CN117809030A - Breast cancer CT image identification and segmentation method based on artificial neural network - Google Patents
Breast cancer CT image identification and segmentation method based on artificial neural network Download PDFInfo
- Publication number
- CN117809030A CN117809030A CN202311776787.6A CN202311776787A CN117809030A CN 117809030 A CN117809030 A CN 117809030A CN 202311776787 A CN202311776787 A CN 202311776787A CN 117809030 A CN117809030 A CN 117809030A
- Authority
- CN
- China
- Prior art keywords
- segmentation
- breast cancer
- image
- neural network
- artificial neural
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000011218 segmentation Effects 0.000 title claims abstract description 69
- 208000026310 Breast neoplasm Diseases 0.000 title claims abstract description 42
- 206010006187 Breast cancer Diseases 0.000 title claims abstract description 39
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 15
- 230000003902 lesion Effects 0.000 claims abstract description 14
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 abstract description 2
- 238000002591 computed tomography Methods 0.000 description 28
- 238000003745 diagnosis Methods 0.000 description 7
- 206010028980 Neoplasm Diseases 0.000 description 6
- 201000011510 cancer Diseases 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 238000011282 treatment Methods 0.000 description 6
- 210000001165 lymph node Anatomy 0.000 description 5
- 230000004913 activation Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000002059 diagnostic imaging Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 206010006272 Breast mass Diseases 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/03—Recognition of patterns in medical or anatomical images
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to the technical field of medical image processing, and particularly relates to a method for identifying and dividing a lesion area in a breast cancer CT image. A breast cancer CT image identification and segmentation method based on an artificial neural network comprises the following steps: s1: CT images of breast cancer patients are acquired and preprocessed; s2, inputting the preprocessed CT image into a pre-trained SAM model to obtain a pre-segmentation result; and S3, splicing the pre-segmentation result with the images before segmentation, and inputting the spliced images as a data set into a pre-trained U-Net segmentation model to obtain a mask map of the lesion area. According to the method, the mask map of the lesion area in the CT image of the breast cancer patient is obtained by adopting the artificial neural network model, and the method has the advantages of smooth area segmentation, high segmentation accuracy, high segmentation speed and the like.
Description
Technical Field
The invention belongs to the technical field of medical image processing, and particularly relates to a method for identifying and dividing a lesion area in a breast cancer CT image.
Background
Breast cancer is one of the most common cancers today. It was counted that 230 ten thousand women diagnosed with breast cancer worldwide in 2020, accounting for about 1/5 of all cancer diagnosed cases, 68.5 of which die. By the year 2020, over the last five years, a total of 780 ten thousand women were diagnosed with breast cancer in the world. The latest cancer report in china in 2019 was published: breast cancer is the first cancer onset in women in China, and is one of the most serious malignant tumors affecting the health of women. Therefore, it is particularly important to study diagnosis and treatment of breast cancer.
Over the past several decades, various medical imaging techniques have been widely used for early detection, diagnosis and treatment of cancer, and Computed Tomography (CT) is one of the most important medical imaging detection techniques for breast cancer diagnosis. In order to diagnose whether a patient has cancer and to carry out subsequent treatments, it is necessary to accurately delineate the lesion area in the medical image. Clinically, interpretation and delineation of medical images is mainly done by experts such as imaging doctors, but the process is very time consuming and highly dependent on the experience of the doctor.
In recent years, with the development of artificial intelligence technology, deep learning algorithms play an increasingly important role in the field of intelligent medical treatment, and domestic and foreign experts develop a great deal of research on auxiliary diagnosis of breast cancer based on artificial neural networks. The computer-aided technology is used for replacing manual reading of medical images, identification and marking of lesion areas are carried out, diagnosis and treatment of breast cancer can be assisted by doctors, and the method has important significance.
Therefore, the depth fusion of the artificial neural network and the breast cancer CT image is a direction worthy of research in the future.
Disclosure of Invention
In order to overcome the defects and shortcomings of the prior art, the invention provides a breast cancer CT image identification and segmentation method based on an artificial neural network, which realizes identification and segmentation of lesion areas (breast tumors and axillary lymph nodes) in the breast cancer CT image. The method has the advantages of smooth region segmentation, high segmentation accuracy, high segmentation speed and the like.
In order to solve the technical problems of the invention, the invention adopts the following technical scheme: a breast cancer CT image identification and segmentation method based on an artificial neural network comprises the following steps:
s1, acquiring CT images of breast cancer patients, and preprocessing;
s2, inputting the preprocessed CT image into a pre-trained SAM model to obtain a pre-segmentation result;
and S3, splicing the pre-segmentation result with the images before segmentation, and inputting the spliced images as a data set into a pre-trained U-Net segmentation model to obtain a mask map of the lesion area.
Further, in the step S1, the specific method of the pretreatment is as follows: and converting the breast cancer CT image of the patient from an NRRD format to a PNG format, and converting the PNG format to an RGB three-channel NPY format.
