# US 10,307,108 B2 · Pseudo-CT generation from MR data using a feature regression model

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 10,307,108 B2
- Document ID: US10307108B2
- Status: Granted
- Kind: B2
- Application number: 14/881,903
- Filed: 2015-10-13
- Published: 2017-04-13
- Granted: 2019-06-04
- Earliest claimed: 2015-10-13
- Assignees: Elekta Limited
- Applicants: Elekta, Inc.
- Inventors: Xiao Han
- Web page: https://patentgrep.com/patent/US10307108B2

## CPC

- A61B 5/7267: {involving training the classification device}
- A61B 5/0035: {adapted for acquisition of images from more than one imaging mode, e.g. combining MRI and optical tomography}
- A61B 5/055: involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
- A61B 5/7278: {Artificial waveform generation or derivation, e.g. synthesizing signals from measured signals}
- A61N 5/1039: {using functional images, e.g. PET or MRI}
- G01R 33/5608: {Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels (image data processing or generation, in general G06T)}
- G06T 5/60: using machine learning, e.g. neural networks
- G06T 5/90: Dynamic range modification of images or parts thereof
- G06T 12/30: Image post-processing, e.g. metal artefact correction
- G16H 30/20: for handling medical images, e.g. DICOM, HL7 or PACS
- G16H 30/40: for processing medical images, e.g. editing
- G16H 50/50: for simulation or modelling of medical disorders
- A61B 2090/3762: {using computed tomography systems [CT]}
- A61B 2576/00: Medical imaging apparatus involving image processing or analysis (A61B1/00009, A61B6/52 and A61B8/52 take precedence)
- G06T 2207/10081: Computed x-ray tomography [CT]
- G06T 2207/10088: Magnetic resonance imaging [MRI]
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- A61N: ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY (measurement of bioelectric currents A61B; surgical instruments, devices or methods for transferring non-mechanical forms of energy to or from the body A61B18/00; anaesthetic apparatus in general A61M; incandescent lamps H01K; infrared radiators for heating H05B)
- G01R: MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES (indicating correct tuning of resonant circuits H03J3/12)
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA

## Abstract

Systems and methods are provided for generating a pseudo-CT prediction model that can be used to generate pseudo-CT images. An exemplary system may include a processor configured to retrieve training data including at least one MR image and at least one CT image for each of a plurality of training subjects. For each training subject, the processor may extract a plurality of features from each image point of the at least one MR image, create a feature vector for each image point based on the extracted features, and extract a CT value from each image point of the at least one CT image. The processor may also generate the pseudo-CT prediction model based on the feature vectors and the CT values of the plurality of training subjects.

## Claims

### Claim 1

A system for generating a pseudo-CT image, comprising:
a database configured to store training data comprising Magnetic Resonance (MR) data and Computerized Tomography (CT) data of a plurality of training subjects, each training subject having at least one MR image and at least one CT image;
a computer processor circuitry device communicatively coupled to the database for accessing information stored in the database;
a storage device communicatively coupled to the computer processor circuitry device, the storage device storing instructions that, when executed by the computer processor circuitry device, configures the computer processor circuitry device to perform operations comprising:
accessing the database to retrieve the training data including the at least one MR image and the at least one CT image for each of the plurality of training subjects;
for each training subject:
extracting a plurality of features from each image point of the at least one MR image;
creating a feature vector for each image point based on the extracted features by storing each of the extracted features in a respective dimension of a plurality of dimensions of the feature vector;
extracting a CT value from each image point of the at least one CT image; and
aligning the at least one MR image and the at least one CT image for each of the plurality of training subjects;
training a machine learning technique to model a relationship between a CT value parameter H, a tissue value parameter, and a β parameter, the training comprising:
denoting the CT value parameter H to be the extracted CT value of each image point and the tissue value parameter to be the feature vector of each image point; and
computing the β parameter by performing a data fitting technique on the denoted CT value parameter H and the tissue value parameter, such that H is approximately equal to a mathematical operation of the β parameter and the tissue value parameter; and
generating, based on the trained machine learning technique, the pseudo-CT image from one or more features of a new MR image by denoting the tissue value parameter modeled by the machine learning technique to be the one or more features of the new MR image and computing the CT value parameter H modeled by the machine learning technique by applying the computed β parameter modeled by the machine learning technique to the tissue value parameter denoted to be the one or more features of the new MR image.

### Claim 2 (depends on 1)

The system of claim 1, wherein the training data include multiple CT images for at least one of the plurality of training subjects, and the operations comprise:
averaging CT values of corresponding image points of the multiple CT images.

### Claim 3 (depends on 1)

The system of claim 1, wherein the relationship modeled by the machine learning technique corresponds to:
H≈β1P+β2T1+β3T*2,
where H corresponds to the CT value parameter H, β1, β2, β3 correspond to the β parameter, and P, T1, T*2 correspond to the tissue value parameter.

### Claim 4 (depends on 1)

The system of claim 1, wherein the operations comprise:
reducing the dimensions of the feature vectors; and
generating the pseudo-CT image based on the feature vectors having reduced dimensions and the CT values of the plurality of training subjects.

