3D Bone Imaging from 2D X-rays via ML Posture Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D imaging technologies, such as CT scans, expose patients to high radiation and are costly and bulky, and existing methods for generating stereoscopic images from two-dimensional images fail to produce satisfactory results for human bones and joints, especially when input images are not orthogonally positioned.
Innovation Solution
A method using an image processing engine to predict posture parameters from two X-ray images and generate stereoscopic images, employing machine learning algorithms like convolutional neural networks (CNNs) and interference image removal techniques to optimize the reconstruction of 3D images from non-orthogonal input images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If CT scan is used for 3D imaging, then high-resolution internal structure information is obtained, but high radiation dose and high cost are incurred
Solution Approach 1:
The patent creates a 3D copy of the bone structure by reconstructing it from two 2D X-ray images using machine learning algorithms. Instead of using CT scan to directly obtain 3D data, the system generates a virtual 3D model that replicates the bone's anatomical features, thereby avoiding the harmful radiation exposure associated with CT scanning while maintaining diagnostic quality
Solution Approach 2:
The patent replaces the mechanical CT scanning system with an computational image processing system. Rather than physically rotating the X-ray source and detector around the patient as in CT, the system uses machine learning algorithms to computationally reconstruct 3D bone structure from two stationary 2D X-ray images, eliminating the need for multiple X-ray exposures
2Measurement precision
If CT scanner is used for 3D imaging, then accurate bone structure data is obtained, but the device is expensive and bulky
Solution Approach 1:
The system creates an accurate 3D digital copy of the bone structure through computational reconstruction from 2D images. This virtual model captures essential anatomical features needed for diagnosis and surgical planning without requiring expensive CT scanning hardware
Solution Approach 2:
The patent uses standard, widely available X-ray imaging equipment instead of expensive CT scanners. The system processes ordinary 2D X-ray images through machine learning algorithms to generate 3D reconstructions, making the technology more accessible and cost-effective while maintaining diagnostic accuracy
3Ease of operation
If existing methods generate stereoscopic images from two-dimensional images, then 3D visualization is achieved, but satisfactory results are not obtained when input images are not orthogonally positioned
Solution Approach 1:
The patent performs preliminary detection of the actual angular positions of the two X-ray images before reconstruction. By measuring the precise orientation of each image relative to the bone structure, the system can pre-calculate the appropriate transformation parameters needed for accurate 3D reconstruction, even when images are not perfectly orthogonal
Solution Approach 2:
The system dynamically adjusts reconstruction parameters based on the detected angular positions of the input images. Instead of assuming fixed orthogonal positioning, the machine learning algorithm modifies the reconstruction geometry to accommodate various angular configurations, thereby maintaining accuracy across different imaging scenarios
Data Source
AI summary
The present invention relates to a method for generating three-dimensional image from two-dimensional images, and more specifically, a method for generating three-dimensional image of human bones from two 2D planar images thereof. The method comprises the steps of: providing a first X-ray planar image and a second X-ray planar image; predicting one set of predicted posture parameters for each of the first X-ray planar image and the second X-ray planar image; and generating the data of a stereoscopic image according to the first X-ray planar image, the second X-ray planar image, and the predicted posture parameters. The present invention also relates to a method for training an artificial intelligence to perform three-dimensional image generation described above.