Operating Room Object Detection via Audio-Video Correlation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current operating room systems face challenges in providing intuitive and user-friendly interaction with medical devices, effective data collection, and efficient data tagging and archiving, particularly in complex environments with multiple high-tech tools, leading to difficulties in identifying important information from large amounts of video and audio data.

Innovation Solution

A method and system that combines video and audio data streams to accurately detect and track objects within the operating room, using object recognition and audio event association to enhance object identification and control, thereby improving device control and data collection by linking visual and auditory cues.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If multiple high-tech medical devices are added to the operating room to improve patient safety and surgical capabilities, then the functional capabilities are enhanced, but the device complexity and difficulty of monitoring tracking increase

Engineering Contradiction:
Improvepatient safetyVSAvoidOR environment complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary system comprising video cameras, audio sensors, and processing units that mediate between the multiple medical devices and the monitoring personnel. This intermediary automatically tracks objects, detects gestures, recognizes audio commands, and correlates data from various devices, reducing the complexity burden on the medical team while maintaining enhanced safety capabilities through integrated monitoring.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Quantity of substance

If video and audio monitoring systems are deployed to collect large amounts of data for improved analysis, then the data collection capability is enhanced, but the difficulty of identifying important information increases

Engineering Contradiction:
Improvedata volumeVSAvoidinformation identification difficulty
Core Design Contradiction:
Quantity of substanceVSDifficulty of detecting and measuring

Solution Approach 1:

The patent replaces manual review of video and audio data with automated computational systems. Machine learning algorithms analyze video streams to detect gestures and objects, audio processing systems recognize commands and speech patterns, and correlation algorithms automatically identify important information by cross-referencing multiple data streams, thereby reducing the difficulty of information identification while maintaining comprehensive data collection.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system implements feedback mechanisms where detected gestures, audio commands, and object tracking information are continuously correlated and used to adjust monitoring focus. The system provides feedback loops that automatically highlight important events, tag relevant data segments, and guide attention to critical information, making large volumes of data manageable and easily analyzable.

Inventive Principle:
Principle #23Feedback

3Measurement precision

If manual review of recorded video and audio data is performed to identify important information, then thorough analysis is achieved, but the loss of time increases

Engineering Contradiction:
Improveinformation analysis thoroughnessVSAvoiddata review time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary automated analysis of video and audio data during the procedure itself. Object tracking, gesture detection, and audio recognition occur in real-time or near-real-time, with important events being automatically tagged and flagged as they happen. This preliminary action eliminates the need for lengthy post-procedure manual review while maintaining thorough analysis through continuous automated monitoring.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4276777A1Object detection in an operating room
Publication Date: 2023.11.15 BAXTER MEDICAL SYST GMBH & CO KG
  • EP4276777A1 patent drawingFigure 1
  • EP4276777A1 patent drawingFigure 2
  • EP4276777A1 patent drawingFigure 3

AI summary

A method of object detection in an operating room is provided. A method of detecting objects in a video stream of an operating room in provided. The method comprises receiving a video stream of an operating room; receiving an audio stream of the operating room; detecting, from video data of the video stream, a first data object corresponding to a first physical object within the operating room; detecting, from the audio stream, audio data corresponding to an audio event within the operating room; determining that the audio data is associated with the first data object; and in response to determining that the audio data is associated with the first data object, detecting, from video data of the video stream, a second data object in dependence on the first data object, the second data object corresponding to a second physical object located in the operating room.