Image analysis method for extracting local density distribution features of collagen fibers based on pixel intensity changes

Abstract

The present invention discloses an image analysis method for extracting characteristics of local density distribution of collagen fibers based on pixel intensity changes; the method includes optical imaging, binarization and window filtering steps of collagen fibers, wherein optical imaging is achieved by second harmonic generation microscopic imaging, and threshold filtering is used in binarization and window filtering to obtain the local density distribution matrix of collagen fibers, and the corresponding false color coding can vividly and intuitively reflect the density distribution of collagen fibers at the pixel level. The present invention can reflect the two-dimensional local density distribution and overall density of collagen fibers, and has the advantages of being fast, concise, efficient, and accurate, and has the advantages of multiple visual effects output presentation such as two-dimensional analysis, three-dimensional analysis, and one-dimensional information in the depth direction. In addition, the present invention also has a strong adaptability to adapt to a variety of fibrous structure analyses and to be combined with a variety of subsequent data analysis methods.

Description

Technical Field

[0001] The invention belongs to the technical field of quantitative characterization of biological tissues, and in particular relates to an image analysis method for extracting local density distribution characteristics of collagen fibers based on pixel intensity changes. Background Art

[0002] Second Harmonic Generation (SHG) microscopy technology can achieve label-free imaging of collagen fibers with high nonlinear polarizability and asymmetric central structure. In addition, SHG has the advantages of strong penetration, non-invasiveness, high signal-to-noise ratio and high resolution, so it has an inherent advantage in imaging collagen fibers.

[0003] The results of second harmonic generation microscopy of extracellular collagen fibers contain rich information about the spatial structure and distribution of collagen fibers. The spatial distribution of collagen fibers has many characteristics, such as orientation order, curvature and density. Changes in these characteristics are often closely related to many physiological changes such as damage and healing of biological tissues, occurrence and evolution of diseases, and development and spread of cancer. Therefore, this technology has been widely used in the fields of biomedicine and life sciences as an auxiliary tool.

[0004] At present, the analysis based on collagen fiber images mainly relies on manual qualitative evaluation by experienced pathologists, which has disadvantages such as insufficient detail identification and high chance. However, automated quantitative analysis technology for auxiliary diagnosis is relatively rare. Therefore, it is urgent to develop accurate and efficient quantitative analysis technology for the spatial distribution of collagen fibers. Analysis based on collagen fiber imaging results is of great significance in multidisciplinary fields such as optics, biology and medicine.

[0005] Some parameters have been proposed in the field of quantitative characterization of fibrous structures. Among them, the fiber density value based on threshold filtering can simply reflect the collagen density of a single imaging result. However, this single-value indicator is too rough and ignores the detailed information of most collagen fiber distribution contained in the image. It is especially difficult to use to identify collagen fiber images with similar density values ​​but different distribution characteristics. Relatively speaking, the local density parameter distribution can more accurately reflect the spatial structural characteristics of the dense distribution of collagen fibers. Therefore, it is necessary to develop a method to obtain the local density distribution of collagen fibers at the pixel resolution level. Summary of the invention

[0006] In view of the defects existing in the above technical solutions, the present invention provides an image analysis method for extracting the local density distribution features of collagen fibers based on pixel intensity changes. The method can reflect the two-dimensional local density distribution of collagen fibers, has the advantages of being fast, concise, efficient and accurate, and has the advantages of multiple visual effects output presentation such as two-dimensional analysis, three-dimensional analysis and one-dimensional information in the depth direction.

[0007] The present invention is implemented by the following technical solutions:

[0008] An image analysis method for extracting collagen fiber local density distribution characteristics based on pixel intensity changes comprises the following steps:

[0009] 1) Imaging the collagen fibers in the extracellular matrix using second harmonic generation microscopy to obtain a series of two-dimensional grayscale original images;

[0010] 2) For the two-dimensional grayscale original image, the pixel intensity threshold filtering method is used to identify pixels greater than the initial threshold as the collagen fiber area and assign a value of 1, and pixels less than the initial threshold are identified as the background area and assigned a value of 0 to obtain a binary image. The initial threshold is determined based on experience;

[0011] 3) Take a square window centered on any target pixel in the binary image, calculate the ratio of the number of collagen fiber pixels in the window to the total number of pixels in the window, record it as the local density of the central pixel, and traverse all pixels to obtain the local density distribution matrix. Further averaging the local density can reflect the overall density of collagen fibers.

[0012] In the above technical solution, further, the square window centered at any target pixel in the binary image is a square window of equal size with an odd number of pixels, and a repeated border filling window is used for the image boundary pixels;

[0013] Furthermore, the size of the output local density distribution matrix is ​​the same as the size of the original image, and the pixel values ​​are normalized and distributed between 0 (sparse local density) and 1 (dense local density); the local density distribution matrix is ​​subjected to false color coding to generate a local density distribution image, wherein pixels with a value of 0 are blue and pixels with a value of 1 are red;

[0014] The beneficial effects of the present invention are:

[0015] The present invention obtains the collagen fiber local density distribution matrix and its false color coded image by analyzing the second harmonic generation imaging results of collagen fibers. The present invention effectively extracts high-precision local density distribution characteristics of collagen fibers, and the output collagen fiber local density distribution matrix can directly distinguish collagen fiber images with similar density values ​​and different local density detail distributions, becoming a powerful tool for distinguishing collagen fiber density distribution. The output pixel-level collagen fiber local density distribution image can intuitively and vividly reflect the characteristics of collagen fiber local density distribution. The local density distribution matrix can provide a data basis for further analyzing and comparing the local density distribution of collagen fibers in the area of ​​interest or the specified sub-area in the collagen fiber image, and also provide technical support for high-precision analysis of local density changes with depth in three-dimensional conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the process of the present invention;

