the comparison of differences and trends in lysosome capacity. For new cell slices, based on the same processing steps, the corresponding second quantity and second volume sum can be calculated. Since there may be multiple new cell slices, there will also be multiple second quantities and second volume sums. Therefore, based on the changes in the second quantity and second volume sum, quantity change sequences and volume change sequences can be generated. These sequences can describe the differences and trends in lysosome capacity of cell samples prepared using different methods.

[0059] Step S104: Construct an analysis model. The analysis model outputs biological analysis results of lysosomes in cell samples based on the quantitative change sequence and the volume change sequence.

[0060] Specifically, the analytical model refers to selecting an appropriate statistical analysis model or machine learning model to analyze the sequence of changes in quantity and volume, and output the biological analysis results of lysosomes, including but not limited to the trend of changes in the number and volume of lysosomes, the distribution characteristics of lysosomes, and the change patterns of lysosomes under different conditions.

[0061] In one specific embodiment, before classifying and recognizing the first image, the first image set is preprocessed, including:

[0062] (1) Extract two adjacent first images from the first image set and set them as the target image and reference image respectively. Based on the image segmentation algorithm, identify the recognition positions of the target image and reference image respectively.

[0063] (2) Map the identified location in the reference image to the target image and generate a mapping function. Set the size of the search area and search the target image with the identified location as the center and the size of the search area as the size to generate multiple search areas.

[0064] (3) Obtain the first pixel value of the vertex position of any search region, compare the second pixel value of the recognition position of the reference image with all the first pixel values, if the second pixel value belongs to the interval corresponding to the largest first pixel value and the smallest first pixel value, then the blank pixel is 0, if the second pixel value is less than the smallest first pixel value, then the blank pixel is the difference between the second pixel value and the smallest first pixel value, if the second pixel value is greater than the largest first pixel value, then the blank pixel is the difference between the second pixel value and the largest first pixel value.

[0065] (4) Create a blank image with the same size as the reference image and assign the blank pixels to the corresponding recognition positions in the blank image.

[0066] (5) Repeat this step until all identified locations in the reference image have been preprocessed, and set the assigned blank image as the new first image.

[0067] Specifically, preprocessing is required to more clearly classify and identify the first image set. Any two adjacent images in the first image set will exhibit morphological differences; preprocessing can more accurately identify subtle morphological variations. The identification location refers to multiple interest pixels corresponding to each organelle in the target image and reference image. The identification location can represent the pixel region of any organelle, but may not be complete.

[0068] In the target image, the retrieval region is set based on the recognition location in the reference image. Therefore, an image registration method is needed to map the recognition location of the reference image onto the target image. The size of the retrieval region can be set by scaling down the size of the first image so that the shape of the retrieval region is consistent with that of the first image. If the first image is rectangular, the retrieval region will be similar to a small rectangular area. The retrieval region is equivalent to defining the range of pixels to be searched around the recognition location.

[0069] Vertex position refers to the four vertex pixels of the retrieval region. If the second pixel falls within the range of several first pixel values ​​(i.e., the range corresponding to the largest and smallest first pixel values), it indicates that the pixel value at that identification location has no significant difference between the two images. If the second pixel value is less than the smallest first pixel value, it indicates that the identification location is darker in the reference image than in the target image. By calculating the difference (negative value), the degree of reduction in pixel value in the reference image relative to the target image is quantified. If the second pixel value is greater than the largest first pixel value, it indicates that the pixel value at that identification location in the reference image is higher than all pixel values ​​within the corresponding retrieval region in the target image. By calculating the difference (positive value), the degree of increase in pixel value in the reference image relative to the target image is quantified.

[0070] The blank image is used to remove pixels that are not necessary for analysis. The pixels at the identification locations in the blank image are assigned values ​​based on their size. This preprocessing step is repeated for all identification locations to generate a fully assigned blank image, facilitating subsequent classification and identification. Therefore, this assigned blank image is set as the new first image, and the classification and identification of various organelles are performed on this new first image.

[0071] In one specific embodiment, all new first images are combined into a new first image set based on a preset shooting interval, and lysosomes are identified by classifying the new first images, including:

[0072] (1) Obtain the organelle region and background region in the new first image.

