# US 12,002,204 B2 · Patient anatomy and task specific automatic exposure control in computed tomography > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,002,204 B2 - Document ID: US12002204B2 - Status: Granted - Kind: B2 - Application number: 17/471,532 - Filed: 2021-09-10 - Published: 2023-03-16 - Granted: 2024-06-04 - Earliest claimed: 2021-09-10 - Assignees: GE Precision Healthcare LLC; The Board of Trustees of the Leland Stanford Junior University - Applicants: GE Precision Healthcare LLC; The Board of Trustees of the Leland Stanford Junior University - Inventors: Adam S. Wang; Debashish Pal; Abdullah-Al-Zubaer Imran; Sen Wang; Evan Zucker; Bhavik Natvar Patel - Web page: https://patentgrep.com/patent/US12002204B2 ## CPC - A61B 6/488: {involving pre-scan acquisition} - G06T 7/0012: {Biomedical image inspection} - A61B 6/032: {Transmission computed tomography [CT]} - A61B 6/542: {involving control of exposure} - A61B 6/545: {involving automatic set-up of acquisition parameters} - G06N 3/08: Learning methods - G06N 20/00: Machine learning - G06V 10/25: Determination of region of interest [ROI] or a volume of interest [VOI] - G06T 2207/10081: Computed x-ray tomography [CT] - G06T 2207/10144: Varying exposure - G06T 2207/20081: Training; Learning - G06T 2207/30168: Image quality inspection - A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D) - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract 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. ## Claims ### Claim 1 A system, comprising: a memory that stores computer executable components; and a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise: a training component that trains a first deep learning network to determine expected organ doses absorbed by one or more organs of an anatomical region under different automatic exposure control patterns of a computed tomography scan based on scout images captured of the anatomical region for different patients, and a second deep learning network to determine expected measures of image quality in target regions and background regions of computed tomography scan images captured under the different automatic exposure control patterns based on the scout images; and an optimization component that employs the first deep learning network and the second deep learning network to determine optimal automatic exposure control patterns of the different automatic exposure control patterns for performance of the computed tomography scan for the different patients based on the scout images. ### Claim 2 (depends on 1) The system of claim 1, wherein the optimization component determines the optimal automatic exposure control patterns using an objective function that selects the optimal automatic exposure control patterns based on maximizing image quality in the target regions and minimizing the expected organ doses. ### Claim 3 (depends on 1) The system of claim 1, wherein the training component trains the first deep learning network using a supervised machine learning process and ground truth data comprising a plurality of computed tomography volumes generated from computed tomography scans of the anatomical region under the different automatic exposure control patterns and organ segmentation dose maps generated from the computed tomography volumes. ### Claim 4 (depends on 1) The system of claim 1, wherein the training component trains the second deep learning network using a supervised machine learning process and ground truth data image quality data generated using a plurality of computed tomography volumes generated from computed tomography scans of the anatomical region under the different automatic exposure control patterns, wherein the ground truth image quality data provides measures of image quality in the target regions and the background regions as represented in the computed tomography volumes under the different automatic exposure control patterns. ### Claim 5 (depends on 4) The system of claim 4, wherein the computer executable components further comprise: a simulation component that generates simulated noise projections using the computed tomography volumes and generates the ground truth image quality data based on the simulated noise projections. ### Claim 6 (depends on 1) The system of claim 1, wherein the different automatic exposure control patterns reflect different tube current modulations. ### Claim 7 (depends on 1) The system of claim 1, wherein the different automatic exposure control patterns reflect different acquisition parameter values for one or more acquisition parameters selected from the group consisting of: tube voltage, collimation, filtration, bowtie, pitch, and start angle. ### Claim 8 (depends on 1) The system of claim 1, wherein the target regions and the background regions vary for different tasks and the optimal automatic exposure control patterns respectively represent different scout image and task combinations of the scout images and the different tasks, and wherein the training component further trains a third deep learning network to determine the optimal automatic exposure control patterns for each of the different scout image and task combinations. ### Claim 9 (depends on 8) The system of claim 8, wherein the computer executable components further comprise: an inferencing component that employs the third deep learning network to determine an optimal automatic exposure pattern for a new computed tomography scan of the anatomical region of a patient based on one or more new scout images captured of the anatomical region of the patient and a selected task of the different clinical tasks. ### Claim 10 (depends on 9) The system of claim 9, wherein the inferencing component further employs the first deep learning network to determine, based on the one or more new scout images, expected organ doses absorbed by the one or more organs of the patient under the optimal automatic exposure pattern, and employs the second deep learning network to determine an expected measure of image quality in the target and background regions of scan images to be captured under the optimal automatic exposure pattern. ### Claim 11 (depends on 9) The system of claim 9, wherein the computer executable components further comprise: a reception component that receives the one or more new scout images from a computed tomography imaging device prior to performance of the new computed tomography scan using the imaging device; and a control component operatively coupled to the computed tomography imaging device that controls performance of the