[0017] Figure 2 It is a schematic diagram of each link of the quantitative analysis process of the present invention in combination with the two-dimensional image results of the second harmonic generation microscopy of collagen fibers;

[0018] Figure 3 The present invention relates to a schematic diagram of the difference in the second harmonic three-dimensional images of collagen fibers of the quantitative analysis of normal tissue, benign tumor tissue and malignant tumor of the mouse ovary; as well as a schematic diagram of the analysis of the local density distribution images at different depths in the three-dimensional image stack and the change of the local density average value at different depths with the imaging depth. DETAILED DESCRIPTION

[0019] In order to describe the present invention more specifically, the technical solution of the present invention is described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0020] The image analysis method for extracting the local density distribution characteristics of collagen fibers based on pixel intensity changes of the present invention specifically comprises the following steps: Figure 1 Shown: Optical imaging of collagen fibers, binarization, and window filtering steps.

[0021] Combined with the specific collagen fiber second harmonic generation microscopy imaging results, the details of each step are described as follows Figure 2 As shown:

[0022] The imaging result of collagen fibers is the original two-dimensional grayscale image. First, intensity filtering is performed through a set threshold to obtain a binary image, in which the area with collagen fibers is identified as 1 value (white) and the background area is identified as 0 value (black). The threshold can be adjusted based on experience according to the imaging quality and is generally taken as an integer fraction of the maximum pixel value of the original image.

[0023] On the basis of the binary image, a neighborhood window calculation is performed for each pixel to obtain the local density value of collagen fibers in the neighborhood of the pixel. For any pixel in the binary image, a square window of equal size centered on it is taken, and the window size is an odd number of pixels. The ratio of 1-valued pixels in the window to the total pixels in the window is calculated, which is the local density of collagen fibers in the space corresponding to the pixel. For boundary pixels, that is, pixels whose calculation window exceeds the image boundary, a repeated filling method is used in the window. After traversing all pixels, the normalized local density distribution matrix of collagen fibers can be obtained, and an intuitive local density distribution map of collagen fibers is further generated through false color coding (the area with sparse collagen fibers is blue and the area with dense collagen fibers is red).

[0024] Figure 3 The fiber density analysis results of the second harmonic three-dimensional imaging of collagen fibers of normal tissue, benign tumor tissue and malignant tumor of mouse ovary, the false color coded images of local density distribution at different depths in the image stack and the line graph of the change of local density average value with imaging depth. This shows that the method proposed in the present invention can still maintain strong robustness when extended to the application of three-dimensional image stacks. The successful application of three-dimensional data correlation analysis provides more vivid and clear data for exploring the tissue morphology and fiber distribution of normal tissue and tumor tissue of mouse ovary. In further analysis, more intuitive and rich information on the distribution of collagen fibers in mouse ovarian tissue can be obtained through histograms and depth change graphs; by comparing with the synchronous analysis of the same parameters of healthy tissue, the rapid and accurate extraction of the changes in the distribution characteristics of collagen fibers around mouse ovarian tumors at different stages of evolution can be achieved, which provides a basis for in-depth understanding of the generation, evolution and deterioration mechanism of tumors, and will also enrich the theoretical basis for subsequent treatment processes.

[0025] The above examples show that the present invention realizes the extraction of collagen fiber density features by analyzing the pixel intensity changes. The local density distribution matrix newly defined by window filtering generates a local density distribution map after false color coding, which can quickly, intuitively and accurately reflect the density of collagen fiber distribution at the pixel level. The corresponding false color coded image also greatly improves the visual effect of the output result, and provides a powerful tool for further exploring the fiber distribution differences of different tissues. The present invention has the advantages of being fast, concise, efficient, and accurate, and has the advantages of multiple visual effect output presentations such as two-dimensional analysis, three-dimensional analysis, and one-dimensional information in the depth direction, and has a strong adaptability combined with subsequent multiple data analysis methods.

[0026] The above description of the examples is to facilitate the understanding and application of the present invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to the above embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments, and improvements and modifications made by those skilled in the art based on the disclosure of the present invention should be within the scope of protection of the present invention.

Claims

1. An image analysis method for extracting collagen fiber local density distribution characteristics based on pixel intensity changes, characterized in that: The following steps are involved: 1) Imaging the collagen fibers in the extracellular matrix using second harmonic generation microscopy to obtain a series of two-dimensional grayscale original images; 2) For the two-dimensional grayscale original image, the pixel intensity threshold filtering method is used to identify pixels greater than the initial threshold as the collagen fiber area and assign a value of 1, and pixels less than the initial threshold are identified as the background area and assigned a value of 0 to obtain a binary image. The initial threshold is determined based on experience; 3) Take a square window centered on any target pixel in the binary image, calculate the ratio of the number of collagen fiber pixels in the window to the total number of pixels in the window, record it as the local density of the central pixel, and traverse all pixels to obtain the local density distribution matrix.

2. The image analysis method for extracting collagen fiber local density distribution characteristics based on pixel intensity changes according to claim 1, characterized in that: In step 3), the square window centered at any target pixel in the binarized image is a square window of equal size with an odd number of pixels, and a repeated border filling window is used for the image boundary pixels.

3. The image analysis method for extracting collagen fiber local density distribution characteristics based on pixel intensity changes according to claim 1, characterized in that: The size of the local density distribution matrix is ​​the same as that of the original image, and the pixel values ​​are normalized and distributed between 0 and 1, where 0 represents sparse local density and 1 represents dense local density; the local density distribution matrix generates a local density distribution image through false color coding, where pixels with a value of 0 are blue and pixels with a transition value of 1 are red.

Images