# US 12,005,270 B2 · Systems and methods for accelerated online adaptive radiation therapy

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

- Number: US 12,005,270 B2
- Document ID: US12005270B2
- Status: Granted
- Kind: B2
- Application number: 17/256,339
- Filed: 2019-06-26
- Published: 2021-07-22
- Granted: 2024-06-11
- Earliest claimed: 2018-06-26
- Assignees: The Medical College of Wisconsin, Inc.
- Applicants: The Medical College of Wisconsin, Inc.
- Inventors: X. Allen Li; Ying Zhang; Sara Lim; Jingqiao Zhang; Ergun Ahunbay; Ranjeeta Thapa; Haidy Nasief
- Web page: https://patentgrep.com/patent/US12005270B2

## CPC

- A61N 5/1031: {using a specific method of dose optimization}
- G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems
- A61N 5/1038: {taking into account previously administered plans applied to the same patient, i.e. adaptive radiotherapy}
- G06N 20/00: Machine learning
- G06T 5/40: using histogram techniques
- G06T 7/11: Region-based segmentation
- G06T 7/12: Edge-based segmentation
- G06T 7/149: involving deformable models, e.g. active contour models
- G06T 7/174: involving the use of two or more images
- G06T 7/33: using feature-based methods
- G16H 10/60: for patient-specific data, e.g. for electronic patient records
- G16H 20/40: relating to mechanical, radiation or invasive therapies, e.g. surgery, laser therapy, dialysis or acupuncture
- G16H 30/40: for processing medical images, e.g. editing
- A61B 6/032: {Transmission computed tomography [CT]}
- A61B 6/5217: {extracting a diagnostic or physiological parameter from medical diagnostic data}
- G06T 2207/10081: Computed x-ray tomography [CT]
- G06T 2207/10088: Magnetic resonance imaging [MRI]
- G06T 2207/20081: Training; Learning
- G06T 2207/30008: Bone
- G06T 2207/30096: Tumor; Lesion
- 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)
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)

## Abstract

Systems and methods for accelerated online adaptive radiation therapy (“ART”) are described. The improvements to online ART are generally provided based on the use of textural analysis and machine learning algorithms implemented with a hardware processor and a memory. The described systems and methods enable more efficient and accurate online adaptive replanning (“OLAR”), which can also be implemented in clinically acceptable timeframes. For example, OLAR can be reduced from taking 10-30 minutes down to 5-10 minutes.

## Claims

### Claim 1

A method for validating an accuracy of radiation treatment plan contour data, the method comprising:
(a) accessing with a computer system, previously generated contour data comprising contours associated with a radiation treatment plan for a subject;
(b) accessing with the computer system, an image of the subject;
(c) generating with the computer system, inner shell data by eroding contours in the contour data by a first margin;
(d) generating with the computer system, outer shell data by expanding contours in the contour data by a second margin;
(e) computing with the computer system, image feature data from the image;
(f) generating with the computer system, an inner shell image feature map from the image feature data over contours in the inner shell data;
(g) generating with the computer system, an outer shell image feature map from the image feature data over contours in the outer shell data;
(h) validating the contour data by inputting the inner shell image feature map and the outer shell image feature map to a machine learning algorithm, generating output as labeled contour data, wherein the labeled contour data labels contour data that are validated as accurate contour data contour data that are not validated as inaccurate contour data;
(i) generating a binary mask for a given contour in the contour data by assigning a first binary value to pixels within the contour and assigning a second binary value to pixels outside the contour, and wherein the inner shell data are generated by eroding the binary mask by the first margin and the outer shell data are generated by dilating the binary mask by the second margin; and
(j) generating core region data from the binary mask, wherein the core region data are generated by:
generating a distance map by computing a distance metric between each pixel in the binary mask and a boundary pixel nearest to that pixel in the binary mask; and
defining the core region based on a percent of pixels having highest distance metric values in the distance map.

### Claim 2 (depends on 1)

The method as recited in claim 1, wherein the distance metric is a Euclidean distance metric.

### Claim 3 (depends on 1)

The method as recited in claim 1, wherein the percent is 20 percent.

### Claim 4 (depends on 1)

The method as recited in claim 1, wherein the percent is 30 percent.

### Claim 5 (depends on 1)

The method as recited in claim 1, further comprising generating with the computer system, a core region image feature map from the image feature data over the outer core region.

### Claim 6 (depends on 5)

The method as recited in claim 5, wherein the machine learning algorithm implements a decision tree model that takes as inputs the core region image feature data, the inner shell image feature data, and the outer shell image feature data.

### Claim 7 (depends on 6)

The method as recited in claim 6, wherein the decision tree model comprises:
a first level in which the core region image feature data are evaluated relative to ground truth data;
a second level in which the inner shell image feature data are evaluated relative to ground truth data; and
a third level in which the outer shell image feature data are evaluated relative to ground truth data.

### Claim 8 (depends on 7)

The method as recited in claim 7, wherein the first level of the decision tree model evaluates a selected set of image features in the core region image feature data to similar image features in the ground truth data.

### Claim 9 (depends on 8)

The method as recited in claim 8, wherein the set of image features is selected based on a hierarchical clustering of the image feature data contained in the core region.

### Claim 10 (depends on 7)

The method as recited in claim 7, wherein a contour is validated when its corresponding image feature data pass the first level, the second level, and the third level of the decision tree model.

### Claim 11 (depends on 1)

The method as recited in claim 1, wherein the inner shell image feature map and the outer shell image feature map are each normalized based on image feature data contained in the core region.

### Claim 12 (depends on 1)

The method as recited in claim 1, wherein the first margin and the second margin have a same margin value.

### Claim 13 (depends on 12)

The method as recited in claim 12, wherein the margin value is 4 mm.

