Processing surgical data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The trade-off between imaging a greater range of wavelengths, image resolution, and camera size/weight in surgical robotics poses challenges, particularly in multispectral and hyperspectral imaging, which affects the versatility and safety of surgical robots.
Innovation Solution
A surgical system utilizing a processing device with a trained machine learning model to transform low-quality data from sensing devices into higher-quality data formats, such as higher resolution images or videos, by applying generative models like GANs and CNNs, enabling improved image reconstruction and telemetry data processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If multispectral or hyperspectral cameras image a greater number of wave bands, then more information about tissue properties is revealed, but the camera size and weight increase
Solution Approach 1:
The patent uses a standard RGB camera to capture visible light images and generates multispectral or hyperspectral images through computational processing and machine learning models. This copying approach creates virtual multispectral/hyperspectral data from standard camera inputs, eliminating the need for physically complex multispectral/hyperspectral sensors while still providing enhanced tissue information
Solution Approach 2:
The patent replaces the mechanical/optical system of complex multispectral or hyperspectral camera hardware with a computational system using machine learning models and algorithms. The physical complexity of multiple sensors and optical components is substituted with software-based image processing that reconstructs multispectral/hyperspectral information from standard RGB inputs
2Loss of information
If more sensors or bands per sensor are used to detect more wave bands, then imaging capability improves, but the camera becomes bulkier and more expensive
Solution Approach 1:
The patent creates virtual copies of multispectral/hyperspectral imaging capabilities through computational methods. Instead of physically implementing multiple sensors or spectral bands, the system uses machine learning models to generate synthetic spectral data from standard RGB camera inputs, achieving spectral information without spectral hardware complexity
Solution Approach 2:
The patent transforms the imaging approach by changing from physical spectral parameter detection to computational spectral parameter generation. The system uses trained machine learning models that take standard RGB image parameters and transform them into multispectral or hyperspectral image parameters through mathematical processing and pattern recognition
3Measurement precision
If higher resolution images are transmitted, then surgical precision improves, but data transfer bandwidth requirements increase
Solution Approach 1:
The patent extracts only the essential spectral information needed for surgical precision rather than transmitting complete high-resolution multispectral or hyperspectral datasets. The machine learning model processes and condenses spectral data to extract meaningful features that maintain surgical precision while reducing data volume for transmission
Solution Approach 2:
The patent applies different processing qualities to different regions or aspects of the imaging data. The machine learning model focuses computational resources on extracting and transmitting the most critical spectral information for surgical decision-making, rather than uniformly processing all spectral bands at full resolution
Data Source
AI summary
A surgical system comprising a processing device configured to implement a trained machine learning model, the processing device being configured to: receive first data having a first data format from a sensing device; receive additional data indicating a condition of the surgical system; in dependence on the additional data, input the first data or data derived therefrom to the trained machine learning model; and output second data having a second data format.