[0073] (2) Extract multiple features from any organelle region, calculate the average value of all organelle regions on the features, set extraction conditions based on the average value, and extract feature objects from the organelle region based on the extraction conditions.

[0074] (3) Set up a multi-level classification method, construct multiple sub-classification models based on the multi-level classification method, extract a new first image, and set the category label based on the type of organelle region and input it into any sub-classification model. The sub-classification model outputs the classification accuracy based on the category label corresponding to the feature object, and sets the sub-classification model with the highest classification accuracy as the classification recognition model.

[0075] (4) Input the remaining new first image into the classification and recognition model. The classification and recognition model outputs the category label corresponding to the feature object. The continuous pixel region contained in the feature object with the category label of lysosome label is set as the lysosome region. The number of lysosome regions is set as the first number.

[0076] Specifically, the preprocessed first image contains organelle regions corresponding to various types of organelles. Therefore, according to the above processing steps, the organelle regions and background regions can be separated from the new first image.

[0077] Features are generated through feature extraction, including but not limited to geometric features, shape features, and size features. Since there may be multiple features of the same type, extraction conditions are set by calculating the average value to extract organelle objects, i.e., feature objects, from each organelle region.

[0078] Multilevel classification refers to a set of classification methods based on the number of feature types. For example, the first classification method might be area threshold classification, and the second classification method might be shape type classification. A multilevel classification method is formed by combining the first and second classification methods. Sub-classification models are models used for classification corresponding to the multilevel classification method, such as decision tree models, statistical models, and support vector machine models. Category labels are set according to the type of organelle, including but not limited to labels for lysosomes, mitochondria, and centrosomes. Classification accuracy refers to the classification accuracy of each sub-classification model. Classification accuracy can be calculated through supervised learning of category labels, and the sub-classification model with the highest classification accuracy is set as the classification and recognition model.

[0079] The trained classification and recognition model can accurately analyze the remaining new first images, and then define the continuous pixel regions labeled as lysosomes as lysosome regions. Since discontinuous lysosome regions may belong to the same lysosome or different lysosomes, in order to reduce recognition errors, the discontinuous lysosome regions are first defined as different lysosomes, and the first number can be counted.

[0080] In one specific embodiment, the output estimated form includes:

[0081] (1) Obtain an image of the arrangement of lysosomes based on a preset shooting interval, set an average length value based on the region diameter of the lysosomes, and set a standard length value based on the average length value.

[0082] (2) After initializing each pixel value in the array image, the peak pixel in the array image is output based on the filter, and the peak pixel is assigned to the first center position of the lysosome.

[0083] (3) Taking each first center position as the center and the standard length value as the range radius, extract the pixel positions in the arranged image that are offset in the Z direction relative to the first center position based on the range radius, obtain the background pixel value of the arranged image, filter out the pixel positions with pixel values ​​greater than the background pixel values ​​and set them as the second center positions.

[0084] (4) Mark any second center location with the same identification label as the first center location, sort all first center locations and all second center locations based on distance, and generate a location list.

[0085] (5) Determine whether any center position coincides with the lysosome region in the first image based on the position list. If so, delete the center position from the position list. The center position includes the first center position and the second center position.

[0086] (6) If the center position is the second center position, then determine whether there is a lysosomal region with the same identification label in the adjacent pixel positions in the Z direction. If not, delete the center position.

[0087] (7) Extract the extended region corresponding to the center position from the arranged image, and generate the planar lysosome region by binarizing the pixel values ​​of the extended region.

[0088] (8) Repeat this step until all remaining center positions in the position list are processed to generate multiple planar lysosome regions. Summarize all planar lysosome regions based on the Z direction to generate a three-dimensional lysosome region and set the three-dimensional lysosome region as the estimated morphology.

[0089] Specifically, Figure 2This is a flowchart illustrating the estimated morphology of lysosomes. Since the morphology of individual organelles is dynamic and may be obscured, accurate identification of the three-dimensional lysosomes is necessary. An array image refers to an image formed by stacking new first images of the same cell slice according to the time of capture. Lysosome regions in the new first images have been identified. The region diameter refers to the maximum length of a lysosome within the array image, and the average length value is the average of all region diameters. The standard length value refers to the length of the lysosome in the Z-direction within the array image.

