3D Breast Model Analysis for False Positive Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for analyzing breast images are time-consuming and prone to errors when detecting aberrant regions, especially when dealing with large numbers of images, as they rely on human examination and are not efficient in distinguishing between benign and malignant tissues.
Innovation Solution
An analysis method and electronic apparatus that utilize a machine learning algorithm to create a 3D breast model, identify volumes of interest, and compare them with tissue segmentation results to determine false positives, allowing for rapid and accurate detection of aberrant regions and their properties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If medical personnel manually examine breast images to detect aberrant parts, then detection accuracy can be maintained through human verification, but the analysis time increases significantly when dealing with large numbers of images
Solution Approach 1:
The patent segments the breast image analysis process into multiple independent components: initial aberrant part detection by the computer system, tissue segmentation into glandular and non-glandular regions, and hierarchical verification steps. This segmentation allows parallel processing of different image regions and enables the system to quickly eliminate false positives by checking only relevant tissue types, thereby reducing overall analysis time while maintaining detection accuracy.
Solution Approach 2:
The patent implements preliminary automated detection of aberrant parts using machine learning algorithms before human verification. The system pre-processes images by detecting potential abnormalities and classifying them, so that medical personnel only need to verify suspicious cases rather than examining every image from scratch. This preliminary action significantly reduces the time required for manual review while maintaining high detection accuracy.
2Productivity
If automated detection methods are used to quickly process large numbers of breast images, then productivity increases, but the reliability of detection decreases due to higher error rates
Solution Approach 1:
The patent introduces tissue segmentation results as an intermediary layer between automated aberrant part detection and final diagnosis. The system first segments breast tissue into glandular and non-glandular regions, then uses this segmentation information to verify detected aberrant parts. This intermediary step acts as a filter that eliminates false positives (e.g., abnormalities detected in non-glandular tissue that cannot be malignant) while preserving true positives, thereby maintaining high detection reliability even when processing large numbers of images rapidly.
Solution Approach 2:
The patent implements a feedback mechanism where tissue segmentation results are used to verify and refine the initial automated detection of aberrant parts. The system continuously refines its detection by comparing detected abnormalities against tissue type classifications, learning from previous results to improve subsequent detections. This feedback loop maintains high reliability while enabling rapid processing of large image datasets.
3Reliability
If comprehensive verification of all detected aberrant parts is performed, then false positives are reduced, but the complexity of the analysis process increases
Solution Approach 1:
The patent applies local quality verification by tailoring the verification process to the specific tissue type where each aberrant part is detected. Instead of applying uniform comprehensive verification to all detected abnormalities, the system checks only those located in glandular tissue with further lactiferous duct analysis, while accepting results from non-glandular tissue with appropriate confidence levels. This localized approach reduces false positives without requiring complex verification of every single detection, thereby managing analysis process complexity effectively.
4Measurement precision
If detailed tissue segmentation and verification steps are implemented, then measurement precision improves, but the device complexity increases
Solution Approach 1:
The patent segments the verification process into distinct modular steps: tissue segmentation into glandular/non-glandular regions, aberrant part detection, and conditional verification based on tissue type. Each module performs a specific function and can be independently implemented or modified. This segmentation achieves high measurement precision through systematic verification while managing device complexity by organizing the system into manageable, functionally-separated components rather than a monolithic complex structure.
Data Source
AI summary
An analysis method and an electronic apparatus for breast image are provided. The method includes the following steps. One or more breast ultrasound images are obtained. The breast ultrasound images are used for forming a three-dimensional (3D) breast model. A volume of interest (VOI) in the breast ultrasound image is obtained by applying a detection model on the 3D breast model. The VOI is compared with a tissue segmentation result. The VOI is determined as a false positive according to a compared result between the VOI and the tissue segmentation result. The compared result includes that the VOI is located at a glandular tissue based on the tissue segmentation result. In response to the VOI being located in the glandular tissue of the tissue segmentation result, the VOI is compared with the lactiferous duct in the 3D breast model.