Breast Tomosynthesis Auto-Cropping Algorithm
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current tomosynthesis breast imaging techniques suffer from low resolution along the thickness of the breast due to a narrow range of projection angles, leading to unnecessary layers outside the breast that complicate diagnosis, increase computational costs, and degrade image quality.
Innovation Solution
An X-ray apparatus that uses a boundary calculator, integrated with a position sensor and force sensor, to determine the correct layers to reconstruct within the breast, eliminating unnecessary layers and improving image quality by correlating compression paddle position and force measurements with reconstruction algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If a narrow range of projection angles is used in tomosynthesis, then radiation dose is reduced and image quality is improved, but resolution along the thickness of the breast deteriorates causing thick layers and spill-over between layers
Solution Approach 1:
The patent extracts and removes layers outside the breast from the reconstructed image volume using an auto-cropping algorithm. This eliminates the harmful spill-over effects and low-frequency artifacts from layers containing no breast tissue, thereby improving the effective resolution and diagnostic quality without requiring a wider projection angle range.
2Reliability
If layers outside the breast are included in reconstruction, then complete volumetric coverage is achieved, but diagnostic speed decreases and work efficiency is reduced
Solution Approach 1:
The patent automatically identifies and removes layers outside the breast using intensity and gradient-based algorithms, retaining only the relevant layers containing breast tissue. This extraction process maintains complete coverage of the breast volume while eliminating unnecessary layers that slow down diagnosis, thereby improving both reliability and productivity.
Solution Approach 2:
The auto-cropping algorithm performs preliminary removal of unnecessary layers before the radiologist views the images. By pre-processing the image volume to eliminate layers outside the breast, the system prepares optimized datasets that speed up subsequent diagnostic work without compromising volumetric coverage.
3Productivity
If auto-cropping algorithms remove layers based on maximum intensity or difference, then computational speed improves, but accuracy deteriorates when objects have low contour content near boundaries
Solution Approach 1:
The patent employs a two-stage auto-cropping approach that first removes layers with very low intensity (clearly outside the breast) and then applies gradient-based refinement to accurately identify boundary layers. This partial removal strategy maintains computational efficiency while improving boundary accuracy by focusing detailed analysis only on the critical transition region.
Solution Approach 2:
The algorithm uses gradient magnitude feedback to iteratively refine the identification of breast boundary layers. By measuring the gradient at each layer boundary and adjusting the crop limits based on this feedback, the system accurately identifies layers containing breast tissue even when contour content is low, maintaining both speed and precision.
4Reliability
If all layers are reconstructed, then complete image volume is produced, but reconstruction time increases and computational cost rises
Solution Approach 1:
The patent extracts and excludes layers outside the breast from the reconstruction process entirely. By identifying the breast boundaries using position and force sensor data combined with auto-cropping algorithms, the system reconstructs only the necessary layers containing breast tissue, maintaining complete image volume coverage while reducing reconstruction time and computational cost.
Solution Approach 2:
The system performs preliminary identification of breast boundaries using compression paddle position and force measurements before initiating the full reconstruction process. This preliminary action defines the reconstruction volume in advance, ensuring completeness of breast tissue coverage while avoiding unnecessary computation for layers outside the breast.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach significantly reduces reconstruction time, enhances image quality by focusing only on layers containing human tissue, and eliminates the need for markers or sharp contours, thereby speeding up calculations and improving diagnostic efficiency.
Implementation Method 1
Each projection image is essentially a conventional 2-dimensional digital X-ray image of the examined object
Implementation Method 2
each voxel is essentially a value corresponding to the X-ray attenuation in one point of the real world
Implementation Method 3
the breast is compressed between a patient support and a compression paddle, which reduces radiation dose and enables better image quality
Implementation Method 4
a position sensor indicates the thickness of the breast, which is used for determining exposure parameters for the X-ray source
Implementation Method 5
the force of pressure is also measured, which is displayed to the operator or used to control the motor. In addition, the measured force is also used to correct for paddle deflection in the measured thickness
Data Source
AI summary
To enhance results of an x-ray apparatus the invention relates to an apparatus for three-dimensional imaging of a human breast comprising a reconstruction means for reconstruction of a three-dimensional image volume, a compression paddle for holding said breast, an actuator for controlling the position of said compression paddle, and a position gauge for measurement of the position of said compression paddle. The reconstruction means is operatively arranged to receive signals, corresponding to measurements by said position gauge, from said position gauge and constrain the boundary of said image volume based on said received signal.