# US 11,992,348 B2 · Machine learning model to adjust C-arm cone-beam computed tomography device trajectories > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,992,348 B2 - Document ID: US11992348B2 - Status: Granted - Kind: B2 - Application number: 17/634,597 - Filed: 2020-09-04 - Published: 2022-10-13 - Granted: 2024-05-28 - Earliest claimed: 2019-09-05 - Assignees: The Johns Hopkins University - Applicants: The Johns Hopkins University - Inventors: Mathias Unberath; Jan-Nico Zaech - Web page: https://patentgrep.com/patent/US11992348B2 ## CPC - A61B 6/032: {Transmission computed tomography [CT]} - A61B 6/12: Arrangements for detecting or locating foreign bodies - A61B 6/4085: {Cone-beams} - A61B 6/4441: {the rigid structure being a C-arm or U-arm} - A61B 6/505: {for diagnosis of bone} - A61B 6/5258: {involving detection or reduction of artifacts or noise} - A61B 6/563: {involving image data transmission via a network} - G16H 20/40: relating to mechanical, radiation or invasive therapies, e.g. surgery, laser therapy, dialysis or acupuncture - G16H 30/20: for handling medical images, e.g. DICOM, HL7 or PACS - G16H 30/40: for processing medical images, e.g. editing - A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D) - G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA ## Abstract A device may receive an X-ray image captured by a C-arm CBCT device at a particular position defined by a six-degree of freedom pose relative to an anatomy, and may process the X-ray image, with a machine learning model, to determine a predicted quality of next possible X-ray images provided by the C-arm CBCT device. The device may utilize the machine learning model, to identify a particular X-ray image, of the next possible X-ray images, with a greatest predicted quality and to update the six-degree of freedom pose based on the particular X-ray image. The device may provide, to the C-arm CBCT device, data that identifies the updated six-degree of freedom pose to cause the C-arm CBCT device to adjust to a new position based on the updated six-degree of freedom pose. ## Claims ### Claim 1 A method comprising: receiving, by a device, an X-ray image captured by a C-arm cone-beam computed tomography (CBCT) device at a particular position defined by six-degree of freedom pose relative to an anatomy; processing, by the device, the X-ray image, with a machine learning model, to determine a predicted quality of next possible X-ray images provided by the C-arm CBCT device; utilizing, by the device, the machine learning model, to: identify a particular X-ray image, of the next possible X-ray images, with a greatest predicted quality, and update the six-degree of freedom pose based on the particular X-ray image and to generate an updated six-degree of freedom pose; and providing, by the device and to the C-arm CBCT device, data that identifies the updated six-degree of freedom pose, wherein the data that identifies the updated six-degree of freedom pose cause the C-arm CBCT device to adjust to a new position based on the updated six-degree of freedom pose. ### Claim 2 (depends on 1) The method of claim 1, wherein the data that identifies the updated six-degree of freedom pose cause the C-arm CBCT device to acquire a new X-ray image at the new position. ### Claim 3 (depends on 1) The method of claim 1, wherein the X-ray image is captured by the C-arm CBCT device to verify screw placement in a spine of a patient during spinal fusion surgery. ### Claim 4 (depends on 1) The method of claim 1, wherein the six-degree of freedom pose includes an in-plane angle that is defined according to a circular trajectory where a source and a detector of the C-arm CBCT device move in a plane for an entire scan, and wherein the six-degree of freedom pose includes an out-of-plane angle that is associated with tilting a C-arm, of the C-arm CBCT device, relative to the plane. ### Claim 5 (depends on 1) The method of claim 1, wherein the machine learning model includes a visual geometry group (VGG) convolutional neural network model. ### Claim 6 (depends on 1) The method of claim 1, wherein processing the X-ray image, with the machine learning model, to determine the predicted quality of the next possible X-ray images provided by the C-arm CBCT device comprises: regressing a detectability index for the next possible X-ray images to determine the predicted quality of the next possible X-ray images. ### Claim 7 (depends on 6) The method of claim 6, wherein regressing the detectability index for the next possible X-ray images to determine the predicted quality of the next possible X-ray images comprises: determining a local modulation transfer function based on the X-ray image; determining a local noise power spectrum based on the X-ray image; determining a task function describing properties of an object, provided in the X-ray image, with a quality in Fourier space; and regressing the detectability index for the next possible X-ray images to determine the predicted quality of the next possible X-ray images based on the local modulation transfer function, the local noise power spectrum, and the task function. ### Claim 8 A device, comprising: one or more memories; and one or more processors, communicatively coupled to the one or more memories, configured to: receive an X-ray image captured by a C-arm cone-beam computed tomography (CBCT) device at a particular position defined by a six-degree of freedom pose relative to an anatomy; process the X-ray image, with a machine learning model, to determine a predicted quality of next possible X-ray images provided by the C-arm CBCT device; utilize the machine learning model, to identify a particular X-ray image, of the next possible X-ray images, with a greatest predicted quality; utilize the machine learning model to update the