Bone Suppression in X-ray Radiograms via Gradient Field Smoothing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for removing bone shadows from radiograph images, such as dual energy acquisition and global affine transformations, are inefficient when dealing with curved bone contours, leading to suboptimal suppression of obstructing objects.
Innovation Solution
A system that computes and smooths the gradient field of an image using 2-dimensional convolution kernels, with one kernel parallel to the contour and another normal to it, allowing for effective extraction and suppression of objects regardless of contour orientation or curvature, and integrates these kernels to create an object image that can be subtracted from the original image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dual energy acquisition is used to remove bone shadows, then soft tissue image quality is improved, but equipment complexity and x-ray dose increase
Solution Approach 1:
The patent creates a bone model (a copy of the bone structure) by smoothing the gradient field and integrating it to generate a bone image that replicates the appearance and position of bones in the original radiograph. This bone image is then subtracted from the original image to produce a bone-suppressed image, eliminating the need for complex dual energy equipment while achieving similar soft tissue visualization效果
Solution Approach 2:
The patent replaces the physical/mechanical dual energy x-ray system with a computational image processing approach. Instead of using two different x-ray energy levels to differentially penetrate bone and soft tissue, the method uses gradient field smoothing and integration to synthesize and remove bone structures purely through mathematical operations on the single-energy radiograph
2Measurement precision
If global affine transformation is used to reorient bone shadows, then suppression performance improves for straight bones, but effectiveness deteriorates for curved bone contours
Solution Approach 1:
The patent employs curvature-adaptive smoothing by computing the gradient field and applying smoothing operations that respect the local curvature of bone structures. The gradient field smoothing process naturally follows curved bone contours rather than assuming straight-line orientations, allowing effective bone model generation for both straight and curved bone structures without requiring image reorientation
Solution Approach 2:
The patent uses a dynamic approach where the smoothing kernel and integration process adapt to the local geometry of bone structures. Rather than applying a fixed global transformation, the method dynamically adjusts to local contour characteristics through gradient field computation and smoothing, enabling effective bone suppression for bones with varying orientations and curvatures throughout the image
3Ease of manufacture
If bone model smoothing is applied in fixed directional kernels, then processing simplicity is maintained, but suppression accuracy deteriorates for non-horizontal bone orientations
Solution Approach 1:
The patent computes the gradient field of the radiograph, which provides directional information about bone structures regardless of their orientation. By smoothing this gradient field and integrating it, the method generates a bone model that automatically adapts to the actual orientation and curvature of bones in the image, achieving accurate suppression for bones in any orientation while maintaining the simplicity of a unified processing approach
Data Source
AI summary
The invention relates to a system (100) for extracting an object (Ob) from a source image, said object being delineated by a contour (C), the system (100) comprising a gradient unit (110) for computing the source image gradient field, based on the source image, a smoothing unit (120) for smoothing the source image gradient field, and an integration unit (130) for calculating an object image by integrating the smoothed source image gradient field, thereby extracting the object (Ob) from the source image. At each point of the source image, the smoothing is defined by a 2-dimensional convolution kernel which is a product of a first 1-dimensional convolution kernel in the first direction substantially parallel to the contour (C), and a second 1-dimensional convolution kernel in the second direction substantially normal to the contour (C). The first 1-dimensional convolution kernel defines smoothing within each region separated by the contour, while the second 1-dimensional convolution kernel defines smoothing across the contour separating two regions, independently of the orientation of the object and the contour curvature.