Pulmonary Embolism Detection Using Radiologist-Guided Vessel Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer-aided detection methods for pulmonary embolisms in CT images often incorrectly identify pulmonary veins as arteries, leading to increased time and effort in reading CT sections due to the difficulty in distinguishing between the two vessel types.
Innovation Solution
A method where a radiologist manually identifies and marks pulmonary arteries and veins in CT sections using a cursor and mouse, with the computer combining these markings to differentiate between arteries and veins, thereby reducing the number of vessels that need to be examined for embolisms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If computer-aided detection uses automated algorithms to identify pulmonary arteries, then detection speed is improved, but accuracy deteriorates due to false positives from misidentifying veins as arteries
Solution Approach 1:
The patent merges automated computer algorithms with radiologist expertise by having the computer perform initial vessel segmentation and then use radiologist markings to train and refine the algorithm. This combination allows the system to maintain high detection speed while improving accuracy through human-in-the-loop learning.
Solution Approach 2:
The radiologist performs preliminary manual markings of arteries and veins during initial review, which then serve as training data for the computer algorithm. This preliminary human action enables subsequent automated processing to be both fast and accurate.
2Reliability
If the computer reviews all vessels including false positive veins, then comprehensive detection is improved, but time consumption worsens due to examining unnecessary vessels
Solution Approach 1:
The system performs partial review by using radiologist markings to identify and exclude veins from automated embolism detection. This partial action (reviewing only arteries) maintains reliability for detecting pulmonary embolisms while significantly reducing time consumption by avoiding review of veins where embolisms cannot occur.
3Extent of automation
If the computer autonomously identifies vessels without radiologist input, then automation is improved, but differentiation accuracy worsens between arteries and veins
Solution Approach 1:
The system implements feedback loops where radiologist markings of arteries and veins are fed back to train and refine the computer algorithm. This iterative feedback process enables the system to progressively improve autonomous identification accuracy while maintaining high automation levels.
Data Source
AI summary
In a preferred embodiment a radiologist traces the pulmonary artery and pulmonary veins visible in a set of CT images and identifies the arteries and veins. The radiologist's identification of the pulmonary arteries and pulmonary veins is then received by an image analyzer and combined with the analyzer's identification of the pulmonary arteries to form a combined identification; and the analyzer then reviews this combined identification of the pulmonary arteries to detect any pulmonary embolisms. The radiologist's identification of any pulmonary embolisms is compared with the analyzer's identification of any pulmonary embolisms to determine if there are any embolisms identified by the analyzer that were not identified by the radiologist.