# US 2022/0284687 A1 · Systems and methods for image segmentation

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

- Number: US 2022/0284687 A1
- Document ID: US20220284687A1
- Status: Published application
- Kind: A1
- Application number: 17/664,422
- Filed: 2022-05-22
- Published: 2022-09-08
- Earliest claimed: 2018-12-17
- Assignees: Shanghai United Imaging Intelligence Co., Ltd.
- Applicants: Shanghai United Imaging Intelligence Co., Ltd.
- Inventors: Miaofei HAN; Yaozong GAO; Yu ZHANG; Yiqiang ZHAN
- Web page: https://patentgrep.com/patent/US20220284687A1

## CPC

- G06T 7/11: Region-based segmentation
- A61B 5/1073: {Measuring volume, e.g. of limbs}
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/0495: Quantised networks; Sparse networks; Compressed networks
- G06N 3/09: Supervised learning
- G06T 7/0012: {Biomedical image inspection}
- G06T 7/149: involving deformable models, e.g. active contour models
- G06T 7/62: of area, perimeter, diameter or volume
- G06V 30/2504: {Coarse or fine approaches, e.g. resolution of ambiguities or multiscale approaches}
- A61B 2576/02: specially adapted for a particular organ or body part
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30084: Kidney; Renal
- G06V 2201/031: of internal organs
- G06V 2201/032: of protuberances, polyps nodules, etc.
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A system for image segmentation is provided. The system may obtain a target image including an ROI, and segment a preliminary region representative of the ROI from the target image using a first ROI segmentation model corresponding to a first image resolution. The system may segment a target region representative of the ROI from the preliminary region using a second ROI segmentation model corresponding to a second image resolution. At least one model of the first and second ROI segmentation models may at least include a first convolutional layer and a second convolutional layer downstream to the first convolutional layer. A count of input channels of the first convolutional layer may be greater than a count of output channels of the first convolutional layer, and a count of input channels of the second convolutional layer may be smaller than a count of output channels of the second convolutional layer.

## Claims

### Claim 1

A system for image segmentation, comprising:
at least one storage device including a set of instructions; and
at least one processor configured to communicate with the at least one storage device, wherein when executing the set of instructions, the at least one processor is configured to direct the system to perform operations including:
obtaining a target image including a region of interest (ROI); and
segmenting a target region representative of the ROI using a ROI segmentation model, wherein
the ROI segmentation model includes a plurality of convolutional layers, the plurality of convolutional layers at least including a first convolutional layer and a second convolutional layer downstream to the first convolutional layer, a count of input channels of the first convolutional layer is greater than a count of output channels of the first convolutional layer.

### Claim 2 (depends on 1)

The system of claim 1, wherein a count of input channels of the second convolutional layer is smaller than a count of output channels of the second convolutional layer, and the count of input channels of the first convolutional layer is equal to the count of output channels of the second convolutional layer.

### Claim 3 (depends on 1)

The system of claim 1, wherein the ROI segmentation model further includes a third convolutional layer between the first convolutional layer and the second convolutional layer, and a count of output channels of the third convolutional layer is equal to a count of input channels of the third convolutional layer.

### Claim 4 (depends on 1)

The system of claim 1, wherein the ROI segmentation model further includes at least one of a batch normalization layer or an activation layer between the first convolutional layer and the second convolutional layer.

### Claim 5 (depends on 1)

The system of claim 1, wherein the ROI segmentation model includes a first ROI segmentation model and a second ROI segmentation model, and the segmenting a target region representative of the ROI using a ROI segmentation model comprises:
segmenting, from the target image, a preliminary region representative of the ROI using the first ROI segmentation model; and
segmenting, from the preliminary region, the target region representative of the ROI using the second ROI segmentation model.

### Claim 6 (depends on 5)

The system of claim 5, wherein the first ROI segmentation model corresponds to a first image resolution, the second ROI segmentation model corresponding to a second image resolution, and the first image resolution is lower than the second image resolution.

### Claim 7 (depends on 6)

The system of claim 6, wherein the segmenting, from the target image, a preliminary region representative of the ROI using the first ROI segmentation model comprises:
preprocessing the target image to generate a preprocessed target image;
segmenting, from the preprocessed target image, the ROI by applying the first ROI segmentation model to the preprocessed target image; and
segmenting, from the target image, the preliminary region based on the segmented ROI in the preprocessed target image.

### Claim 8 (depends on 7)

The system of claim 7, wherein the preprocessing the target image to generate a preprocessed target image comprises:
generating a resampled target image having the first image resolution by resampling the target image; and
generating the preprocessed target image by normalizing the resampled target image.