### Claim 14 (depends on 1)

The method as recited in claim 1, wherein the image feature data comprise intensity histogram-based image features.

### Claim 15 (depends on 14)

The method as recited in claim 14, wherein the intensity histogram-based image features comprise at least one of mean, standard deviation, range, skewness, and kurtosis.

### Claim 16 (depends on 1)

The method as recited in claim 1, wherein the image feature data comprise texture features computed based on a gray-level co-occurrence matrix.

### Claim 17 (depends on 1)

The method as recited in claim 1, wherein the image feature data comprise texture features computed based on a gray-level run-length matrix.

### Claim 18 (depends on 1)

The method as recited in claim 1, wherein the machine learning algorithm implements a recursive random forest classification to select a set of image features against which the inner shell image feature map and the outer shell image feature map are compared to validate the contour data.

### Claim 19 (depends on 1)

The method as recited in claim 1, further comprising:
inputting the inaccurate contour data in the labeled contour data to a contour correction algorithm, generating output as corrected contour data.

### Claim 20 (depends on 19)

The method as recited in claim 19, wherein the corrected contour data are further validated by repeating steps (c)-(h) using the corrected contour data as input contour data.

### Claim 21 (depends on 19)

The method as recited in claim 19, wherein the contour correction algorithm comprises:
computing with the computer system, second image feature data for a region-of-interest (ROI) in the image, wherein the ROI is defined by a contour in the inaccurate contour data;
generating with the computer system, an image feature map from the second image feature data, wherein the image feature map depicts image feature values in the ROI;
generating corrected contour data by inputting the image feature map and the inaccurate contour data to an active contour algorithm, generating output as the corrected contour data;
wherein the image feature map is used by the active contour algorithm as an external force that guides an active curve to find a corrected boundary for the inaccurate contour data.

### Claim 22

A method for validating an accuracy of radiation treatment plan contour data, the method comprising:
(a) accessing with a computer system, previously generated contour data comprising contours associated with a radiation treatment plan for a subject;
(b) accessing with the computer system, an image of the subject;
(c) generating with the computer system, inner shell data by eroding contours in the contour data by a first margin;
(d) generating with the computer system, outer shell data by expanding contours in the contour data by a second margin, wherein the first margin and the second margin have a same margin value;
(e) computing with the computer system, image feature data from the image;
(f) generating with the computer system, an inner shell image feature map from the image feature data over contours in the inner shell data;
(g) generating with the computer system, an outer shell image feature map from the image feature data over contours in the outer shell data; and
(h) validating the contour data by inputting the inner shell image feature map and the outer shell image feature map to a machine learning algorithm, generating output as labeled contour data, wherein the labeled contour data labels contour data that are validated as accurate contour data contour data that are not validated as inaccurate contour data.

### Claim 23 (depends on 22)

The method as recited in claim 22, wherein the margin value is 4 mm.

### Claim 24

A method for validating an accuracy of radiation treatment plan contour data, the method comprising:
(a) accessing with a computer system, previously generated contour data comprising contours associated with a radiation treatment plan for a subject;
(b) accessing with the computer system, an image of the subject;
(c) generating with the computer system, inner shell data by eroding contours in the contour data by a first margin;
(d) generating with the computer system, outer shell data by expanding contours in the contour data by a second margin;
(e) computing with the computer system, image feature data from the image;
(f) generating with the computer system, an inner shell image feature map from the image feature data over contours in the inner shell data;
(g) generating with the computer system, an outer shell image feature map from the image feature data over contours in the outer shell data; and
(h) validating the contour data by inputting the inner shell image feature map and the outer shell image feature map to a machine learning algorithm, generating output as labeled contour data, wherein the labeled contour data labels contour data that are validated as accurate contour data contour data that are not validated as inaccurate contour data;
wherein the machine learning algorithm implements a recursive random forest classification to select a set of image features against which the inner shell image feature map and the outer shell image feature map are compared to validate the contour data.

### Claim 25

A method for validating an accuracy of radiation treatment plan contour data, the method comprising:
(a) accessing with a computer system, previously generated contour data comprising contours associated with a radiation treatment plan for a subject;
(b) accessing with the computer system, an image of the subject;
(c) generating with the computer system, inner shell data by eroding contours in the contour data by a first margin;
(d) generating with the computer system, outer shell data by expanding contours in the contour data by a second margin;
(e) computing with the computer system, image feature data from the image;
(f) generating with the computer system, an inner shell image feature map from the image feature data over contours in the inner shell data;
(g) generating with the computer system, an outer shell image feature map from the image feature data over contours in the outer shell data;
(h) validating the contour data by inputting the inner shell image feature map and the outer shell image feature map to a machine learning algorithm, generating output as labeled contour data, wherein the labeled contour data labels contour data that are validated as accurate contour data contour data that are not validated as inaccurate contour data; and
(i) inputting the inaccurate contour data in the labeled contour data to a contour correction algorithm, generating output as corrected contour data.

### Claim 26 (depends on 25)

The method as recited in claim 25, wherein the corrected contour data are further validated by repeating steps (c)-(h) using the corrected contour data as input contour data.

### Claim 27 (depends on 25)

The method as recited in claim 25, wherein the contour correction algorithm comprises:
computing with the computer system, second image feature data for a region-of-interest (ROI) in the image, wherein the ROI is defined by a contour in the inaccurate contour data;
generating with the computer system, an image feature map from the second image feature data, wherein the image feature map depicts image feature values in the ROI;
generating corrected contour data by inputting the image feature map and the inaccurate contour data to an active contour algorithm, generating output as the corrected contour data;
wherein the image feature map is used by the active contour algorithm as an external force that guides an active curve to find a corrected boundary for the inaccurate contour data.