[0090] Initialization processing refers to pixel normalization. A Laplacian Gaussian filter is used to detect and integrate the pixel values ​​in the image, outputting the peak pixel corresponding to the maximum value. There may be multiple peak pixels; the peak pixels with the closest spatial distance are combined into one. The position of the pixels contained in this peak pixel is set as the center position of the lysosome, i.e., the first center position.

[0091] The selection of the second center location is to more accurately identify and locate lysosomes, especially in cases where lysosomes are densely packed or irregularly shaped, providing a more comprehensive description of their shape and location. The Z-direction refers to the stacking direction of the arranged images. A pixel location refers to a region of location offset in the Z-direction with a radius equal to or greater than the first center location. The background pixel value refers to the average pixel value of the background region that does not contain lysosomes. Multiple second center locations are generated by selecting all pixel locations based on the magnitude of the background pixel value.

[0092] Identification tags are used to associate the second center position with the first center position. For example, if the identification tag for the first center position is M1, then the identification tag for the second center position selected based on the first center position is also M1. After labeling the second center positions, they are arranged according to their distance from the referenced first center position to generate a position list.

[0093] In this step, the first image refers to the new first image after the lysosomes have been classified. If the center position coincides with the lysosome region in the first image, it means that the center position overlaps with the already formed two-dimensional lysosome region, and therefore the center position is deleted. The center position includes two types: the first center position and the second center position.

[0094] If the identification labels are different, it means that the second center position corresponding to the Z direction does not belong to the lysosomal region, or there is a discontinuous lysosomal region in the Z direction.

[0095] The corresponding extended region is extracted from the center position in the arranged image, and a planar lysosome region is further generated. The planar lysosome region is a two-dimensional lysosome region without occlusion or overlap. Alternatively, the thresholding of pixel value gradient can be used instead of the binarization of pixel value to improve the recognition accuracy.

[0096] Multiple central locations correspond to multiple planar lysosomal regions. All planar lysosomal regions are stacked and summarized according to the Z direction to generate a three-dimensional lysosome, that is, a three-dimensional lysosomal region, which can completely describe the estimated morphology of each lysosome in the cell slice.

[0097] In one specific embodiment, after uniformly marking the X and Y directions of all first images containing lysosomes, they are stacked and arranged along the Z direction based on the time sequence of a preset shooting interval to form an arranged image.

[0098] Specifically, the X and Y directions represent the planar directions corresponding to the first image. Since multiple first images are generated by continuous shooting at the same preset shooting interval, the dynamic changes of lysosomes in the first image are related to the shooting order of each first image. Therefore, by stacking all the first images along the Z direction, an arranged image can be generated.

[0099] In one specific embodiment, the product of the average length value and a preset coefficient is set as the expansion distance, and the expansion area is generated by expanding outwards from the center position in the X and Y directions with the expansion distance as the center.

[0100] Specifically, the preset coefficients can be set according to the slicing parameters.

[0101] In one specific embodiment, calculating the first volume of the lysosome includes:

[0102] (1) Obtain the number of pixels and the area of ​​any planar lysosome region, and set the product of the pixel area and the number of pixels as the first area of ​​the planar lysosome region.

[0103] (2) Obtain the distance interval in the Z direction based on the electron microscope imaging method, calculate the numerical integral of the first area and the distance interval based on the trapezoidal rule, and generate the first volume.

[0104] Specifically, the formula for calculating the first volume V is: Where I is the total number of first images in the first image set. For distance interval, Let be the first area of ​​the planar lysosome region in the i-th first image. It is the first area of ​​the planar lysosome region in the (i+1)th first image.

[0105] In one specific embodiment, the biological analysis results of lysosomes in the cell sample are output, including:

[0106] (1) The analysis model includes a first model and a second model. The quantity change sequence and volume change sequence are input into the first model and trained using the sliding window method. The lysosome prediction results are output based on the production method.

