CT Automatic Exposure Control Using Machine Learning for Patient-Specific Dose Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automatic exposure control (AEC) techniques in computed tomography (CT) scans are inadequate as they do not account for patient-specific anatomy, image quality based on clinical tasks, and are prone to issues like patient centering, leading to suboptimal radiation dose distribution and image quality.
Innovation Solution
The use of machine learning techniques, specifically deep learning models, to tailor AEC settings to specific patient anatomies and clinical tasks by estimating organ doses and image quality from scout images, thereby optimizing tube current profiles and other AEC parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If conventional automatic exposure control (AEC) techniques are used, then radiation dose can be controlled, but image quality is compromised due to uniform noise level across entire volume and lack of patient-specific anatomy consideration
Solution Approach 1:
The patent applies local quality by differentiating image quality requirements across different anatomical regions. Instead of uniform noise control throughout the entire scanned volume, the system identifies target regions (e.g., organs of interest) and background regions, then optimizes exposure parameters specifically for the target regions while allowing different noise levels in background regions. This enables higher image quality where diagnostically important while reducing overall radiation dose.
Solution Approach 2:
The system performs preliminary action by using scout images to pre-identify patient-specific anatomy, target regions, and organs before the actual CT scan. Machine learning models analyze the scout images to predict organ locations, patient body habitus, and clinically relevant regions, then use this information to configure AEC parameters in advance. This preliminary analysis enables personalized dose optimization without requiring the actual diagnostic scan to be performed at higher doses for exploration.
2Object-affected harmful factors
If conventional AEC algorithms based on general patient shape and size are used, then radiation dose can be reduced, but image quality in target regions is compromised due to lack of task-specific optimization
Solution Approach 1:
The system applies parameter changes by dynamically adjusting AEC parameters (tube current modulation, voltage, collimation, pitch) based on task-specific requirements. Different clinical tasks (e.g., lung nodule detection, liver lesion characterization, bone imaging) have different image quality priorities in different regions. The machine learning model selects and optimizes acquisition parameters according to the specific clinical task, enabling the system to adapt to various diagnostic objectives while maintaining dose efficiency.
Solution Approach 2:
The system performs preliminary action by receiving and analyzing task information before the CT scan to determine which regions require optimized image quality. The machine learning model uses the task description to identify target regions and adjust AEC parameters in advance, ensuring that the diagnostic task priorities are met without unnecessary radiation to non-critical regions.
3Object-affected harmful factors
If conventional AEC techniques are used, then radiation dose can be controlled, but dose distribution is suboptimal due to patient centering issues and lack of patient-specific anatomy awareness
Solution Approach 1:
The system applies local quality by identifying patient-specific anatomy and organs in the scout image, then tailoring the radiation dose distribution to each anatomical region. The machine learning model segments the patient anatomy and determines which organs require protection and which regions need higher image quality. This enables non-uniform dose distribution that is optimized for each patient's unique anatomy, delivering appropriate dose to critical regions while minimizing exposure to sensitive organs.
Solution Approach 2:
The system uses feedback by analyzing the scout image to extract patient-specific anatomical information, then using this information to adjust AEC parameters for the actual scan. The machine learning model continuously refines the dose distribution based on the observed anatomy, creating a feedback loop that optimizes both image quality and dose distribution according to the specific patient being scanned.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Techniques are described for tailoring automatic exposure control (AEC) settings to specific patient anatomies and clinical tasks. According to an embodiment, computer-implemented method comprises receiving one or more scout images captured of an anatomical region of a patient in association with performance of a computed tomography (CT) scan. The method further comprises employing a first machine learning model to estimate, based on the one or more scout images, expected organ doses representative of expected radiation doses exposed to organs in the anatomical region under different AEC patterns for the CT scan. The method can further comprises employing a second machine learning model to estimate, based on the one or more scout images, expected measures of image quality in target and background regions of scan images captured under the different AEC patterns, and determining an optimal AEC pattern based on the expected organ doses and the expected measures of image quality.