new computed tomography scan by the computed tomography imaging device using the optimal automatic exposure pattern. ### Claim 12 (depends on 11) The system of claim 11, wherein the computer executable components further comprise: an interface component that facilitates receiving user input adjusting one or more parameters of the optimal exposure control pattern, resulting in a modified automatic exposure control pattern, and wherein based on reception of the user input, the inferencing component further employs the first deep learning network to determine, based on the one or more new scout images, updated expected organ doses absorbed by the one or more organs of the patient under the updated automatic exposure pattern, and employs the second deep learning network to determine an updated expected measure of image quality in the target and background regions of scan images to be captured under the updated automatic exposure pattern. ### Claim 13 (depends on 9) The system of claim 9, wherein the computer executable components further comprise: an interface component that facilitates receiving user input identifying at least one of, a desired image quality for the target region and a desired radiation dose to the one or more organs, and wherein based on reception of the user input, the inferencing component employs the third deep learning network to determine, based on the one or more scout images, a modified automatic exposure control pattern that achieves the desired image quality and the desired radiation dose. ### Claim 14 A method, comprising: training, by a system operatively coupled to a processor, a first deep learning network to determine expected organ doses absorbed by one or more organs of an anatomical region under different automatic exposure control patterns of a computed tomography scan based on scout images captured of the anatomical region for different patients; training, by the system, a second deep learning network to determine expected measures of image quality in target regions and background regions of computed tomography scan images captured under the different automatic exposure control patterns based on the scout images; and employing, by the system, the first deep learning network and the second deep learning network to determine optimal automatic exposure control patterns of the different automatic exposure control patterns for performance of the computed tomography scan for the different patients based on the scout images. ### Claim 15 (depends on 14) The method of claim 14, wherein the employing further comprises determining the optimal automatic exposure control patterns using an objective function that selects the optimal automatic exposure control patterns based on maximizing image quality in the target regions and minimizing the expected organ doses. ### Claim 16 (depends on 14) The method of claim 14, wherein the training the first deep learning network comprises employing a supervised machine learning process and ground truth data comprising a plurality of computed tomography volumes generated from computed tomography scans of the anatomical region under the different automatic exposure control patterns and organ segmentation dose maps generated from the computed tomography volumes. ### Claim 17 (depends on 14) The method of claim 14, wherein the training the first deep learning network comprises employing a supervised machine learning process and ground truth data image quality data generated using a plurality of computed tomography volumes generated from computed tomography scans of the anatomical region under the different automatic exposure control patterns, wherein the ground truth image quality data provides measures of image quality in the target regions and the background regions as represented in the computed tomography volumes under the different automatic exposure control patterns. ### Claim 18 (depends on 14) The method of claim 14, wherein the target regions and the background regions vary for different tasks and the optimal automatic exposure control patterns respectively represent different scout image and task combinations of the scout images and the different tasks, and wherein the method further comprises: training, by the system, a third deep learning network to determine the optimal automatic exposure control patterns for each of the different scout image and task combinations; employing, by the system, the third deep learning network to determine an optimal automatic exposure pattern for a new computed tomography scan of the anatomical region of a patient based on one or more new scout images captured of the anatomical region of the patient and a selected task of the different clinical tasks. ### Claim 19 A machine-readable storage medium, comprising executable instructions that, when executed by a processor, facilitate performance of operations, comprising: training a first deep learning network to determine expected organ doses absorbed by one or more organs of an anatomical region under different automatic exposure control patterns of a computed tomography scan based on scout images captured of the anatomical region for different patients; training a second deep learning network to determine expected measures of image quality in target regions and background regions of computed tomography scan images captured under the different automatic exposure control patterns based on the scout images; and employing the first deep learning network and the second deep learning network to determine optimal automatic exposure control patterns of the different automatic exposure control patterns for performance of the computed tomography scan for the different patients based on the scout images. ### Claim 20 (depends on 19) The machine-readable storage medium of claim 19, wherein the target regions and the background regions vary for different tasks and the optimal automatic exposure control patterns respectively represent different scout image and task combinations of the scout images and the different tasks, and wherein the operations further comprise: training a third deep learning network to determine the optimal automatic exposure control patterns for each of the different scout image and task combinations; employing the third deep learning network to determine an optimal automatic exposure pattern for a new computed tomography scan of the anatomical region of a patient based on one or more new scout images captured of the anatomical region of the patient and a selected task of the different clinical tasks.