### Claim 9 (depends on 5)

The system of claim 5, wherein the segmenting, from the preliminary region, the target region representative of the ROI using the second ROI segmentation model comprises:
preprocessing the preliminary region to generate a preprocessed preliminary region;
segmenting, from the preprocessed preliminary region, the ROI by applying the second ROI segmentation model to the preprocessed preliminary region; and
generating the target region by resampling the segmented ROI in the preprocessed preliminary region, the target region and the target image having a same image resolution.

### Claim 10 (depends on 9)

The system of claim 9, wherein the preprocessing the preliminary region to generate a preprocessed preliminary region comprises:
generating a resampled preliminary region having the second image resolution by resampling the preliminary region; and
generating the preprocessed preliminary region by normalizing the resampled preliminary region.

### Claim 11 (depends on 1)

The system of claim 1, wherein the at least one processor is further configured to direct the system to perform the operations including:
determining, in the at least one storage device, a first memory space and a second memory space based on the plurality of convolutional layers, wherein
the plurality of convolutional layers include a fourth convolutional layer and a fifth convolutional layer adjacent to each other, and
during the application of the ROI segmentation model,
input data of the fourth convolutional layer are stored in the first memory space,
output data of the fourth convolutional layer are stored in the second memory,
input data of the fifth convolutional layer are stored in the second memory space, and
output data of the fifth convolutional layer are stored in the first memory.

### Claim 12 (depends on 11)

The system of claim 11, wherein the ROI segmentation model includes at least one skip-connection, and data relating to the at least one skip-connection is stored in at least one of the first memory space or the second memory space.

### Claim 13 (depends on 1)

The system of claim 1, wherein the at least one processor is further configured to direct the system to perform the operations including:
determining, based on the target region, a parameter value indicative of a physiological condition of the ROI; and
classifying a functionality of the ROI based on the parameter value of the ROI.

### Claim 14

A system for training a region of interest (ROI) segmentation model corresponding to a target image resolution, comprising:
at least one storage device storing a set of instructions; and
at least one processor configured to communicate with the at least one storage device, wherein when executing the set of instructions, the at least one processor is configured to direct the system to perform operations including:
obtaining a preliminary model including a plurality of convolutional layers; and
generating the ROI segmentation model corresponding to the target image resolution by training the preliminary model, wherein
the plurality of convolutional layers include at least a first convolutional layer and a second convolutional layer downstream to the first convolutional layer, and
a count of input channels of the first convolutional layer is greater than a count of output channels of the first convolutional layer.

### Claim 15 (depends on 14)

The system of claim 14, wherein the generating the ROI segmentation model corresponding to the target image resolution by training the preliminary model comprises:
obtaining at least one training image, each of the at least one training image including an annotated ROI and having the target image resolution; and
generating the ROI segmentation model corresponding to the target image resolution by training the preliminary model using the at least one training image.

### Claim 16 (depends on 15)

The system of claim 15, wherein the obtaining at least one training image comprises:
obtaining at least one image including the annotated ROI and having an image resolution different from the target image resolution;
generating at least one resampled image having the target image resolution by resampling the at least one image; and
generating the at least one training image by normalizing the at least one resampled image.

### Claim 17 (depends on 14)

The system of claim 14, wherein a count of input channels of the second convolutional layer is smaller than a count of output channels of the second convolutional layer, and the count of input channels of the first convolutional layer is equal to the count of the output channels of the second convolutional layer.

### Claim 18 (depends on 14)

The system of claim 14, wherein the plurality of convolutional layers further includes a third convolutional layer between the first convolutional layer and the second convolutional layer, and a count of output channels of the third convolutional layer is equal to a count of input channels of the third convolutional layer.

### Claim 19

A system for image segmentation, comprising:
at least one storage device including a set of instructions; and
at least one processor configured to communicate with the at least one storage device, wherein when executing the set of instructions, the at least one processor is configured to direct the system to perform operations including:
obtaining a target image including a region of interest (ROI);
segmenting, from the target image, a preliminary region representative of the ROI using a ROI segmentation model; and
segmenting, from the preliminary region, a target region representative of the ROI, wherein
the ROI segmentation model includes a plurality of convolutional layers, the plurality of convolutional layers at least including a first convolutional layer and a second convolutional layer downstream to the first convolutional layer, a count of input channels of the first convolutional layer is greater than a count of output channels of the first convolutional layer.

### Claim 20 (depends on 19)

The method of claim 19, wherein a count of input channels of the second convolutional layer is smaller than a count of output channels of the second convolutional layer, and the count of input channels of the first convolutional layer is equal to the count of the output channels of the second convolutional layer.