Tomographic Image Composite Generation for Mammography
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating composite two-dimensional images from tomographic images in mammography fail to effectively highlight structures of interest in depth and can blur fine structures, making it difficult to detect and visualize important features like tumors, spiculae, and calcifications.
Innovation Solution
An image processing device that detects structures of interest in tomographic images, performs frequency decomposition to derive band tomographic images, and selects the most representative images for each pixel based on the frequency band and structure type, generating a composite two-dimensional image that enhances the visibility of structures in depth and maintains fine structural details.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If multiple tomographic images are combined using addition or averaging methods to generate a composite two-dimensional image, then the depth information of structures is improved, but fine structures such as calcifications and spiculae become blurred and difficult to detect
Solution Approach 1:
The patent segments the composite image generation process by frequency bands. It divides the image into multiple frequency components and processes each band separately, selecting images based on the presence of structures of interest in specific frequency ranges. This segmentation allows different frequency components to contribute to different aspects of the composite image, preserving fine structures while maintaining depth information.
Solution Approach 2:
The patent applies local quality by selecting different tomographic images for different spatial regions based on the presence and characteristics of structures of interest. In regions containing fine structures like calcifications, it prioritizes images that preserve these details, while in other regions it may use images that provide better depth information. This localized selection strategy optimizes both depth representation and fine structure visibility in different parts of the image.
2Reliability
If tomographic images are combined using maximum intensity projection or minimum intensity projection methods, then certain structures become more prominent, but other important structures may be lost or obscured
Solution Approach 1:
The patent changes the selection parameter from simple intensity-based projection (maximum or minimum) to a frequency-band-based selection criterion. Instead of projecting based on extreme intensity values, it selects images based on the presence of structures of interest in specific frequency bands, using detection results to guide the selection process. This parameter change allows multiple structures with different intensity characteristics to be preserved simultaneously.
Solution Approach 2:
The patent introduces feedback by using detection results of structures of interest to guide the image selection process. The detection system identifies structures in each tomographic image, and this information feeds back into the selection process, determining which images should be used for composite image generation. This feedback mechanism ensures that structures of interest are preserved while reducing information loss.
3Measurement precision
If a single tomographic image is used to generate the composite two-dimensional image, then fine structures are preserved, but depth information and three-dimensional context are lost
Solution Approach 1:
The patent adds another dimension to the image selection process by incorporating frequency band information. Instead of selecting images based solely on spatial position or intensity, it introduces frequency domain characteristics as an additional selection criterion. This allows the composite image to integrate information from multiple tomographic images across different frequency bands, simultaneously preserving fine structures and depth information.
Data Source
AI summary
A processor detects a structure of interest from a plurality of tomographic images indicating a plurality of tomographic planes of an object. The processor selects a tomographic image from the plurality of tomographic images according to a frequency band in a region in which the structure of interest has been detected. The processor generates a composite two-dimensional image using the selected tomographic image in the region in which the structure of interest has been detected and using a predetermined tomographic image in a region in which the structure of interest has not been detected.