Composite Mammography Image Selection for Tumor Differentiation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods for breast cancer diagnosis using mammography fail to achieve the same diagnostic performance in composite two-dimensional images as tomographic images, leading to increased interpretation burden for radiologists due to the difficulty in distinguishing between tumors and local mammary gland masses.
Innovation Solution
An image processing device and method that generates a composite two-dimensional image by selecting and combining specific tomographic images based on tumor and mammary gland characteristics, using algorithms for tumor detection and center of gravity analysis to differentiate between tumors and local masses, thereby reducing the need for interpreting multiple images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple tomographic images are used for diagnosis, then diagnostic performance is improved, but interpretation burden increases
Solution Approach 1:
The patent combines multiple tomographic images into a single composite two-dimensional image by synthesizing information from different depth planes. This merging process preserves diagnostic information while reducing the number of images radiologists need to interpret, directly resolving the contradiction between diagnostic performance and interpretation burden
Solution Approach 2:
The patent creates a composite image that integrates three-dimensional tomographic data into a two-dimensional representation. By transforming the dimensional representation while preserving diagnostic content, it allows radiologists to view comprehensive information in a single plane without losing the depth resolution benefits of tomosynthesis
2Speed
If simple back projection method is used for reconstruction, then processing speed is improved, but image quality deteriorates
Solution Approach 1:
The patent extracts and combines specific high-quality features from multiple tomographic images reconstructed by simple back projection. By selecting and synthesizing only the most diagnostic information from each image, it achieves high image quality without requiring complex reconstruction algorithms, thus maintaining processing speed
3Measurement precision
If tumor and local mass cannot be distinguished, then diagnostic accuracy is improved, but false positive rate increases
Solution Approach 1:
The patent applies different processing and weighting to different regions within the composite image based on their diagnostic characteristics. By enhancing local features specific to tumors while suppressing characteristics of benign local masses, it improves diagnostic accuracy while reducing false positives through region-specific optimization
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
An image processing device determines whether each tumor candidate regions detected from a plurality of tomographic images indicating a plurality of tomographic planes of an object is a tumor or a local mass of a mammary gland, selects a first tomographic image group from the plurality of tomographic images in a first region determined to be the tumor, selects a second tomographic image group from the plurality of tomographic images in a second region determined to be the local mass of the mammary gland, selects a third tomographic image group from the plurality of tomographic images in a third region other than the first region and the second region, and generates a composite two-dimensional image using the tomographic image groups selected for each of the first region, the second region, and the third region.