Further, the specific method of S3 is as follows: and splicing the pre-segmentation result with the CT image in the NPY format of the RGB three channels to serve as a fourth channel except R, G, B.
Specifically, the size of the original three-channel NPY format CT image is nxwxhx3, the size of the spliced image is nxwxhx4, where N represents the number of images, W represents the width of the image, H represents the height of the image, 3 represents the depth of each pixel, i.e., R, G, B three channels, and 4 represents the depth of each pixel, i.e., R, G, B and four channels as a result of pre-segmentation. The spliced image contains original image information and pre-segmentation information. And finally, inputting the spliced data into an accurate segmentation network to obtain a segmentation Mask of the background and the lesion area.
The invention uses the artificial neural network to realize the identification and segmentation of breast tumors, axillary lymph nodes and background areas in the CT image of breast cancer, and has the advantages of smooth segmentation area, high segmentation accuracy, high segmentation speed and the like. The neural network comprises a pre-segmentation part and an accurate segmentation part, global features can be captured in a preliminary stage by using SAM model pre-segmentation, and the pre-segmentation result and the original image are spliced to provide rich auxiliary information for the follow-up accurate segmentation model, so that the capturing capability of the accurate segmentation model on target boundaries, details and contexts is improved, and the accuracy of the accurate segmentation result is improved. The mask map obtained by final segmentation can be used as an important reference for assisting doctors in diagnosis and developing subsequent treatment, so that the workload and working time of image doctors are greatly reduced, marking deviation caused by insufficient experience of partial doctors can be reduced to a certain extent, and the accuracy of diagnosis is improved.
Drawings
FIG. 1 is a flow chart of a breast cancer CT image identification and segmentation method based on an artificial neural network;
fig. 2 is a schematic diagram of an algorithm structure for breast cancer CT image recognition and segmentation.
Detailed Description
In order that the invention may be readily understood, a more particular description thereof will be rendered by reference to specific embodiments that are illustrated in the appended drawings. Preferred embodiments of the present invention are shown in the drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. These embodiments are provided so that this disclosure will be thorough and complete.
The embodiment provides a breast cancer CT image recognition and segmentation method based on an artificial neural network, and the flow is shown in fig. 1, and the method comprises the following specific steps:
1. data preprocessing
1. And (3) acquiring a breast cancer CT image in an NRRD format, and delineating and marking two lesion areas, namely a breast tumor and an axillary lymph node by using a Slicer software.
2. Converting the breast cancer CT image in the NRRD format into a PNG format, converting the PNG format into an NPY format of RGB three channels, and respectively storing the NPY format. The size of the three-channel NPY file is NxW xH x 3, wherein N is the number of images, H is the height of the images, W is the width of the images, and 3 is the depth (channel number) of the images.
3. And converting the NRRD file marked with the lesion area into an NPY file serving as a target (label) of the training model.
2. Breast cancer CT image recognition and segmentation model construction and training
1. And inputting the PNG-format breast cancer CT image into a pre-trained SAM model for segmentation, and storing the obtained segmentation result as an NPY file as a pre-segmentation result. The pre-segmentation result is a segmentation result diagram with the same size as the original image, wherein each pixel is allocated a predicted class label. The pre-segmentation result has a size of nxwxhx1, where 1 represents one predicted class label to which each pixel is assigned.
2. And splicing the pre-segmentation result with the CT image in the NPY format of the RGB three channels to serve as a fourth channel except R, G, B. Specifically, the CT image of the original three-channel NPY format with a size of nxw×h×3 is spliced with the pre-segmentation result with a size of nxw×h×1, and the spliced image has a size of nxw×h×4. And taking the NPY file obtained after splicing as a data set input to the U-Net.
3. The U-Net is composed of two parts, namely an encoder and a decoder. The encoder is composed of four downsampling modules, each downsampling module is composed of two 3×3 convolution layers and a 2×2 pooling layer; the decoder consists of four upsampling modules and a 1 x 1 convolutional layer, each upsampling module consisting of a bilinear interpolation, a feature concatenation and two 3 x 3 convolutional layers. The convolutional layer in each downsampling and upsampling module is followed by a Relu activation function:
Relu＝max(0，x)
where x is the data input to the activation function.
4. And inputting the NPY file obtained after the splicing to the U-Net for training, and verifying and testing. If the requirements are not met, the training parameters are adjusted to continue training the model, and verification and test are carried out until the breast cancer CT image identification and segmentation model meeting the requirements is obtained. The loss function L in the model training process adopts a cross entropy loss function, and the calculation method of the cross entropy loss function is as follows:
where M is the number of categories, in this example, M equals 3 (breast mass, axillary lymph node, background area); y is ic As a sign function (0 or 1), y is the true class of sample i is equal to c ic Taking 1, otherwise taking 0; p is p ic The predicted probability that sample i belongs to category c is observed.
