Radiological Report Viewer with Semantic Highlighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Physicians face challenges in efficiently comparing and diagnosing medical images due to the time-consuming and error-prone process of manually reviewing and correlating structured BIRADS reports with legacy unstructured free-text reports, which hinders effective patient diagnosis and treatment.
Innovation Solution
A method and system that utilize text recognition and semantic analysis to identify and highlight relevant sentences in unstructured reports matching BIRADS descriptors, allowing for efficient comparison and correlation between structured and unstructured medical reports, using a processor and modules like report analyzers, ontology engines, and reasoning engines to translate and match descriptors with report interpretations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physicians manually review and correlate structured BIRADS reports with legacy unstructured free-text reports, then diagnostic accuracy is maintained, but time consumption increases significantly
Solution Approach 1:
The patent introduces an intermediary system comprising a report analyzer, ontology engine, and reasoning engine that acts as a mediator between structured BIRADS reports and unstructured legacy reports. This intermediary automatically performs semantic analysis, extracts relevant information, and presents correlated findings to physicians, thereby maintaining diagnostic accuracy while eliminating the time-consuming manual review process.
Solution Approach 2:
The patent replaces the mechanical manual process of reading and correlating reports with an automated computational system. The report analyzer uses text recognition and semantic analysis algorithms to automatically extract and correlate information from both structured and unstructured reports, substituting the physician's manual cognitive work with automated intelligent processing while preserving diagnostic accuracy.
2Reliability
If physicians manually compare current images with prior images and reports, then lesion progression is accurately detected, but workflow efficiency decreases
Solution Approach 1:
The system performs preliminary automated analysis by pre-extracting and organizing relevant information from prior reports and images before the physician reviews the case. The report analyzer预先 identifies key findings, measurements, and observations from historical data, presenting them in a correlated manner that enables physicians to quickly assess lesion progression without manually searching through multiple prior studies.
Solution Approach 2:
The patent segments the complex task of comparing current and prior images and reports into distinct automated processing steps: text extraction from unstructured reports, semantic analysis to identify relevant findings, structured data correlation, and presentation of comparative results. This segmentation allows each step to be optimized independently while collectively improving workflow efficiency without compromising lesion progression detection accuracy.
3Adaptability or versatility
If unstructured free-text reports are converted to structured format using text recognition, then data integration is improved, but system complexity increases
Solution Approach 1:
The patent implements a universal ontology engine that serves multiple functions: it provides a standardized vocabulary for structuring extracted data, enables semantic analysis across different report formats, facilitates correlation between structured and unstructured data, and supports various query types. This multi-functional ontology system improves data integration capability while managing complexity through a unified approach rather than separate specialized systems.
Solution Approach 2:
The ontology engine acts as an intermediary layer between the text recognition process and the structured data storage system. It translates extracted text into standardized semantic representations using controlled vocabularies and relationships, thereby improving data integration capability while isolating the complexity of unstructured text processing from the structured data infrastructure.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A method and a report viewer for viewing a structured report, such as medical report describing radiological images using descriptors selected from a predefined list of descriptors, includes the acts of opening the medical report; and in response to the opening act, searching for a further report related to the descriptors of the medical report, and highlighting words and/or sentences in the further report that match keywords derived from the descriptors. The medical report and the further report may be displayed simultaneously with the words and/or sentences being highlighted. The further report may include an unstructured text report, and the method further includes mapping the descriptors to findings in the text report and highlighting the findings.