Endoscope Image Recognition for Report Input Automation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current endoscope examination report creation methods require time-consuming input of information on sites, disease names, and characteristics, disrupting the examination process.
Innovation Solution
An information processing apparatus that uses image recognition to detect specific features during an endoscope examination, displaying relevant options on a screen for selection, allowing for efficient input of report information without interrupting the examination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If information on site, disease name, and characteristics is input collectively during endoscope examination, then complete report information is obtained, but examination process is interrupted and time consumption increases
Solution Approach 1:
The system performs preliminary recognition and analysis of endoscopic images using AI algorithms during the examination process. The recognizer automatically identifies sites, disease names, and characteristics, pre-processing the information that would otherwise require manual input, thereby reducing both information loss and time consumption
Solution Approach 2:
The system enables self-service by allowing the recognizer to automatically generate and populate report information without requiring continuous manual intervention. The recognized information is automatically recorded and associated with the examined site, making the system serve itself in information gathering while the physician continues the examination
2Measurement precision
If manual input of report information is required during examination, then accurate data entry is achieved, but operational efficiency decreases
Solution Approach 1:
The system replaces the mechanical manual input process with an automated image recognition system. The recognizer analyzes endoscopic images and automatically extracts and records information about sites, disease names, and characteristics, substituting the manual typing and selection process with automated computer vision and machine learning algorithms
Solution Approach 2:
The recognizer acts as an intermediary between the endoscopic image and the final report. It processes the visual information from the endoscope and translates it into structured report data, mediating between the raw examination data and the required report format while maintaining accuracy
3Loss of information
If multiple selection screens are displayed sequentially for site, disease name, and characteristics, then comprehensive information is collected, but user operation complexity increases
Solution Approach 1:
The system merges multiple sequential selection screens into a single integrated display. The recognizer presents all necessary selections (site, disease name, characteristics) simultaneously on one screen based on the analyzed image, combining what were previously separate sequential steps into a unified user interaction interface
Solution Approach 2:
The system performs preliminary analysis of the endoscopic image to pre-determine the relevant sites, disease names, and characteristics before displaying the selection screen. This preliminary recognition allows the system to prepare and display only the relevant options that match the recognized features, reducing the complexity of user operation
Data Source
AI summary
There are provided an information processing apparatus, an information processing method, an endoscope system, and a report creation support device which can efficiently input information necessary for generating a report. Images captured by an endoscope are acquired, and the acquired images are displayed on a display unit. In addition, the acquired images are input to a plurality of recognizers, and a recognizer that has output a specific recognition result is detected from among the plurality of recognizers. Options for an item corresponding to the detected recognizer is displayed on the display unit, and an input of selection for the displayed options is accepted.