3. Identification and segmentation of breast cancer CT images
1. And converting the acquired CT image of the breast cancer patient into PNG format and RGB three-channel NPY format.
2. And inputting the CT image in the PNG format into a SAM model for segmentation, and storing the obtained segmentation result as an NPY file as a pre-segmentation result.
3. And taking the pre-segmentation result as a fourth channel except R, G, B, splicing the pre-segmentation NPY file and the RGB NPY file, and inputting the pre-segmentation NPY file into a pre-trained U-Net network model to obtain segmentation Mask sketching results of breast tumors, axillary lymph nodes and background areas.
4. Thus, the identification and segmentation of the lesion area of the CT image of the breast cancer are completed.
Claims (3)
1. The breast cancer CT image identification and segmentation method based on the artificial neural network is characterized by comprising the following steps of:
s1, acquiring CT images of breast cancer patients, and preprocessing;
s2, inputting the preprocessed CT image into a pre-trained SAM model to obtain a pre-segmentation result;
and S3, splicing the pre-segmentation result with the images before segmentation, and inputting the spliced images as a data set into a pre-trained U-Net segmentation model to obtain a mask map of the lesion area.
2. The breast cancer CT image recognition and segmentation method based on the artificial neural network according to claim 1, wherein the method comprises the following steps: in the step S1, the specific method for preprocessing is as follows: and converting the breast cancer CT image of the patient from an NRRD format to a PNG format, and converting the PNG format to an RGB three-channel NPY format.
3. The breast cancer CT image recognition and segmentation method based on artificial neural network according to claim 1 or 2, characterized in that: the specific method of the S3 is as follows: and splicing the pre-segmentation result with the CT image in the NPY format of the RGB three channels to serve as a fourth channel except R, G, B, and inputting the spliced data into an accurate segmentation network to obtain segmentation masks of the background and the lesion area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311776787.6A CN117809030A (en) | 2023-12-22 | 2023-12-22 | Breast cancer CT image identification and segmentation method based on artificial neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311776787.6A CN117809030A (en) | 2023-12-22 | 2023-12-22 | Breast cancer CT image identification and segmentation method based on artificial neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117809030A true CN117809030A (en) | 2024-04-02 |
Family
ID=90424454
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311776787.6A Pending CN117809030A (en) | 2023-12-22 | 2023-12-22 | Breast cancer CT image identification and segmentation method based on artificial neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117809030A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118587443A (en) * | 2024-08-07 | 2024-09-03 | 之江实验室 | Image segmentation method and device based on self-training and priori guidance |
-
2023
- 2023-12-22 CN CN202311776787.6A patent/CN117809030A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118587443A (en) * | 2024-08-07 | 2024-09-03 | 之江实验室 | Image segmentation method and device based on self-training and priori guidance |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113674253B (en) | Automatic segmentation method for rectal cancer CT image based on U-transducer | |
| CN111091589A (en) | Ultrasonic and nuclear magnetic image registration method and device based on multi-scale supervised learning | |
| CN111882560B (en) | Lung parenchyma CT image segmentation method based on weighted full convolution neural network | |
| CN113034505B (en) | Glandular cell image segmentation method and glandular cell image segmentation device based on edge perception network | |
| CN113223005B (en) | Thyroid nodule automatic segmentation and grading intelligent system | |
| CN112365464A (en) | GAN-based medical image lesion area weak supervision positioning method | |
| CN114092439A (en) | Multi-organ instance segmentation method and system | |
| CN110689525A (en) | Method and device for recognizing lymph nodes based on neural network | |
| CN112884788B (en) | Cup optic disk segmentation method and imaging method based on rich context network | |
| CN110766670A (en) | Mammary gland molybdenum target image tumor localization algorithm based on deep convolutional neural network | |
| CN115512110A (en) | Medical image tumor segmentation method related to cross-modal attention mechanism | |
| CN114202545A (en) | UNet + + based low-grade glioma image segmentation method | |
| CN109214388B (en) | Tumor segmentation method and device based on personalized fusion network | |
| CN117809030A (en) | Breast cancer CT image identification and segmentation method based on artificial neural network | |
| CN112927237A (en) | Honeycomb lung focus segmentation method based on improved SCB-Unet network | |
| CN116524315A (en) | Mask R-CNN-based lung cancer pathological tissue section identification and segmentation method | |
| CN110992309A (en) | Fundus image segmentation method based on deep information transfer network | |
| CN114581474A (en) | Automatic clinical target area delineation method based on cervical cancer CT image | |
| CN111210398A (en) | White blood cell recognition system based on multi-scale pooling | |
| CN118512278A (en) | AI modeling method and device used before tooth 3D printing | |
| CN114612478B (en) | Female pelvic cavity MRI automatic sketching system based on deep learning | |
| CN113487579B (en) | Multi-mode migration method for automatically sketching model | |
| Nasim et al. | Review on multimodality of different medical image fusion techniques | |
| CN115526898A (en) | Medical image segmentation method | |
| CN114882282A (en) | Neural network prediction method for colorectal cancer treatment effect based on MRI and CT images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |