CN113902746A - Method and system for extracting blood vessel guide wire in medical image, electronic device and medium - Google Patents
Method and system for extracting blood vessel guide wire in medical image, electronic device and medium Download PDFInfo
- Publication number
- CN113902746A CN113902746A CN202111513444.1A CN202111513444A CN113902746A CN 113902746 A CN113902746 A CN 113902746A CN 202111513444 A CN202111513444 A CN 202111513444A CN 113902746 A CN113902746 A CN 113902746A
- Authority
- CN
- China
- Prior art keywords
- blood vessel
- image
- guide wire
- medical image
- segmentation
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/48—Diagnostic techniques
- A61B6/481—Diagnostic techniques involving the use of contrast agents
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/50—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications
- A61B6/504—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment specially adapted for specific body parts; specially adapted for specific clinical applications for diagnosis of blood vessels, e.g. by angiography
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
- A61M25/00—Catheters; Hollow probes
- A61M25/01—Introducing, guiding, advancing, emplacing or holding catheters
- A61M25/09—Guide wires
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Radiology & Medical Imaging (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- Heart & Thoracic Surgery (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- High Energy & Nuclear Physics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Optics & Photonics (AREA)
- Surgery (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Pathology (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Dentistry (AREA)
- Vascular Medicine (AREA)
- Pulmonology (AREA)
- Anesthesiology (AREA)
- Hematology (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention discloses a method, a system, electronic equipment and a medium for extracting a blood vessel guide wire in a medical image, which relate to the field of artificial intelligence and medical image processing, and the method comprises the steps of inputting a current frame medical image into a trained blood vessel guide wire segmentation model, and acquiring blood vessel position information and guide wire position information in the current frame medical image; the trained blood vessel guide wire segmentation model is obtained based on a deep learning algorithm and training of multiple groups of historical images; the historical images comprise historical medical images, and blood vessel segmentation annotation images and guide wire segmentation annotation images which correspond to the historical medical images; the vessel segmentation annotation image and the guide wire segmentation annotation image are determined after morphological operation is carried out on the vessel subtraction image, and the vessel subtraction image is determined by carrying out difference determination on the first frame of historical medical image and other frames of historical medical images. The invention further improves the segmentation precision of the medical image by improving the image annotation efficiency and the annotation precision.
Description
Technical Field
The invention relates to the field of artificial intelligence and medical image processing, in particular to a method, a system, electronic equipment and a medium for extracting a blood vessel guide wire in a medical image.
Background
In recent years, cerebrovascular diseases seriously threaten the life health of human beings, and cerebrovascular intervention surgery is gradually the best treatment mode. At present, some technologies assisted by robots and computers are closely combined with clinical practice, and have important clinical value and practical significance for doctors. As a necessary link of the robot-assisted minimally invasive surgery, the position of the guide wire and the position of the blood vessel can greatly reduce the burden of medical workers and provide more reliable auxiliary information. Therefore, real-time guidewire morphology segmentation and vessel segmentation are essential.
For the blood vessel segmentation, the traditional methods, such as a threshold method, a region growing-based method, a tracking-based method, a clustering-based method, a model-based segmentation method, and the like, have a good segmentation effect on simple blood vessel images, and are not ideal for the segmentation effect of blood vessel images with high complexity.
For guidewire segmentation, conventional methods, such as histogram construction based on pixel values, histogram identification based, statistical-based methods, etc., are not suitable for real-time dynamic surgical environments, and such methods have poor universality and robustness, especially in complex or noisy environments. In contrast, segmentation methods based on deep learning have higher accuracy than other conventional methods.
In recent research, the U-net network structure shows the development potential, becomes a mainstream framework, and shows very good performance in the field of medical image segmentation. However, due to the complex structure of the cerebral vessels, in order to obtain well-labeled training data, a professional doctor needs to spend a lot of time and energy, and the professional knowledge and experience are heavily relied on, so that a lot of subjective differences exist. Therefore, at present, a large amount of labeled data are obtained at high cost to develop deep learning research so as to segment the cerebrovascular image phenomenon, and meanwhile, a large amount of unlabeled data are not well applied, and the method does not meet the urgent requirement in practical clinical application.
Disclosure of Invention
The invention aims to provide a method, a system, electronic equipment and a medium for extracting a blood vessel guide wire in a medical image, which are used for further improving the segmentation precision of the medical image by improving the image labeling efficiency and the labeling precision.
In order to achieve the purpose, the invention provides the following scheme:
a method for extracting a vascular guide wire in a medical image comprises the following steps:
acquiring a current frame medical image to be processed; the current frame medical image is a scanned image acquired by special equipment, and the scanned image comprises a blood vessel which appears due to contrast in an interventional operation and a guide wire for guiding a stent to be placed into the blood vessel;
inputting the current frame medical image into a trained blood vessel guide wire segmentation model to acquire blood vessel position information and guide wire position information in the current frame medical image;
the trained blood vessel guide wire segmentation model is obtained based on a deep learning algorithm and training of multiple groups of historical images; the history image includes: the method comprises the following steps that a historical medical image, a blood vessel segmentation annotation image corresponding to the historical medical image and a guide wire segmentation annotation image corresponding to the historical medical image are obtained;
the blood vessel segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation, and the guide wire segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation; the vessel subtraction image is determined by differencing the first frame of historical medical image and the other frames of historical medical images.