[0107] (2) Construct a second model based on the lysosome prediction results, set biological questions, define state variables and parameters, obtain change patterns and special time points based on the lysosome prediction results, and output biological analysis results based on the change patterns and special time points.

[0108] Specifically, Figure 3 The flowchart illustrates the output of biological analysis results from the analytical model. The first model is an ARIMA-LSTM (Autoregressive Integral Moving Average-Long Short-Term Memory) hybrid model, capable of handling both linear and non-linear dependencies in quantity and volume change sequences. The quantity and volume change sequences are input into the first model and trained inversely. The sliding window method uses past data as input features to help the first model learn temporal dependencies, thereby improving prediction accuracy. Lysosome prediction results include both quantity and volume change outcomes.

[0109] The second model refers to a biological model, where the biological problem includes, but is not limited to, the target problem corresponding to the dynamic changes of lysosomes such as generation, degradation, fusion, and autophagy. State variables refer to the key variables to be described in the second model, such as the number, volume, and enzyme activity of lysosomes. Parameters include control parameters (such as temperature and pH) and internal parameters (generation rate, degradation rate, fusion rate). Change patterns refer to the fluctuation patterns of the lysosome prediction results, and specific time points refer to key time points where a significant trend of change exists. The biological analysis results are the biological conclusions output by the second model based on the biological problem and its predictions of lysosomes.

[0110] The above describes the lysosomal volume statistical analysis method based on morphological analysis in the embodiments of this application. The following describes the lysosomal volume statistical analysis system based on morphological analysis in the embodiments of this application. Please refer to [link to relevant documentation]. Figure 4 One embodiment of the lysosomal volume statistical analysis system based on morphological analysis in this application includes:

[0111] The image acquisition module 201 is used to prepare cell samples into cell slices, use an electron microscope to take pictures of the cell slices at preset shooting intervals to generate first images, and summarize all the first images based on the preset shooting intervals to generate a first image set.

[0112] The morphology analysis module 202 is used to classify and identify lysosomes in the first image, count the first number of all lysosomes, measure the size of any lysosome in the first image set based on morphology analysis, output the estimated morphology, calculate the first volume of the lysosome, and summarize the first volume corresponding to the lysosomes in the cell slice based on the first number to generate the first volume sum.

[0113] The summary module 203 is used to switch the cell sample preparation method, generate multiple new cell slices, calculate the second quantity and second volume sum corresponding to the new cell slices, summarize all volume sums to generate a volume change sequence, and summarize all quantities to generate a quantity change sequence. The volume sum includes the first volume sum and the second volume sum, and the quantity includes the first quantity and the second quantity.

[0114] The bioanalysis module 204 is used to output the biological analysis results of lysosomes in cell samples based on the quantitative change sequence and the volume change sequence.

[0115] Through the collaborative efforts of the aforementioned components, the first image set is preprocessed, including image segmentation, pixel value comparison, and blank image creation. Multiple sub-classification models are constructed based on a multi-level classification approach. Feature extraction and classification are performed using features (such as geometric, shape, and size features), accurately identifying lysosomes and effectively removing noise and unnecessary background information from the images, thus improving the accuracy of lysosome identification and ensuring the accuracy of lysosome quantity statistics. Then, based on the arrangement of images and the peak pixels output by the filter, the center position of the lysosome is determined. The three-dimensional morphology of the lysosome is determined through Z-direction offset and pixel value filtering. The first area of ​​the planar lysosome region is calculated using the trapezoidal rule, and combined with the distance interval in the Z-direction, the first volume of the lysosome is calculated. This provides a more comprehensive description of the morphological and volume changes of the lysosome, offering a reliable data foundation for subsequent biological analysis, reducing computational errors, and improving the reliability of the analysis results. Finally, an analytical model was constructed, including a first model and a second model. The sliding window method was used to train the sequence of changes in quantity and volume, and the lysosome prediction results were output. Biological questions were set, state variables and parameters were defined, and biological analysis results were output. This model can reveal the changing trends and patterns of lysosomes under different conditions, and achieve accurate statistical analysis of lysosome capacity and quantity.

