Intraoperative Tumor Classification Using Images and Anatomical Position
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neurosurgical methods struggle to accurately determine tumor type during operations due to limitations in preoperative imaging, leading to increased costs and risks, and require additional hardware for real-time identification.
Innovation Solution
A method using a trained machine learning model that predicts tumor type based on captured images and categorical position information, eliminating the need for preoperative image evaluation and reducing resource requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If histopathological frozen section examination is performed to determine tumour type during surgery, then tumour type identification accuracy is improved, but operation time and anaesthesia risk increase
Solution Approach 1:
The patent replaces the mechanical/historical method of frozen section examination with a machine learning-based image recognition system. The trained machine learning model processes intraoperative images to predict tumour type in real-time, eliminating the need for waiting time associated with tissue processing and laboratory examination.
Solution Approach 2:
The patent uses digital copies of tissue images captured during surgery as input to the machine learning model. Instead of physically processing and examining actual tissue sections through frozen section, the system analyzes image representations of the tissue, providing instantaneous results without the time delay of physical processing.
2Loss of time
If preoperative imaging techniques (MRI, CT) are used to determine tumour type, then preparation time is reduced, but differentiation accuracy between tumour types remains insufficient
Solution Approach 1:
The patent performs preliminary training of the machine learning model using preoperative imaging data and known tumour types. This pretraining phase allows the system to learn patterns and features associated with different tumour types before the actual surgery, enabling accurate real-time classification during the operation.
Solution Approach 2:
The patent introduces a trained machine learning model as an intermediary between preoperative imaging and intraoperative decision-making. The model processes images captured during surgery, combining preoperative planning with real-time visual data to achieve accurate tumour type identification that neither preoperative imaging nor intraoperative frozen section can achieve alone.
3Measurement precision
If additional hardware systems (Raman spectroscopy, confocal endomicroscopy) are introduced for real-time tumour identification, then tumour type determination accuracy is improved, but investment costs and operating theatre space requirements increase
Solution Approach 1:
The patent makes the existing surgical microscope multi-functional by adding image processing capabilities through machine learning. Instead of requiring separate specialized hardware systems for tumour identification, the system uses the standard surgical microscope to capture images that are then processed by the trained machine learning model, achieving multiple functions (visual inspection and tumour classification) with a single device.
Solution Approach 2:
The patent replaces expensive specialized hardware systems (Raman spectroscopy, confocal endomicroscopy) with a computational approach using digital image copies. The trained machine learning model analyzes standard surgical microscope images to provide tumour type identification, eliminating the need for additional physical hardware while maintaining or improving accuracy.
Data Source
AI summary
A method for assisting a neurosurgical operation, wherein at least one image representation of an operating region on a patient is captured by means of a medical visualization system wherein categorical position information describing an anatomical position of a tumor present in the operating region is acquired and/or obtained, wherein the at least one captured image representation and the acquired and/or obtained categorical position information are supplied to a trained machine learning model as input data, wherein a tumor type is predicted by means of the trained machine learning model using the at least one captured image representation and the acquired and/or obtained categorical position information as a starting point, wherein the trained machine learning model provides tumor type information describing the tumor type as output data, and wherein the tumor type information is output. The invention further relates to an arrangement for assisting a neurosurgical operation.