Surgical Tool Recognizer Update via CG Data Evaluation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing surgical tool recognizers struggle to accurately recognize new surgical tools due to environmental influences, lacking a method to confirm recognition capability after updating.

Innovation Solution

An information processing device and method that generates CG data from images of new surgical tools, learns a recognizer using this data, and evaluates recognition results to ensure sufficient recognition capability before updating the recognizer.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a new learning model is generated to recognize a new surgical tool, then the recognizer can be updated to include the new tool, but there is no method to confirm whether the new tool is actually recognized after updating

Engineering Contradiction:

Improverecognizer's ability to recognize new surgical toolsVSAvoidconfirmation of recognition capability

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent implements a feedback mechanism where the updated recognizer processes images of the new surgical tool, and the recognition results are evaluated against ground truth data. This closed-loop feedback system confirms whether the recognizer successfully learned the new tool, resolving the contradiction between adaptability and reliability by providing empirical verification of recognition capability.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent performs preliminary evaluation actions before the recognizer is deployed for actual use. By testing the recognizer on validation images of the new surgical tool immediately after updating, the system confirms recognition capability in advance, ensuring reliability before the recognizer is put into production use.

Inventive Principle:

Principle #10Preliminary action

2Adaptability or versatility

If the surgical tool recognizer is updated with a new learning model, then it can potentially recognize the new surgical tool, but the recognition may fail due to environmental influences

Engineering Contradiction:

Improverecognizer's capability to handle new surgical toolsVSAvoidenvironmental influences on recognition accuracy

Core Design Contradiction:

Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

The patent employs extensive data augmentation techniques that generate variations of surgical tool images under different environmental conditions (lighting, angles, backgrounds). By training with excessive diverse samples beyond what would normally be encountered, the recognizer becomes robust to environmental influences while maintaining adaptability to new tools.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent prepares the recognizer in advance against environmental variations by training with augmented images that simulate different surgical environments. This beforehand cushioning ensures that when the recognizer encounters actual environmental variations during use, it remains reliable while maintaining adaptability to new surgical tools.

Inventive Principle:

Principle #11Beforehand cushioning (Prior cushioning)

3Measurement precision

If learning is performed continuously to improve recognition accuracy, then the recognizer becomes more accurate, but the time and computational resources required increase

Engineering Contradiction:

Improverecognition accuracyVSAvoidlearning time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary data preparation and augmentation before the actual learning process. By pre-processing images, generating augmentations, and preparing the training dataset in advance, the subsequent learning phase is optimized and completed faster, achieving high accuracy without excessive learning time.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements dynamic learning strategies where the learning process adapts based on performance metrics. Learning continues only as long as accuracy improves, and the system automatically stops when convergence is reached or when validation performance plateaus, optimizing the balance between accuracy and learning time.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS20250139946A1Information processing device, information processing method, and program

Publication Date: 2025.05.01 SONY GROUP CORP

AI summary

The present disclosure relates to an information processing device, an information processing method, and a program for enabling update and use of a surgical tool recognizer after confirming that the surgical tool recognizer is in a state of being capable of recognizing a new surgical tool. Computer graphics (CG) data serving as learning data to be used to learn a recognizer that recognizes a new surgical tool is generated on the basis of an image including the new surgical tool, the recognizer that recognizes the new surgical tool is caused to learn on the basis of the CG data, an evaluation value that evaluates a recognition result of the new surgical tool by the recognizer based on the image including the new surgical tool is calculated, and the learning of the recognizer based on the CG data is continued according to the evaluation value. The present disclosure can be applied to a surgical image display system.