[0116] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0117] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0118] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A statistical analysis method for lysosomal capacity based on morphological analysis, characterized in that, The lysosomal volume statistical analysis method based on morphological analysis includes: Cell samples are prepared into cell slices, and the cell slices are photographed using an electron microscope at preset shooting intervals to generate first images. All the first images are summarized based on the preset shooting intervals to generate a first image set. The first image is classified and identified to identify lysosomes, and the first number of all the lysosomes is counted. The size of any lysosome in the first image set is measured based on morphological analysis, the estimated morphology is output, and the first volume of the lysosome is calculated. Based on the first number, the first volumes corresponding to the lysosomes in the cell slice are summarized to generate the first volume sum. Switch the cell sample preparation method to generate multiple new cell slices, calculate the second quantity and second volume of the new cell slices, summarize all volumes to generate a volume change sequence, and summarize all quantities to generate a quantity change sequence. The volume sum includes the first volume sum and the second volume sum, and the quantity includes the first quantity and the second quantity. An analytical model is constructed, which outputs biological analysis results of lysosomes in the cell sample based on the quantity change sequence and the volume change sequence.

2. The lysosomal capacity statistical analysis method based on morphological analysis according to claim 1, characterized in that, Before classifying and recognizing the first image, the first image set is preprocessed, including: Two adjacent first images are extracted from the first image set and set as the target image and the reference image, respectively. The recognition positions of the target image and the reference image are identified based on the image segmentation algorithm. The recognition position refers to multiple interest pixels corresponding to each organelle in the target image and the reference image. The recognition position represents the pixel region of any organelle. The identification location in the reference image is mapped to the target image, and a mapping function is generated. The size of the retrieval region is set, and the target image is searched with the identification location as the center and the size of the retrieval region as the size, generating multiple retrieval regions. Obtain the first pixel value of the vertex position of any of the search regions, where the vertex position refers to the four vertex pixels of the search region. Compare the second pixel value of the recognition position of the reference image with all the first pixel values. If the second pixel value belongs to the interval corresponding to the largest and smallest first pixel values, then the blank pixel is 0. If the second pixel value is less than the smallest first pixel value, then the blank pixel is the difference between the second pixel value and the smallest first pixel value. If the second pixel value is greater than the largest first pixel value, then the blank pixel is the difference between the second pixel value and the largest first pixel value. Create a blank image with the same size as the reference image, and assign the blank pixels to the corresponding recognition positions in the blank image; Repeat this step until all the identified locations in the reference image have been preprocessed, and then set the assigned blank image as the new first image.

3. The lysosomal capacity statistical analysis method based on morphological analysis according to claim 2, characterized in that, All the new first images are combined into a new first image set based on the preset shooting interval. Lysosomes are then classified and identified from the new first images, including: Obtain the organelle region and background region from the new first image; Extract multiple feature quantities for any of the organelle regions, calculate the average value of all the organelle regions on the feature quantities, set extraction conditions based on the average value, and extract feature objects from the organelle regions based on the extraction conditions; A multi-level classification method is set up, and multiple sub-classification models are constructed based on the multi-level classification method. A new first image is extracted, and a category label is set based on the type of the organelle region. The image is then input into any of the sub-classification models. The sub-classification models output the classification accuracy based on the category label corresponding to the feature object. The sub-classification model with the highest classification accuracy is set as the classification recognition model. The remaining new first image is input into the classification and recognition model, which outputs the category label corresponding to the feature object. The continuous pixel region contained in the feature object with the category label of lysosome is set as the lysosome region, and the number of the lysosome regions is set as the first number.

