Bone Marrow White Blood Cell Localization via Saturation Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for localizing bone marrow white blood cells in medical images face challenges due to varying cell densities and color distributions, especially when cells overlap with darker red blood cells, limiting their effectiveness in diverse scenarios.
Innovation Solution
A method utilizing median filtering, conversion to HSV color space, K-means clustering, decision tree algorithms, and morphological processing to accurately select white blood cell areas, combining these techniques to enhance precision and adaptability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If threshold algorithms (e.g., Ostu) are used to segment white blood cells, then the method is simple and fast, but it fails when white blood cell color distribution overlaps with red blood cells or when cell density varies significantly
Solution Approach 1:
The patent transforms the image from RGB color space to HSV color space, changing the parameter representation of cell colors. This allows selective use of the Saturation channel for clustering, which enhances the distinction between white blood cells and red blood cells while maintaining algorithmic simplicity.
Solution Approach 2:
The patent applies K-means clustering to segment the Saturation channel into multiple regions, then selects specific clusters (P1 or P1+P2) that correspond to white blood cell areas. This segmentation approach overcomes the limitations of global thresholding by locally adapting to different color distributions.
2Adaptability or versatility
If clustering algorithms are applied to segment white blood cells, then adaptability to different color distributions improves, but performance degrades when color distribution is scattered or cells overlap with darker red blood cells
Solution Approach 1:
The patent changes from using all three RGB channels to specifically using the Saturation channel for clustering. This parameter selection addresses the problem of scattered color distributions by focusing on the saturation property that best distinguishes white blood cells from red blood cells in the given context.
Solution Approach 2:
The patent applies median filtering before clustering to preliminarily remove noise from the image. This preliminary action improves the quality of input data for clustering, enhancing performance even when cells overlap or color distribution is scattered.
3Device complexity
If global thresholding is applied to separate white blood cells from background and red blood cells, then the assumption of equal area distribution simplifies the process, but it fails when white blood cell proportion varies significantly or colors overlap
Solution Approach 1:
The patent changes from global thresholding in RGB space to local clustering in HSV Saturation space. This parameter transformation maintains relatively simple processing while significantly improving adaptability to diverse scenarios including varying cell proportions and color overlaps.
Solution Approach 2:
The patent introduces dynamic selection of clustering results (P1 or P1+P2) based on image characteristics. This dynamic approach allows the algorithm to adapt to different scenarios while maintaining a unified framework, overcoming the rigidity of global thresholding.
Data Source
AI summary
A saturation clustering-based method for positioning bone marrow white blood cells: first, pre-processing a bone marrow white blood cell image to eliminate partial noise points and simultaneously smooth the image; using K-means clustering to cluster saturation channels of the bone marrow white blood cell image, and select the type of the white blood cells according to a decision tree algorithm; next, eliminating irrelevant areas in a binary image of the white blood cells by means of a morphology processing algorithm, and simultaneously filling in point holes in the white blood cells; and finally, positioning the white blood cells. The present method is simple and effective, and is suitable for a wide range of applications compared to existing threshold-based algorithms, while rendering a final result more accurate by integrating the decision tree algorithm.