Hierarchical Feature Image Synthesis for Breast Tissue Tumor Visibility
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing breast imaging technologies, such as mammography and tomosynthesis, struggle to accurately identify and enhance clinically relevant three-dimensional objects like malignant breast masses and tumors in two-dimensional synthesized images, often leading to incomplete or inaccurate representations due to tissue overlap and structure noise.
Innovation Solution
A method involving multi-level feature modules that decompose high-dimensional breast tissue data into lower-dimensional patterns, using filters and a learning library-based combiner to generate feature maps and object maps, which are then combined to create a 2D synthesized image that accurately represents high-dimensional objects, enhancing their visibility and morphology.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional 2D mammography or tomosynthesis is used to image breast tissue, then the imaging process is simple and fast, but tissue overlap and structure noise prevent accurate identification of three-dimensional objects like tumors
Solution Approach 1:
The patent segments the breast tissue volume into multiple thin 2D slices along the longitudinal axis. Each slice is processed independently through multi-level feature modules that extract specific anatomical features. This segmentation allows the system to analyze three-dimensional structures by breaking them down into manageable two-dimensional components, reducing the harmful effect of tissue overlap while maintaining the ability to reconstruct 3D objects accurately.
Solution Approach 2:
The patent transforms 3D breast tissue data into 2D synthesized images by projecting high-dimensional volumetric data onto 2D planes. The multi-level feature modules process 2D slices and generate feature maps that are then combined to create enhanced 2D representations. This dimensionality change allows radiologists to view complex 3D structures in a 2D format that eliminates tissue overlap while preserving diagnostic information about tumor morphology and position.
2Measurement precision
If multi-level feature modules and high-dimensional grid decomposition are used to accurately represent 3D objects, then diagnostic accuracy improves, but system complexity and processing time increase
Solution Approach 1:
The patent divides the complex task of 3D object recognition into multiple simpler 2D processing stages. Each multi-level feature module handles specific feature extraction on individual 2D slices, rather than attempting to process the entire 3D volume at once. This segmentation reduces computational complexity while maintaining accuracy, as each module can be optimized for its specific function and processed in parallel.
Solution Approach 2:
The patent applies multiple levels of feature extraction and processing beyond what a single 2D image analysis would provide. The system generates multiple feature maps at different levels of abstraction and combines them to create a comprehensive representation. This partial or excessive action ensures that all relevant diagnostic features are captured, even though it increases processing requirements, but the modular architecture allows for selective application based on diagnostic needs.
3Loss of information
If multiple 2D slices are processed to create synthesized images, then complete representation of breast tissue is achieved, but processing time and computational resources increase
Solution Approach 1:
The patent processes breast tissue as a series of discrete 2D slices rather than as a continuous 3D volume. Each slice is independently analyzed by the multi-level feature modules, allowing for parallel processing and efficient resource utilization. This segmentation enables complete tissue representation while reducing processing time compared to traditional 3D reconstruction methods, as each 2D slice can be processed quickly and then combined to form the complete picture.
Solution Approach 2:
The patent performs feature extraction and image synthesis in advance, creating processed 2D slices and feature maps before final diagnostic review. The system pre-processes the volumetric data into multiple 2D representations with extracted features already organized and ready for analysis. This preliminary action reduces the time required during actual diagnostic evaluation, as the complex processing has already been completed and the results are available for immediate review.
Data Source
AI summary
A method for processing breast tissue image data includes processing the image data to generate a set of image slices collectively depicting the patient's breast; for each image slice, applying one or more filters associated with a plurality of multi-level feature modules, each configured to represent and recognize an assigned characteristic or feature of a high-dimensional object; generating at each multi-level feature module a feature map depicting regions of the image slice having the assigned feature; combining the feature maps generated from the plurality of multi-level feature modules into a combined image object map indicating a probability that the high-dimensional object is present at a particular location of the image slice; and creating a 2D synthesized image identifying one or more high-dimensional objects based at least in part on object maps generated for a plurality of image slices.