Combined Projection Image Generation for Digital Breast Tomosynthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical imaging techniques face challenges in generating high-quality projection images from large data volumes, particularly in Digital Breast Tomosynthesis, where high resolution is needed for accurate diagnosis, but results in large data sets that complicate data transfer and increase radiologist workload, and require thick slabs for evaluating lesions that can extend over 10 mm in any direction.
Innovation Solution
A method involving capturing initial projection images, reconstructing three-dimensional volumes, generating and weighting re-projection images under specific geometries, and combining them to produce a combined projection image that enhances sharpness and contrast while reducing noise, using techniques like filtered back-projection and Maximum or Average Intensity Projections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution volumes are used to visualize fine clinical details, then diagnostic accuracy is improved, but data volume increases significantly
Solution Approach 1:
The patent segments the large DBT data volume into multiple three-dimensional volumes, each reconstructed from a different subset of initial projection images. This allows the system to work with smaller, more manageable data sets while still providing comprehensive diagnostic information through multiple perspectives.
Solution Approach 2:
The patent transforms the problem from working with a single large high-resolution volume to generating multiple lower-resolution volumes that are then re-projected and combined. This dimensional transformation in the data processing space allows achieving high diagnostic quality without requiring one extremely large data set.
2Adaptability or versatility
If thick slabs are generated to evaluate lesions extending over 10 mm, then lesion evaluation capability is improved, but image sharpness decreases
Solution Approach 1:
The patent divides the thick slab evaluation task into multiple smaller three-dimensional volumes, each reconstructed from different projection subsets. By generating multiple re-projection images from these segmented volumes and combining them with appropriate weighting, the system achieves both thick-slab coverage and maintained image sharpness.
Solution Approach 2:
The patent applies different weighting factors to different regions and structures within the re-projection images, allowing diagnostically relevant information to be highlighted while maintaining sharpness. This local quality adjustment enables selective enhancement of important features throughout the thick slab volume.
3Manufacturing precision
If multiple re-projection images are combined with weighting, then image quality and contrast are improved, but computational complexity increases
Solution Approach 1:
The patent reconstructs multiple three-dimensional volumes from different subsets of initial projection images rather than using all projections for a single reconstruction. This partial action approach distributes the computational workload and allows parallel processing, reducing overall computational complexity while still achieving high image quality through the combination of multiple re-projection images.
4Loss of information
If full data sets are used for reconstruction, then image completeness is improved, but artifact accumulation increases
Solution Approach 1:
The patent segments the complete set of initial projection images into multiple subsets, with each subset used to reconstruct a separate three-dimensional volume. This segmentation prevents artifact accumulation that would occur if all projections were used in a single reconstruction, while still providing comprehensive image coverage through the combination of multiple re-projections from different subsets.
Data Source
AI summary
A method for generating a combined projection image from a medical inspection object, includes steps of capturing a set of initial projection images; reconstructing a first and a second three-dimensional volume from the set of initial projection images; generating a first re-projection image from the first three-dimensional volume and a second re-projection image from the second three-dimensional volume; weighting the first re-projection image and the second re-projection image; and combining the weighted first re-projection image and second re-projection image for generating the combined projection image.