DBT Image Fusion for Lesion Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high volume of digital breast tomosynthesis (DBT) images increases the workload for doctors, who need to refer to 2D plain images for accurate diagnosis, leading to inefficient image reading and interpretation.
Innovation Solution
An image processing method that generates a target image sequence with fusion images by reconstructing and fusing tomographic images from multiple angles, using techniques like maximum intensity projection and weighted summation, to improve diagnostic efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple DBT tomographic images are reconstructed for accurate diagnosis, then diagnostic accuracy is improved, but the workload and time required for image reading increases
Solution Approach 1:
The patent segments the large volume of DBT tomographic images into multiple 2D planar images representing different anatomical planes (axial, coronal, sagittal). This segmentation allows radiologists to view specific planes of interest without reviewing all volumetric data, reducing reading time while maintaining diagnostic accuracy through targeted plane examination.
Solution Approach 2:
The patent transforms 3D volumetric DBT data into 2D planar representations by reconstructing images along different anatomical planes. This dimensionality change enables efficient 2D viewing and interpretation while preserving the diagnostic information contained in the original 3D data, allowing radiologists to leverage their existing 2D interpretation skills.
2Measurement precision
If multiple DBT tomographic images are reconstructed to understand tissue overlap, then lesion localization accuracy is improved, but the complexity of image interpretation increases
Solution Approach 1:
The patent segments the volumetric data into distinct 2D planar images along different anatomical planes. This segmentation eliminates tissue overlap by displaying only the relevant plane, making lesion localization straightforward without requiring interpretation of multiple overlapping 3D images.
Solution Approach 2:
By reconstructing images along axial, coronal, and sagittal planes, the patent provides multiple dimensional views of the breast tissue. This allows radiologists to localize lesions accurately by examining them from different anatomical perspectives without dealing with the complexity of interpreting tissue overlap in single 2D or volumetric 3D images.
3Measurement precision
If 2D plain images are taken for reference during diagnosis, then diagnostic accuracy is improved, but the overall imaging time and workflow efficiency worsen
Solution Approach 1:
The patent merges the benefits of 3D volumetric imaging with efficient 2D planar visualization by reconstructing multiple planes from the same DBT volumetric data. This eliminates the need to acquire separate 2D plain films, as the reconstructed 2D planes provide equivalent reference information without requiring additional imaging time or workflow steps.
Solution Approach 2:
The patent performs preliminary reconstruction of multiple 2D planar images from the volumetric DBT data during the initial imaging process. This preliminary action ensures that all necessary 2D reference images are already available when the radiologist begins interpretation, eliminating the need for additional 2D imaging and streamlining the diagnostic workflow.
Data Source
AI summary
The embodiments of the present disclosure provide methods, systems and computer storage mediums for processing an image. The method may include: obtaining a plurality of projection images generated at a plurality of angles; reconstructing, based on the plurality of projection images, a plurality of tomographic images of a plurality of slices; and obtaining a target image sequence based on the plurality of tomographic images of the plurality of slices, wherein the target image sequence including one or more fusion images, and the one or more fusion images are generated based on one or more tomographic images of the plurality of tomographic images corresponding to one or more slices of the plurality of slices.