Method for segmenting white blood cell image

An image segmentation and white blood cell technology, applied in image analysis, image enhancement, image data processing, etc., can solve the problems of slow algorithm speed, sensitivity to glitches and noise, and inability to fully realize morphological segmentation of cells. Guaranteed accuracy

Inactive Publication Date: 2011-05-04
HUAZHONG UNIV OF SCI & TECH
5 Cites 40 Cited by

AI-Extracted Technical Summary

Problems solved by technology

[0005] Disadvantages: Since the selection of seed points is through continuous erosion or through distance transformation first and then a certain domain value selection, it is very sensitive to weak edges
[0008] Disadvantages: The rules for region merging are difficult to determine, which may lead to over-segmentation
[0011] Disadvantages: Since the selection of the concave point is based on the contour, the algorithm is extremely sensitive to burrs and noise on the contour
[0014] Disadvantages: The cells in the actual situation are often not round, the red blood cells are hollow, and the monocytes often have vacuoles, which will lead to segmentation errors; in addition, the adhesion of white blood cells and ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Abstract

The invention provides a method for segmenting a white blood cell image, which is characterized by comprising the following steps: firstly carrying out binaryzation on the green component picture of a color white blood cell image to obtain an initial interested region based on a significance attention mechanism of human vision; carrying out cancellation and mergence on a region by labeling the initial interested region to obtain an adaptive significance window of each cell; and finally realizing segmentation of nucleuses and cytoplasm in each adaptive significance window by a boundary-extending method. The segmentation method is used to realize the fast segmentation of the white blood cells, particularly the accurate segmentation for high-capacity pictures which are the non-standard dyed and contain overlapped cells.

Application Domain

Technology Topic

Cytoplasmic StructureSelf adaptive +7

Examples

  • Experimental program(1)

Example Embodiment

[0032] The present invention will be further described in detail below in conjunction with specific embodiments.
[0033] A white blood cell image segmentation method includes the following steps:
[0034] (1) Extraction of the initial region of interest (ROI):
[0035] ① Take the green component map of the color image of stained white blood cells as the input image.
[0036] ②The input image is reduced in resolution.
[0037] The size of the resolution reduction multiple dRate depends on the median cellSize of the length and width of the smallest cell in the image. In this example, dRate=Round(cellSize/10), and Round() is the rounding function.
[0038] ③ Count the histogram of the input image after resolution reduction.
[0039] ④ Obtain the gray level Hmax where the nth pixel of the histogram is located;
[0040] n=pRate×(input image length/dRate)×(input image width/dRate), pRate is the ratio of the cell pixel to the entire input image pixel, which can be obtained by prior knowledge.
[0041] ⑤Find the trough H of the histogram between the gray value 0 and Hmax;
[0042] ⑥ Binarize the green component image of the color image with the trough H as the threshold to obtain the initial ROI;
[0043] (2) Obtain the cell adaptive salient window:
[0044] ①Mark ROI: remove the initial ROI that is too large and too small, and then label the remaining area (i.e. ROI). The labels of different connected regions are arranged in increments of position to obtain the number of labels. Oversized and undersized areas refer to areas larger than a first size and smaller than a second size respectively, which can be obtained from prior knowledge. In this embodiment, the first size is set to cellSize, and the second size is set to 0.01×cellSize , That is, the connected area smaller than 0.01×cellSize is too small, and the area larger than cellSize is too large. Count the number of regions and calculate the average region size. Marks larger than the average area size are candidate ROIs (AROI), marks smaller than one-tenth of the average area size are non-ROIs (CROI), and the remaining marks are possible ROIs (BROI).
[0045] ②Expand all ROIs, and check whether there is overlap between each AROI or BROI expansion area (ie detection area) and the adjacent detected area (including BROI expansion area and/or CROI expansion area). If there is overlap, then The label of the overlapping detected area is changed to the label of the detected area until all ROIs are detected. The labels are re-sorted in ascending order starting from 1, the largest label is the number of adaptive salient windows, the cell location marker map is obtained, and each ROI is the label area.
[0046] ③For each area with the same label (may contain multiple connected areas), look for a rectangular area with a horizontal or vertical side tangent to the irregular area (that is, the smallest rectangle that can contain the labeled area, whose side length is Horizontal and vertical), the length and width of the rectangle are obtained, and a square area with a side length of cellSize×2+1 is obtained with the center point of the rectangle as the center. Thus, a set of cell adaptive salient windows is obtained.
[0047] (3) Obtain the precise cell area in the adaptive saliency window
[0048] ① Obtain the precise nuclear area and initial cytoplasmic area in the adaptive saliency window:
[0049] Extract color information for each adaptive salient window and mark the label area. Using the Otsu double-threshold method for the green component label map of each window, the thresholds of nucleus and cytoplasm and the threshold of cytoplasm and background are obtained. At the same time, the corresponding pixels are classified into three categories: nucleus, cytoplasm and background. Fill the cytoplasm area with holes, that is, if the inner area of ​​the cytoplasm contains background dots, then such background dots are converted to cytoplasm dots.
[0050] ②Get the seed area in the adaptive window:
[0051] a. Perform distance transformation on the cell area of ​​each adaptive saliency window (the precise nucleus area and the initial cytoplasm area obtained in the above steps). In this example, the distance transformation method uses a 3×3 template chamfer distance transformation. Find the maximum distance point for each cell area (each area may have more than one) and get the maximum distance.
[0052] b. Use a circle with a radius of a quarter of the maximum distance to expand each point at the maximum distance to obtain a seed area within the cell area and whose shape is similar to the shape of the cell.
[0053] ③Get the precise cytoplasmic area in the adaptive saliency window:
[0054] a. Extract the contour of the seed region as the initial boundary, and calculate the boundary length;
[0055] b. Search for the neighboring points of each point on the current boundary. If the neighboring points are in the cell area, mark them as peripheral boundary points;
[0056] c. After searching for the neighboring points of each current boundary point, if the outer boundary is not a temporary boundary, and the length is not less than one-third of the current boundary length or the outer boundary contains the nucleus area, the outer boundary is marked as the current boundary and return Go to step b; otherwise go to step d.
[0057] d. If there is no temporary boundary, the current outer boundary is marked as a temporary boundary and used as the current boundary, and return to step b. If there is a temporary boundary, and the length of the current boundary is greater than the temporary boundary, delete the temporary boundary and the current boundary, and execute step e; if the length of the current boundary is less than the temporary boundary, delete the temporary boundary and return to step b.
[0058] e. Stop expanding the boundary. At this time, all the border areas and the original seed area together constitute a precise cell area. Excluding the marked precise nuclear area, the rest is the precise cytoplasmic area.
[0059] So far, the precise nucleus area and plasma area of ​​each cell in the color cell image have been obtained, and all the steps of white blood cell adhesion image segmentation have been completed.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Similar technology patents

Classification and recommendation of technical efficacy words

  • Guaranteed accuracy
  • Reduce complexity

Identifying software execution behavior

ActiveUS20130185798A1Reduce complexityDecompilation/disassemblyMemory loss protectionSoftware executionComputer program
Owner:ACCESSDATA GRP INC

Thin client intelligent transportation system and method for use therein

InactiveUS20080114530A1Reduce complexityElimination of map integrity issueAnalogue computers for vehiclesAnalogue computers for trafficMap matchingIntelligent transportation system
Owner:PETRISOR GREGORY C +2

Non-systematic coded error correction

ActiveUS20060248434A1Facilitate detection and correctionReduce complexityStatic storageRedundant data error correctionData bitsData security
Owner:OVONYX MEMORY TECH LLC
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products