six-degree of freedom pose based on the particular X-ray image and to generate an updated six-degree of freedom pose; and provide, to the C-arm CBCT device, data that identifies the updated six-degree of freedom pose, wherein the data that identifies the updated six-degree of freedom pose cause the C-arm CBCT device to: adjust to a new position based on the updated six-degree of freedom pose, and acquire a new X-ray image at the new position. ### Claim 9 (depends on 8) The device of claim 8, wherein the machine learning model includes a convolutional neural network model adapted to perform regression instead of classification. ### Claim 10 (depends on 8) The device of claim 8, wherein the updated six-degree of freedom pose includes an updated in-plane angle that is approximately five degrees greater that an in-plane angle, and wherein the updated six-degree of freedom pose includes an updated out-of-plane angle is in a range of approximately twenty-five degrees less than an out-of-plane angle to approximately twenty-five degrees greater than the out-of-plane angle. ### Claim 11 (depends on 8) The device of claim 8, wherein the one or more processors are further configured to: train the machine learning model based on a first dataset and a second dataset, and prior to receiving the X-ray image, wherein the first dataset is based on historical human chest CT images, and wherein the second dataset is based on a semi-anthropomorphic representation of a human chest. ### Claim 12 (depends on 8) The device of claim 8, wherein the one or more processors are further configured to: utilize a synthetic dataset to test the machine learning model prior to receiving the X-ray image; utilize a real dataset to train the machine learning model prior to receiving the X-ray image; and utilize batch normalization and data augmentation with the machine learning model prior to receiving the X-ray image. ### Claim 13 (depends on 8) The device of claim 8, wherein the data that identifies the updated six-degree of freedom pose reduces artifacts in future X-ray images captured by the C-arm CBCT device. ### Claim 14 (depends on 8) The device of claim 8, wherein the one or more processors are further configured to: identify a problem with a spinal fusion surgery based on one or more of the X-ray image or the next possible X-ray images; and provide a notification that identifies the problem. ### Claim 15 A non-transitory computer-readable medium storing a set of instructions, the set of instructions comprising: one or more instructions that, when executed by one or more processors of a device, cause the device to: train a machine learning model based on a first dataset and a second dataset and to generate a trained machine learning model; receive an X-ray image captured by a C-arm cone-beam computed tomography (CBCT) device at a particular position defined by a six-degree of freedom pose relative to an anatomy; process the X-ray image, with the trained machine learning model, to determine a predicted quality of next possible X-ray images provided by the C-arm CBCT device; utilize the trained machine learning model to identify a particular X-ray image, of the next possible X-ray images, with a greatest predicted quality; utilize the trained machine learning model to update the six-degree of freedom pose based on the particular X-ray image and to generate an updated six-degree of freedom pose; and provide, to the C-arm CBCT device, data that identifies the updated six-degree of freedom pose, wherein the data that identifies the updated six-degree of freedom pose cause the C-arm CBCT device to adjust to a new position based on the updated six-degree of freedom pose. ### Claim 16 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the six-degree of freedom pose includes an in-plane angle that is defined according to a circular trajectory where a source and a detector of the C-arm CBCT device move in a plane for an entire scan, and wherein the six-degree of freedom pose includes an out-of-plane angle that is associated with tilting a C-arm, of the C-arm CBCT device, relative to the plane. ### Claim 17 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the one or more instructions, that cause the device to process the X-ray image, with the trained machine learning model, to determine the predicted quality of the next possible X-ray images provided by the C-arm CBCT device, cause the device to: determine a local modulation transfer function based on the X-ray image; determine a local noise power spectrum based on the X-ray image; determine a task function describing properties of an object, provided in the X-ray image, with a quality in Fourier space; and regress a detectability index for the next possible X-ray images, to determine the predicted quality of the next possible X-ray images based on the local modulation transfer function, the local noise power spectrum, and the task function. ### Claim 18 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the updated six-degree of freedom pose includes an updated in-plane angle is approximately five degrees greater that an in-plane angle, and wherein the updated six-degree of freedom pose includes an updated out-of-plane angle is in a range of approximately twenty-five degrees less than an out-of-plane angle to approximately twenty-five degrees greater than the out-of-plane angle. ### Claim 19 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the first dataset is based on historical human chest CT images, and wherein the second dataset is based on a semi-anthropomorphic representation of a human chest. ### Claim 20 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the data that identifies the updated six-degree of freedom pose reduces artifacts in future X-ray images captured by the C-arm CBCT device.