CAD System Grouping Image Frames for Consistent Lesion Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Computer Aided Diagnosis (CAD) systems face challenges in providing a unified classification result for lesions across multiple image frames, leading to inconsistent diagnoses due to limited human perception and time-consuming analysis, as they struggle to combine information effectively from different image frames.
Innovation Solution
A CAD system comprising a Region of Interest (ROI) detector, a categorizer, a classifier, and a result combiner that groups successive image frames, classifies ROIs, and combines classification results using BI-RADS Lexicon or Category information, calculating scores and determining malignancy based on histogram analysis and model fitness, while also using histogram clustering to combine similar groups and select representative images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a CAD system classifies each region of interest (ROI) detected from a different image frame independently, then the classification process is simple and fast, but different classification results are obtained regarding image frames of the same lesion, leading to inconsistency
Solution Approach 1:
The patent merges multiple independent classification results from different image frames by grouping frames that contain the same lesion and combining their classification outcomes. This is achieved through a result combiner that aggregates classification results across multiple frames, ensuring consistent diagnosis for the same lesion while maintaining efficient processing.
2Measurement precision
If human medical professionals analyze each medical image manually, then detailed examination is possible, but it requires a large amount of time, great attention, and care
Solution Approach 1:
The patent replaces the mechanical system of manual human analysis with an automated computer-based classification system. The CAD system uses algorithms to automatically detect, classify, and combine results from multiple image frames, eliminating the need for time-consuming manual examination while maintaining or improving diagnostic accuracy through consistent application of classification criteria.
3Productivity
If the CAD system processes each image frame separately, then the processing is efficient, but information from all image frames for each lesion is not combined, leading to multiple classification results
Solution Approach 1:
The patent introduces an intermediary component (result combiner) that bridges the gap between independent frame processing and integrated lesion diagnosis. This intermediary aggregates classification results from multiple frames, combines information about the same lesion detected across different frames, and produces a unified classification outcome, thereby preventing information loss while maintaining processing efficiency.
Data Source
AI summary
Disclosed are Computer Aided Diagnosis (CAD) apparatus and method to combine information on sequential image frames and to provide a superior classification result for the ROI in the image frame. The CAD apparatus may include a Region of Interest (ROI) detector configured to detect an ROI from image frames, a categorizer configured to create groups of image frames having successive ROI sections from among the image frames based on a result of the detection, a classifier configured to classify an ROI detected from each of the image frames belonging to the groups, and a result combiner configured to combine classification results for the image frames belonging to each group from the groups and to calculate a group result for the each group.