AI Pulmonary Function Grading From X-Rays Despite Image Variation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AI-based systems for assessing lung function in COVID-19 patients are limited by variations in X-ray image quality due to the wide range of X-ray machines, and there is a need for an objective metric to differentiate severe from mild/moderate cases efficiently.
Innovation Solution
A CAD system using AI and machine learning techniques that extracts rotation, scale, and translation-invariant imaging markers from X-ray images to grade COVID-19 severity, and an alternative approach using CT images for accurate diagnosis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If AI-based systems are used to assess lung function in COVID-19 patients, then diagnostic accuracy is improved, but the system performance deteriorates due to variations in X-ray image quality from different machines
Solution Approach 1:
The patent transforms X-ray images into a standardized representation space by converting them to polar coordinates and applying intensity normalization. This parameter transformation adjusts the image data to be independent of the original imaging device, thereby maintaining diagnostic accuracy while eliminating variations caused by different X-ray machine qualities and settings.
Solution Approach 2:
The patent introduces an intermediary representation layer that acts as a mediator between the input X-ray images and the diagnostic AI model. By converting images to polar coordinates and normalizing intensities, this intermediary layer standardizes the data format, allowing the AI system to process images from various sources consistently without performance degradation.
2Measurement precision
If manual analysis of X-ray images is performed by radiologists, then diagnostic accuracy is maintained, but productivity decreases due to time-consuming analysis and high workload
Solution Approach 1:
The patent implements a self-service diagnostic system where the AI model automatically analyzes X-ray images without requiring radiologist intervention for each case. The system performs autonomous image processing, severity classification, and risk assessment, thereby maintaining diagnostic accuracy while significantly increasing productivity by eliminating manual analysis time.
Solution Approach 2:
The patent replaces the mechanical process of manual radiological analysis with an automated AI-based system. The AI model substitutes for radiologist expertise in image interpretation, enabling rapid processing of numerous images while maintaining diagnostic accuracy through trained neural networks that replicate and enhance human diagnostic capabilities.
3Measurement precision
If CT scans are used for diagnosing COVID-19, then measurement precision is improved, but the harmful factors increase due to higher radiation dose
Solution Approach 1:
The patent leverages the abundance of low-cost, low-radiation chest X-ray images that are routinely obtained for COVID-19 screening. Instead of relying on expensive and high-radiation CT scans, the system processes these readily available X-ray images through AI algorithms, thereby maintaining diagnostic accuracy while avoiding the harmful radiation exposure associated with CT imaging.
Solution Approach 2:
The patent transforms standard chest X-ray images through parameter changes (polar coordinate conversion and intensity normalization) to enhance the diagnostic information content. This transformation allows the system to extract accurate diagnostic features from low-radiation X-rays that would otherwise be insufficient, replacing the need for high-radiation CT scans while maintaining measurement precision.
Data Source
AI summary
Assessment of pulmonary function in coronavirus patients includes use of a computer aided diagnostic system to assess pulmonary function and risk of mortality in patents with coronavirus disease 2019. The CAD system processes thoracic X-ray data from a patient, extracts imaging markers, and grades disease severity based at least in part on the extracted imaging markers, thereby distinguishing between higher risk and lower risk patients. An alternative approach is to use an automatic CAD system to grade COVID-19 from computed tomography (CT) images to determine an accurate diagnosis of lung function.