### Claim 5 (depends on 4)

The system of claim 4, wherein reducing the dimensions of the feature vectors comprises:
reducing the dimensions of the feature vectors based on the MR images of the plurality of training subjects by reducing noise or redundancy in the feature vectors.

### Claim 6 (depends on 4)

The system of claim 4, wherein reducing the dimensions of the feature vectors comprises:
reducing the dimensions of the feature vectors based on the MR images and the CT images of the plurality of training subjects by removing signals irrelevant to the CT images in the feature vectors.

### Claim 7 (depends on 1)

The system of claim 1, wherein the extracted features include at least one of local pattern feature, landmark feature, or context feature.

### Claim 8

A system for generating a pseudo-CT image, comprising:
a computer processor circuitry device;
a storage device communicatively coupled to the computer processor circuitry device, the storage device storing instructions that, when executed by the computer processor circuitry device, configures the computer processor circuitry device to perform operations comprising:
receiving an Magnetic Resonance (MR) image of a patient;
wherein a machine learning technique is trained to model a relationship between a Computerized Tomography (CT) value parameter H, a tissue value parameter, and a β parameter by:
extracting a plurality of features from each image point of one or more MR images, each extracted feature corresponding to a different type of MR image information;
creating a feature vector for each image point based on the extracted features;
extracting a CT value for each image point of one or more MR images;
aligning the one or more MR images and a CT image corresponding to the extracted CT value for each of a plurality of training subjects;
denoting the CT value parameter H to be the extracted CT value and the tissue value parameter to be the feature vector of each image point; and
computing the β parameter by performing a data fitting technique on the denoted CT value parameter H and the tissue value parameter, such that H is approximately equal to a mathematical operation of the β parameter and the tissue value parameter; and
generating the pseudo-CT image based on the machine learning technique for all image points by denoting the tissue value parameter modeled by the machine learning technique to be one or more features of a new MR image and computing the CT value parameter H modeled by the machine learning technique by applying the computed β parameter modeled by the machine learning technique to the tissue value parameter denoted to be the one or more features of the new MR image.

### Claim 9 (depends on 8)

The system of claim 8, wherein the different types of MR image information include at least one numerical value and at least one categorical property.

### Claim 10 (depends on 9)

The system of claim 9, wherein the at least one numerical value includes at least one of an intensity value, energy, contrast, homogeneity, a correlation, a joint feature, or a coordinate location of the feature and the at least one categorical property includes at least one of a tissue type, a local pattern feature, a landmark feature, a context feature, or a structure label.

### Claim 11 (depends on 8)

The system of claim 8, wherein the operations comprise:
segmenting the MR image based on tissue class;
generating a classification map based on the segmentation;
extracting one or more image features from the classification map; and
combining the one or more image features extracted from the classification map with the features extracted from the MR image.

### Claim 12

A system for generating a pseudo-CT image for a patient, comprising:
a computer processor circuitry device;
a storage device communicatively coupled to the computer processor circuitry device, the storage device storing instructions that, when executed by the computer processor circuitry device, configures the computer processor circuitry device to perform operations comprising:
receiving an Magnetic Resonance (MR) image of the patient;
wherein a machine learning technique is trained to model a relationship between a Computerized Tomography (CT) value parameter H, a tissue value parameter, and a β parameter by:
extracting a plurality of features from each image point of at least one MR image;
creating a feature vector for each image point based on the extracted features by storing each of the extracted features in a respective dimension of a plurality of dimensions of the feature vector;
extracting a CT value from each image point of at least one CT image;
aligning the at least one MR image and at least one CT image for each of a plurality of training subjects;
denoting the CT value parameter H to be the extracted CT value of each image point and the tissue value parameter to be the feature vector of each image point; and
computing the β parameter by performing a data fitting technique on the denoted CT value parameter H and the tissue value parameter, such that H is approximately equal to a mathematical operation of the β parameter and the tissue value parameter; and
generating, based on the trained machine learning technique, the pseudo-CT image from one or more features of a new MR image by denoting the tissue value parameter modeled by the machine learning technique to be the one or more features of the new MR image and computing the CT value parameter H modeled by the machine learning technique by applying the computed β parameter modeled by the machine learning technique to the tissue value parameter denoted to be the one or more features of the new MR image.

### Claim 13 (depends on 12)

The system of claim 12, wherein the operations comprise:
receiving the MR image from an image acquisition device.

### Claim 14 (depends on 12)

The system of claim 12, wherein the operations comprise:
receiving the MR image from an MR image database.

### Claim 15 (depends on 12)

The system of claim 12, wherein the new image is a classification image.

### Claim 16 (depends on 12)

The system of claim 12, wherein the new image is an intermediate pseudo-CT image.

### Claim 17 (depends on 1)

The system of claim 1, wherein the operations comprise visually representing an anatomical feature of a subject based on the pseudo-CT image.

### Claim 18 (depends on 8)

The system of claim 8, wherein the operations comprise visually representing an anatomical feature of a subject based on the pseudo-CT image.

### Claim 19 (depends on 12)

The system of claim 12, wherein the machine learning model corresponds to:
H≈β1P+β2T1+β3T*2,
where H corresponds to the CT value parameter H, β1, β2, β3 correspond to the β parameter, and P, T1,T*2 correspond to the tissue value parameter.