ML Image Quality Assessment for Medical Imaging Retake Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical imaging technologies face challenges in efficiently assessing image quality, leading to unnecessary retakes that increase patient stress, imaging equipment wear, and X-ray dosage, as low-quality images often require re-acquisition without adequate support for immediate decision-making.
Innovation Solution
A system that utilizes an image quality assessment module and an imaging triaging module to retrieve and assess prior images of similar patients, allowing for informed decisions on whether to re-take images based on pre-defined quality scores, thereby reducing unnecessary retakes by using high-quality prior images as substitutes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If ML-based quality assessment is implemented to prevent low-quality studies, then image quality is improved, but device complexity increases
Solution Approach 1:
An image quality assessment module acts as an intermediary between the imaging apparatus and the radiological evaluation process. This module automatically evaluates image quality metrics (such as exposure adequacy, motion artifacts, and anatomical coverage) and provides feedback to determine whether a study is diagnostic or requires retake, thereby improving reliability without requiring direct complex intervention in the imaging process itself
Solution Approach 2:
The system performs self-service quality assessment by automatically evaluating its own output images through ML algorithms. The image quality assessment module independently analyzes acquired images against predefined quality criteria and generates quality scores, enabling the system to self-regulate and identify non-diagnostic studies without external intervention, thus managing complexity internally
2Reliability
If unnecessary image retakes are performed, then diagnostic accuracy is maintained, but patient stress and X-ray dosage increase
Solution Approach 1:
The image quality assessment is performed preliminarily and immediately after image acquisition, before the radiologist begins evaluation. This preliminary quality check identifies non-diagnostic studies in advance, allowing the system to flag them for potential retake only when necessary. By assessing quality before clinical interpretation, the system prevents unnecessary patient recall and radiation exposure while ensuring diagnostic studies are not missed
Solution Approach 2:
A feedback mechanism is established where image quality metrics are automatically evaluated and fed back to the imaging workflow. The system provides real-time feedback on quality parameters (exposure, positioning, motion) and communicates quality scores to operators, enabling immediate corrective actions if needed while avoiding unnecessary retakes of adequate images, thus reducing patient stress and radiation exposure
3Productivity
If immediate quality assessment is performed after acquisition, then productivity is improved, but measurement precision requirements increase
Solution Approach 1:
The image quality assessment is segmented into multiple independent evaluation components, each focusing on specific quality metrics such as exposure adequacy, motion artifacts, anatomical coverage, and positioning accuracy. This segmentation allows parallel processing of different quality dimensions, enabling comprehensive assessment to be completed rapidly without compromising precision, thus maintaining high imaging throughput
Data Source
AI summary
System (SYS) and related method for imaging support. The system comprises an input interface (IN) for receiving an input image (I1) of a patient acquired by an imaging apparatus (IA). The input image was previously assessed by an image quality assessment module (IQM) and was awarded an image quality, IQ, score. The system's imaging triaging module (TL) retrieves from an image database (DBI) of prior assessed images, a corresponding image (I0) that corresponds to the input image. The corresponding image is a prior image of patient. The imaging triaging module (TL) provides a decision, based on the first IQ score and a second IQ score of the corresponding image (I0), if any, whether or not to acquire a new image (I2) of the patient by the imaging apparatus (IA). The risk of unnecessary retakes can be reduced, thus saving time, dose, and reducing wear-and-tear of the imaging apparatus.