Urine Particle Image Region Segmentation Using Density Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional urine sediment examination methods face challenges in accurately segmenting particles with varying stain levels and tones, leading to inaccurate region extraction and classification due to light refraction and reflection effects, as well as overlapping density histograms.
Innovation Solution
A method involving the extraction of first and second object regions using RGB images, with density distribution calculations and threshold processing based on predetermined groups, allowing for stable segmentation of particles with different sizes and tones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single region segmentation method is used for all particles, then the device complexity is reduced, but the measurement precision deteriorates due to varying particle properties
Solution Approach 1:
The patent divides particles into multiple groups based on density distribution characteristics, and applies different region segmentation methods to each group. This segmentation approach allows the system to handle diverse particle types (sufficiently-stained, hard-to-stain, and hardly-stained particles) with appropriate methods, improving measurement precision without significantly increasing overall system complexity.
Solution Approach 2:
The patent applies different region segmentation approaches to different particle groups based on their local characteristics. For sufficiently-stained particles, one method is used, while for hard-to-stain and hardly-stained particles, alternative methods are applied. This local quality principle ensures optimal segmentation accuracy for each particle type while maintaining manageable device complexity.
2Ease of operation
If a fixed density threshold is used for region segmentation, then the ease of operation is improved, but the measurement precision deteriorates due to overlapping density histograms of different particle types
Solution Approach 1:
The patent dynamically adjusts the density threshold based on the particle group being analyzed. Instead of using a fixed threshold for all particles, the system selects appropriate threshold values from different density histograms corresponding to different particle types. This dynamic adaptation improves measurement precision while maintaining ease of operation through automated threshold selection.
Solution Approach 2:
The patent changes the density parameter threshold according to the particle group characteristics. By storing multiple density histograms and selecting the appropriate one based on particle properties, the system optimizes the density threshold parameter for each particle type, thereby improving region extraction accuracy without complicating the operational process.
3Productivity
If region segmentation is performed without considering particle density distribution, then the productivity is improved, but the measurement precision deteriorates due to background region misclassification
Solution Approach 1:
The patent performs preliminary analysis of particle density distribution by storing density histograms for different particle types before region segmentation. This preliminary action allows the system to quickly identify the appropriate density characteristics for each particle group, enabling accurate region segmentation without significantly impacting examination speed. The pre-stored histograms facilitate rapid threshold determination.
Solution Approach 2:
The patent enables the system to automatically determine appropriate density thresholds and segmentation parameters by utilizing the stored density histograms and particle group characteristics. This self-service mechanism eliminates the need for manual parameter adjustment for each particle type, maintaining high productivity while improving measurement precision through automated, accurate region segmentation.
Data Source
AI summary
An objective is to provide a method and an apparatus for accurately extracting a region of an object particle from a urine particle image obtained by taking an image of urine particles in a urine specimen having varying properties. First, a first object region is extracted using one or more of an R image, a G image, and a B image of a urine particle image taken by an image input optical system configured to input particle images. Then, a density distribution and a size of the first object region of one or more of the R image, the G image, and the B image are calculated. Based on these feature parameters, the first object region is classified into a predetermined number of groups. A second object region is extracted from a local region including the first object region, by using one or more of the R image, the G image, and the B image, depending on each of the groups. This configuration allows stable region segmentation for each particle image even for a urine specimen in which urine particles having different sizes and tones coexist.