4. The lysosomal capacity statistical analysis method based on morphological analysis according to claim 3, characterized in that, The output estimation form includes: The lysosome arrangement image is acquired based on the preset shooting interval. An average length value is set based on the region diameter of the lysosome, and a standard length value is set based on the average length value. The arrangement image refers to the image formed by stacking the various new first images of the same cell slice according to the shooting time. The region diameter refers to the maximum length of the lysosome in the arrangement image. The average length value refers to the average value of the diameters of all regions. The standard length value refers to the length of the lysosome in the Z direction in the arrangement image. The Z direction refers to the stacking direction of the arrangement image. After initializing each pixel value in the arranged image, the peak pixel in the arranged image is output based on the filter, and the peak pixel is corresponding to the first center position of the lysosome. Using each of the first center positions as the center and the standard length value as the range radius, based on the range radius, pixel positions that are offset in the Z direction relative to the first center positions are extracted in the arranged image, the background pixel value of the arranged image is obtained, and the pixel positions with pixel values ​​greater than the background pixel value are selected and set as the second center positions; For any second center location, assign an identification label that is the same as the first center location; sort all the first center locations and all the second center locations based on distance to generate a location list. Based on the location list, it is determined whether any center position coincides with the lysosome region in the first image. If so, the center position is deleted from the location list, wherein the center position includes the first center position and the second center position. If the center position is the second center position, then determine whether there is a lysosomal region with the same identification tag in the adjacent pixel positions in the Z direction. If not, then delete the center position. Extract the extended region corresponding to the center position from the arranged image, and generate a planar lysosome region by binarizing the pixel values ​​of the extended region. Repeat this step until all remaining center positions in the position list have been processed to generate multiple planar lysosome regions. All planar lysosome regions are summarized based on the Z direction to generate a three-dimensional lysosome region, and the three-dimensional lysosome region is set as the estimated morphology.

5. The lysosomal capacity statistical analysis method based on morphological analysis according to claim 4, characterized in that, After uniformly labeling all the first images containing lysosomes with the X and Y directions, where the X and Y directions respectively represent the planar directions corresponding to the first images, they are stacked and arranged along the Z direction based on the time sequence of the preset shooting interval to form the arranged images.

6. The lysosomal capacity statistical analysis method based on morphological analysis according to claim 5, characterized in that, The product of the average length value and a preset coefficient is set as the expansion distance. The expansion area is generated by expanding outwards from the center position in the X and Y directions with the expansion distance as the center.

7. The lysosomal capacity statistical analysis method based on morphological analysis according to claim 4, characterized in that, The calculation of the first volume of the lysosome includes: Obtain the number of pixels and the area of ​​pixels in any of the planar lysosome regions, and set the product of the pixel area and the number of pixels as the first area of ​​the planar lysosome region; The distance interval in the Z direction is obtained by using an electron microscope, and the first volume is generated by calculating the numerical integral of the first area and the distance interval based on the trapezoidal law.

8. The lysosomal volume statistical analysis method based on morphological analysis according to claim 1, characterized in that, The output of biological analysis results of lysosomes in the cell sample includes: The analysis model includes a first model and a second model. The quantity change sequence and the volume change sequence are input into the first model, and the sliding window method is used for training. Based on the fabrication method, the lysosome prediction results are output. A second model is constructed based on the lysosome prediction results. A biological problem is set, state variables and parameters are defined, change patterns and specific time points are obtained based on the lysosome prediction results, and the biological analysis results are output based on the change patterns and specific time points.

9. A lysosomal volume statistical analysis system based on morphological analysis, characterized in that, The lysosomal volume statistical analysis system based on morphological analysis includes: The image acquisition module is used to prepare cell samples into cell slices, use an electron microscope to take pictures of the cell slices at a preset shooting interval to generate a first image, and summarize all the first images based on the preset shooting interval to generate a first image set; The morphology analysis module is used to classify and identify lysosomes in the first image, count the first number of all the lysosomes, measure the size of any lysosome in the first image set based on morphology analysis, output the estimated morphology, calculate the first volume of the lysosome, and summarize the first volume corresponding to the lysosomes in the cell slice based on the first number to generate the first volume sum. The summarization module is used to switch the cell sample preparation method, generate multiple new cell slices, calculate the second quantity and second volume sum corresponding to the new cell slices, summarize all volume sums to generate a volume change sequence, and summarize all quantities to generate a quantity change sequence. The volume sum includes the first volume sum and the second volume sum, and the quantity includes the first quantity and the second quantity. The bioanalysis module is used to output the biological analysis results of lysosomes in the cell sample based on the quantity change sequence and the volume change sequence.

Images