Atlas-Based Prior Relevancy Model for Medical Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying relevant prior medical studies are inefficient due to vague procedure names and codes, leading to delays and missed studies, especially when dealing with large numbers of patient records, as they require manual examination of reports and thumbnail images.
Innovation Solution
An atlas-based prior relevancy and stickman relevancy model is introduced, using graphical tools to match imaging studies with anatomical atlases, allowing users to select relevant body parts and visualize anatomical coverage, enabling rapid filtering and precise characterization of anatomical span.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual examination of reports and thumbnail images is used to identify prior studies, then comprehensive review is possible, but time consumption increases significantly
Solution Approach 1:
The patent replaces the manual mechanical process of examining reports and thumbnail images with an automated computer-based system that uses natural language processing, entity recognition, and anatomical mapping algorithms to automatically identify and retrieve relevant prior studies, thereby eliminating time-consuming manual work while maintaining comprehensive review capability
Solution Approach 2:
The patent introduces an intermediary automated retrieval system that acts as a bridge between the user's information needs and the large volume of stored medical studies. This intermediary system pre-processes and organizes studies by extracting anatomical entities and relationships, enabling rapid retrieval without requiring manual examination of each study
2Productivity
If rule-based methods are used to identify relevant priors, then automation is achieved, but accuracy decreases due to vague procedure names and codes
Solution Approach 1:
The patent replaces simple rule-based matching with advanced natural language processing and entity recognition systems that can interpret vague procedure names and codes by extracting meaningful anatomical entities and relationships from unstructured text, thereby maintaining automation while significantly improving accuracy
Solution Approach 2:
The patent transforms the matching parameters from simple procedure names and codes to detailed anatomical entities, relationships, and contextual features extracted through NLP. This parameter transformation enables more precise matching by capturing the semantic meaning behind vague procedure descriptions rather than relying on literal string matching
3Adaptability or versatility
If comprehensive lists of rules are maintained for procedure names, then coverage is improved, but system complexity increases and maintenance becomes difficult
Solution Approach 1:
The patent replaces the complex mechanical system of maintaining comprehensive rule lists with an intelligent NLP-based system that dynamically extracts anatomical entities and relationships from procedure descriptions. This system adapts to new procedure names and codes automatically through language understanding rather than requiring manual rule updates, thereby maintaining versatility while reducing complexity
Solution Approach 2:
The patent enables the system to self-adapt to new procedure names and codes by using NLP to extract anatomical meaning from unstructured text. The system automatically learns and adjusts to new terminology through its language processing capabilities without requiring manual intervention to update rule lists, thereby maintaining comprehensive coverage while simplifying system maintenance
Data Source
AI summary
User interfaces for navigating medical studies are provided. In various embodiments, a human avatar having a plurality of selectable regions is displayed. Indications of the presence of prior studies are displayed corresponding to the plurality of selectable regions. A selection of a region of the plurality of selectable regions is received from a user. An indication of one or more prior study is displayed corresponding to the selected region.