Diagnosis Support Apparatus for Automated Finding Sentence Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current diagnosis support systems for medical image interpretation are inefficient in creating finding sentences, as existing technologies rely on keyword searches that require extensive manual retrieval and revision, leading to increased workload and fatigue for interpretation doctors.
Innovation Solution
A diagnosis support apparatus that stores image feature information and finding sentences correspondingly, allowing for the acquisition and search of similar image features to automatically generate finding sentences by modifying existing descriptions based on the target area's features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If keyword search is used to retrieve finding sentences, then finding sentences can be retrieved from past interpretation reports, but extensive manual retrieval and revision are required, increasing workload and time consumption
Solution Approach 1:
The patent replaces the mechanical keyword search process with an image-based similarity search system. Instead of manually searching through text keywords, the system automatically compares image features (such as lesion characteristics, anatomical structures, and imaging parameters) to retrieve relevant finding sentences, thereby eliminating the need for extensive manual retrieval and revision work
Solution Approach 2:
The system enables self-service by automatically generating and retrieving finding sentences based on image similarity comparison. The interpretation doctor simply needs to input the current image, and the system autonomously performs the search, retrieval, and presentation of appropriate finding sentences, reducing the doctor's workload to minimal supervision and verification
2Reliability
If interpretation doctors manually create finding sentences, then accuracy can be maintained, but workload increases and fatigue sets in, leading to mistakes and oversights
Solution Approach 1:
The patent utilizes copying by retrieving and adapting finding sentences from past interpretation reports that have similar image features. Instead of creating finding sentences from scratch, the system copies proven, accurate descriptions from previous cases and modifies them to fit the current image, thereby maintaining accuracy while significantly reducing the effort and time required
Solution Approach 2:
The system performs preliminary action by pre-processing and storing finding sentences from past reports in a database organized by image features. When a new image is interpreted, the system has already prepared relevant finding sentences through similarity comparison, so the interpretation doctor only needs to review and confirm the results rather than creating descriptions from scratch
3Reliability
If CAD apparatus is used to detect disease candidate images, then disease oversight can be reduced, but additional work is required to reexamine detected images
Solution Approach 1:
The patent merges the functions of disease detection and finding sentence generation into a unified workflow. The system combines CAD-based disease candidate detection with automatic finding sentence retrieval based on image similarity, so that the reexamination process is integrated with the report creation process, reducing the perceived workload by combining multiple tasks into a single coherent workflow
Data Source
AI summary
A diagnosis support apparatus includes a storage unit which stores pieces of image feature information and finding sentences as interpretation results in correspondence with each other. This apparatus acquires image feature information of a target area designated on an image to be interpreted, searches the storage unit for image feature information similar to the acquired image feature information, acquires a finding sentence stored in correspondence with the retrieved image feature information from the storage unit, and creates a finding sentence concerning interpretation of the designated target area by changing a description of the acquired finding sentence based on image feature information of the designated target area.