Lesion-Specific Reconstruction in Digital Breast Tomosynthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital breast tomosynthesis reconstruction methods require significant computational resources and storage space, as they reconstruct the entire volume uniformly, which can lead to overlooked diagnostically relevant information due to resolution limitations, and do not allow for real-time adjustment of reconstruction parameters based on user selection.
Innovation Solution
A computer-implemented method that includes a pre-computation phase to detect regions of interest and determine lesion-type specific reconstruction parameters, allowing for pre-computed or online computation of selected areas with different reconstruction algorithms and parameters, enabling user-selectable visualization options for improved diagnostic efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire digital breast tomosynthesis volume is reconstructed with uniform high resolution, then diagnostic accuracy is improved, but data processing time and storage requirements increase significantly
Solution Approach 1:
The patent applies different reconstruction resolutions to different regions of the breast volume based on diagnostic importance. Regions containing lesions or abnormalities are reconstructed with high resolution to maintain diagnostic accuracy, while normal regions are reconstructed with lower resolution to reduce processing time and storage requirements. This selective approach resolves the contradiction by maintaining high diagnostic accuracy where needed while reducing overall processing burden.
Solution Approach 2:
The patent segments the digital breast tomosynthesis volume into multiple regions based on diagnostic relevance. The volume is divided into lesion-containing regions and normal regions, with each segment processed differently during reconstruction. This segmentation allows the system to focus computational resources on diagnostically critical areas while using more efficient processing for non-critical areas, thereby reducing overall processing time without compromising diagnostic accuracy in important regions.
2Measurement precision
If the entire digital breast tomosynthesis volume is reconstructed with uniform high resolution, then diagnostic accuracy is improved, but storage space requirements increase significantly
Solution Approach 1:
The patent implements local quality by storing high-resolution reconstruction data only in regions containing lesions or abnormalities, while storing lower-resolution data in normal regions. This approach maintains diagnostic accuracy for lesion detection and characterization while significantly reducing the total storage space required for the complete breast volume, directly resolving the contradiction between diagnostic accuracy and storage requirements.
3Productivity
If thick slices are used for volume reconstruction, then data processing and storage are reduced, but diagnostically relevant information may be overlooked
Solution Approach 1:
The patent applies local quality by using different slice thicknesses for different regions. In regions containing lesions or abnormalities, thin slices are used to preserve fine diagnostic details and prevent information loss. In normal regions, thick slices are used to improve processing efficiency and reduce data volume. This region-specific approach resolves the contradiction between processing efficiency and information preservation.
4Device complexity
If a single reconstruction algorithm is used for the entire volume, then processing is simplified, but lesion-specific optimization is lost
Solution Approach 1:
The patent implements local quality by selecting different reconstruction algorithms for different regions based on lesion type and characteristics. For example, iterative reconstruction algorithms may be applied to regions containing subtle abnormalities, while faster algorithms are used for normal regions. This approach maintains high lesion visualization quality where needed while keeping overall process complexity manageable through automated algorithm selection.
Solution Approach 2:
The patent applies dynamics by making the reconstruction algorithm selection adaptive rather than static. The system dynamically chooses appropriate reconstruction algorithms based on the detected lesion type, size, and characteristics in each region. This dynamic adaptation allows the system to optimize for both speed and quality automatically, resolving the contradiction between simplicity and precision.
Data Source
AI summary
A method, a control unit and a system for image reconstruction and visualization of a tomosynthesis volume. Different region of interests are detected in the volume and specific types of lesions are determined. Based on the type of lesion, different reconstruction parameters and different reconstruction algorithms are applied in order to reconstruct a sub-volume or a region in a projection. After displaying the digital breast tomosynthesis volume, a user selection signal is received, in order to identify a selection area. The selection area identifies a region in the volume which should be reconstructed differently from the remaining volume and typically with higher resolution, because it refers to a region of specific interest. After having received the user selection signal the selection area is defined and a pre-computed or online-computed reconstruction of the selection area is visualized. The reconstruction is executed according to the determined lesion type specific reconstruction parameters.