Image Normalization for Stained Blood Cell Visibility
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Microscopic images of stained blood cells often suffer from significant brightness differences between stained and unstained areas, making it difficult to distinguish cells from the background due to similar brightness levels, which hinders visual recognition.
Innovation Solution
An image normalization method and device that perform a cell visualization operation on each pixel using a specific equation (y = 1 / (1 + e^(c * (1 - x/t))) incorporating a contrast adjustment value, brightness adjustment value, and color distribution center to generate an output image, where the color distribution center is obtained using machine learning or deep-learning methods, ensuring output values are between 0 and 1.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If traditional image processing is applied to stained cell samples, then the brightness difference between stained and unstained areas is reduced, but the color difference between stained cells and stained background area cannot be distinguished
Solution Approach 1:
The patent transforms the image processing approach by changing from traditional brightness-based methods to a color space transformation method. By converting to HSV color space and operating on the Hue channel, the system preserves color information while enhancing contrast. The specific parameter changes involve adjusting hue values through histogram analysis and mapping, which maintains the distinctive color characteristics of stained cells against the background while improving visibility.
2Illumination intensity
If the brightness of the stained area is increased to make cells visible, then the contrast between cells and background is reduced
Solution Approach 1:
The patent segments the image processing into distinct color channel operations. By separating the HSV color space into Hue, Saturation, and Value channels and processing them independently, the system can enhance the Hue channel to improve cell-background contrast without affecting the overall brightness in the Value channel. This segmentation allows selective enhancement of color differences while maintaining the natural brightness distribution of the stained area.
3Illumination intensity
If conventional image normalization is used, then the overall brightness is improved, but cell information is lost due to excessive brightness filtering
Solution Approach 1:
The patent applies local quality enhancement by performing histogram analysis and color mapping specifically on the Hue channel rather than the entire image. This localized approach to color space transformation ensures that cell information is preserved through selective hue enhancement, while the Saturation and Value channels maintain their original characteristics. The method avoids excessive brightness filtering by operating locally in the color domain rather than globally in the intensity domain.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
An image normalization method and an image processing device (100) are provided. The image processing method includes: obtaining an input image; performing a cell visualization operation on each pixel of the input image, wherein the cell visualization operation generates an output value of each pixel of the input image according to an input value of each pixel of the input image, a color distribution center corresponding to at least one cell, a contrast adjustment value, and a brightness adjustment value.