Deep Learning Bone Suppression for Chest Radiographs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for bone suppression in chest radiographs are inefficient, as they often require multiple exposures or image manipulations, and do not effectively suppress rib content, which can obscure important medical information.
Innovation Solution
A deep learning convolutional neural network module is trained using pairs of chest x-ray images and their bone-suppressed counterparts to generate enhanced bone-suppressed images directly from original radiographic images, eliminating the need for multiple exposures and traditional image manipulation techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional bone suppression algorithms are applied to chest radiographs, then bone structures are partially suppressed, but rib content remains and image clarity is insufficient
Solution Approach 1:
The patent replaces traditional mechanical/image manipulation-based bone suppression algorithms with a deep learning neural network system. The neural network is trained on pairs of original and manually bone-suppressed images to learn the complex pattern recognition and suppression task, achieving superior rib suppression while preserving soft tissue details that traditional methods fail to capture.
Solution Approach 2:
The patent transforms the bone suppression task from direct algorithmic manipulation of radiographic parameters to a learned mapping between input and target images. By changing from rule-based parameter adjustment to data-driven parameter optimization through training, the system achieves more accurate and complete bone suppression.
2Reliability
If multiple exposures or traditional image manipulation techniques are used for bone suppression, then some bone suppression is achieved, but the process becomes complex and time-consuming
Solution Approach 1:
The patent replaces complex multi-step mechanical image manipulation procedures with a single neural network inference step. The trained network processes the original radiograph through learned feature transformations to produce the bone-suppressed image directly, eliminating the need for multiple exposures and sequential processing steps.
Solution Approach 2:
The patent performs the complex bone suppression task in advance during the training phase, where the neural network learns from numerous examples of original and target images. This preliminary learning action enables the system to execute accurate bone suppression during clinical use without requiring complex real-time processing or multiple exposures.
3Ease of manufacture
If conventional bone suppression methods are applied, then processing can be performed with existing tools, but diagnostic accuracy is limited due to insufficient bone suppression
Solution Approach 1:
The patent replaces conventional image processing tools with a deep learning-based neural network system. This substitution maintains implementation feasibility through standard computational infrastructure while dramatically improving diagnostic accuracy by achieving complete rib suppression and preserving soft tissue contrast that conventional methods cannot achieve.
Data Source
AI summary
A system and method for generating a rib suppressed radiographic image using deep learning computation. The method includes using a convolutional neural network module trained with pairs of a chest x-ray image and its counterpart bone suppressed image. The bone suppressed image is obtained using a bone suppression algorithm applied to the chest x-ray image. The convolutional neural network module is then applied to a chest x-ray image or the bone suppressed image to generate an enhanced bone suppressed image.