Stereoscopic Image Generation from Monoscopic Endoscope Views
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern 3D endoscopes are unable to be miniaturized to the same degree as 2D endoscopes due to their larger size and reduced field of view, making them unsuitable for certain surgical procedures, particularly those involving critical neurological structures, where depth perception is essential for precise operation.
Innovation Solution
A system and method for generating stereoscopic 3D images from monoscopic 2D endoscopic views using a machine learning algorithm, such as a convolutional neural network, to create a target image that, when combined with the original view, provides an artificial stereoscopic view for the surgeon, enhancing depth perception and precision during surgery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a traditional stereoscopic endoscope with two cameras is used, then depth perception is improved, but the device size increases and field of view decreases
Solution Approach 1:
The patent creates a synthetic copy of the second eye's view by using a machine learning model to generate a target image from the single input image. This copied view simulates the perspective of a second camera without physically installing one, thereby providing depth perception while maintaining the original device's compact size and field of view.
Solution Approach 2:
The patent replaces the mechanical system of a second physical camera with a computational system using machine learning algorithms. The neural network processes the single input image to generate a synthetic second view, substituting physical optical components with digital image processing to achieve stereoscopic depth perception.
2Measurement precision
If a traditional stereoscopic endoscope with two cameras is used, then depth perception is improved, but the device complexity and miniaturization capability are reduced
Solution Approach 1:
The patent replaces the mechanical complexity of housing and aligning two cameras with a computational approach using a trained machine learning model. The system uses a single camera combined with software-based view synthesis, significantly reducing device complexity while achieving the same depth perception effect.
Solution Approach 2:
Instead of physically duplicating camera hardware, the patent creates a virtual copy of the second eye's perspective through machine learning. This synthetic copying approach eliminates the need for complex dual-camera mechanical systems while providing authentic stereoscopic depth perception.
3Measurement precision
If a traditional stereoscopic endoscope with two cameras is used, then depth perception is improved, but the device size increases
Solution Approach 1:
The patent substitutes the physical volume required for two cameras with minimal computational processing. The machine learning model runs on existing hardware resources, replacing the need for additional camera modules and their associated mechanical mounting structures, thereby maintaining a compact device size.
Solution Approach 2:
The patent creates a virtual second view through computational copying rather than physical duplication of camera hardware. This approach generates a synthetic stereoscopic image from a single camera input, eliminating the volume increase that would result from installing a second physical camera.
Data Source
AI summary
A system for generating a target image comprises an endoscope having an image collection component, a computing device communicatively connected to the image collection component of the endoscope, comprising a non-transitory computer-readable medium with instructions stored thereon, which when executed by a processor perform steps comprising receiving at least one input image from the image collection component of the endoscope, providing the at least one input image as an input to a machine learning algorithm, generating a target image from the at least one input image using the machine learning algorithm, and providing the at least one input image and the target image to a display driver, and a display device, communicatively connected to the computing device, and configured to display the images provided to the display driver. A method of training a machine learning algorithm and a method of generating a stereoscopic image are also disclosed.