Optionally, the training method of the vascular guidewire segmentation model specifically includes:
acquiring a plurality of groups of historical medical images, blood vessel segmentation annotation images corresponding to the historical medical images and guide wire segmentation annotation images corresponding to the historical medical images;
and inputting a plurality of groups of the historical medical images, the blood vessel segmentation annotation images corresponding to the historical medical images and the guide wire segmentation annotation images corresponding to the historical medical images into a blood vessel guide wire segmentation model to be trained, and performing iterative training on the blood vessel guide wire segmentation model to be trained to obtain the trained blood vessel guide wire segmentation model.
Optionally, the acquiring multiple sets of historical medical images, the blood vessel segmentation annotation image corresponding to the historical medical image, and the guide wire segmentation annotation image corresponding to the historical medical image specifically includes:
acquiring a plurality of groups of historical medical images; the total frame number of the historical medical images is N;
sequencing the historical medical images according to the time sequence to obtain a first frame of historical medical image;
performing one-to-one difference between the first frame of historical medical image and other frames of historical medical images to obtain N-1 frames of blood vessel subtraction images;
performing morphological operation on each frame of the blood vessel subtraction image to determine a blood vessel segmentation annotation image;
and performing morphological operation on each frame of the blood vessel subtraction image to determine a guide wire segmentation annotation image.
Optionally, the inputting a plurality of sets of the historical medical images, the blood vessel segmentation labeling images corresponding to the historical medical images, and the guide wire segmentation labeling images corresponding to the historical medical images into a blood vessel guide wire segmentation model to be trained, and performing iterative training on the blood vessel guide wire segmentation model to be trained to obtain the trained blood vessel guide wire segmentation model specifically includes:
preprocessing the historical medical image, the blood vessel segmentation annotation image corresponding to the historical medical image and the guide wire segmentation annotation image corresponding to the historical medical image; the preprocessing operation comprises size transformation, image data augmentation and gray level binarization processing;
inputting a plurality of groups of image data after preprocessing operation into a to-be-trained blood vessel guide wire segmentation model, and performing iterative training on the to-be-trained blood vessel guide wire segmentation model to obtain a trained blood vessel guide wire segmentation model;
the image data after the preprocessing operation comprises a historical medical image after the preprocessing operation, a blood vessel segmentation and annotation image after the preprocessing operation corresponding to the historical medical image after the preprocessing operation and a guide wire segmentation and annotation image after the preprocessing operation corresponding to the historical medical image after the preprocessing operation.
Optionally, the vessel guidewire segmentation model is a network model built by adopting a Python language and a tensrflow framework on the basis of a U-Net network architecture; the position of a dense connecting block in the blood vessel guide wire segmentation model is the same as the position of a convolution layer in the U-Net network architecture; the vascular guide wire segmentation model comprises four down-sampling processes and four up-sampling processes.
Optionally, the loss function of the vascular guidewire segmentation model is a binary cross entropy loss function.
Optionally, the current frame medical image is input into a trained blood vessel guidewire segmentation model to obtain blood vessel position information and guidewire position information in the current frame medical image, and the method specifically includes:
preprocessing the current frame medical image;
inputting the current medical image after the preprocessing operation into a trained blood vessel guide wire segmentation model to obtain a blood vessel segmentation annotation image and a guide wire segmentation annotation image corresponding to the current medical image after the preprocessing operation;
and processing the blood vessel segmentation annotation image and the guide wire segmentation annotation image corresponding to the current medical image after the preprocessing operation by adopting a morphological processing algorithm, and determining the blood vessel position information and the guide wire position information in the current medical image.
A vascular guidewire extraction system in medical images, comprising:
the current frame medical image acquisition module is used for acquiring a current frame medical image to be processed; the current frame medical image is a scanned image acquired by special equipment, and the scanned image comprises a blood vessel which appears due to contrast in an interventional operation and a guide wire for guiding a stent to be placed into the blood vessel;
the blood vessel position information and guide wire position information determining module is used for inputting the current frame medical image into a trained blood vessel guide wire segmentation model so as to acquire blood vessel position information and guide wire position information in the current frame medical image;
the trained blood vessel guide wire segmentation model is obtained based on a deep learning algorithm and training of multiple groups of historical images; the history image includes: the method comprises the following steps that a historical medical image, a blood vessel segmentation annotation image corresponding to the historical medical image and a guide wire segmentation annotation image corresponding to the historical medical image are obtained; the blood vessel segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation, and the guide wire segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation; the vessel subtraction image is determined by differencing the first frame of historical medical image and the other frames of historical medical images.
An electronic device, comprising:
one or more processors;
storage means for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement a method of vascular guidewire extraction in a medical image.
A storage medium containing computer executable instructions for performing a method of vascular guidewire extraction in medical images when executed by a computer processor.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
1. compared with the traditional segmentation, namely only the roadmap is obtained firstly and then the interventional operation is carried out according to the roadmap, the method can accurately and quickly segment the guide wire and the blood vessel in the medical image simultaneously, can realize the real-time extraction of the guide wire and the blood vessel in the medical image, helps doctors to identify the guide wire and estimate the position of the guide wire in the blood vessel, and greatly improves the timeliness of the segmentation.
2. In the prior art, the marking data used for deep learning are marked manually, which wastes time and labor, and a large amount of marking data sets can be obtained through subtraction technology and morphological operation and are used as training data for model training.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a schematic flow chart of a method for extracting a vascular guidewire from a medical image according to the present invention;
fig. 2 is a schematic structural diagram of a vascular guidewire extraction system in a medical image according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
How to process medical images based on a deep neural network, and improving segmentation efficiency and segmentation precision are one of the technical problems which are urgently needed to be solved at present. In view of the above, the invention applies the deep learning network based on the U-net network structure to the method for simultaneously segmenting and extracting the blood vessel and the guide wire.
How to acquire a large amount of labeled data at low cost to develop deep learning research so as to segment the cerebrovascular image phenomenon is one of the technical problems which are urgently needed to be solved at present. In view of the above, the invention can obtain a large amount of labeled data through an automatic means without manual labeling, thereby carrying out model training; and then, accurately and quickly segmenting the guide wire and the blood vessel in the medical image according to the trained model, realizing the real-time extraction of the guide wire and the blood vessel in the medical image, and helping a doctor to identify the guide wire and estimate the position of the guide wire in the blood vessel.
In conclusion, the method and the device finish the high-efficiency and accurate extraction of the blood vessel and the guide wire through high-quality marking data and network optimization, and have the advantages of high efficiency, high accuracy, low cost and the like.
Example one
As shown in fig. 1, the method for extracting a vascular guidewire from a medical image according to this embodiment includes:
step 101: acquiring a current frame medical image to be processed; the current frame medical image is a scanning image acquired by a special device, and the scanning image comprises a blood vessel which is visualized by contrast in an interventional operation and a guide wire for guiding the placement of a stent into the blood vessel.
Step 102: and inputting the current frame medical image into a trained blood vessel guide wire segmentation model to acquire blood vessel position information and guide wire position information in the current frame medical image.
The trained blood vessel guide wire segmentation model is obtained based on a deep learning algorithm and training of multiple groups of historical images; the history image includes: the system comprises a historical medical image, a blood vessel segmentation annotation image corresponding to the historical medical image and a guide wire segmentation annotation image corresponding to the historical medical image.
The blood vessel segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation, and the guide wire segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation; the vessel subtraction image is determined by differencing the first frame of historical medical image and the other frames of historical medical images.
One example is: the medical image is a DSA image, and the DSA image is a scanned image acquired by a DSA device.
Further, the method for training the vascular guidewire segmentation model according to this embodiment specifically includes:
acquiring a plurality of groups of historical medical images, blood vessel segmentation annotation images corresponding to the historical medical images, and guide wire segmentation annotation images corresponding to the historical medical images.
And inputting a plurality of groups of the historical medical images, the blood vessel segmentation annotation images corresponding to the historical medical images and the guide wire segmentation annotation images corresponding to the historical medical images into a blood vessel guide wire segmentation model to be trained, and performing iterative training on the blood vessel guide wire segmentation model to be trained to obtain the trained blood vessel guide wire segmentation model.
The acquiring of multiple sets of historical medical images, the blood vessel segmentation annotation images corresponding to the historical medical images, and the guide wire segmentation annotation images corresponding to the historical medical images specifically includes:
acquiring a plurality of groups of historical medical images; the total frame number of the historical medical images is N.
And sequencing the historical medical images according to the time sequence to obtain a first frame of historical medical image, namely a background image.
Performing one-to-one difference between the first frame of historical medical image and other frames of historical medical images to obtain N-1 frames of blood vessel subtraction images;
and performing morphological operation on each frame of the blood vessel subtraction image to determine a blood vessel segmentation and annotation image.
And performing morphological operation on each frame of the blood vessel subtraction image to determine a guide wire segmentation annotation image.
The method comprises the steps of inputting a plurality of groups of historical medical images, blood vessel segmentation annotation images corresponding to the historical medical images and guide wire segmentation annotation images corresponding to the historical medical images into a blood vessel guide wire segmentation model to be trained, and performing iterative training on the blood vessel guide wire segmentation model to be trained to obtain the trained blood vessel guide wire segmentation model, and specifically comprises the following steps:
preprocessing the historical medical image, the blood vessel segmentation annotation image corresponding to the historical medical image and the guide wire segmentation annotation image corresponding to the historical medical image; the preprocessing operation includes size transformation, image data augmentation, and grayscale binarization processing.
And inputting the plurality of groups of image data after preprocessing operation into the to-be-trained blood vessel guide wire segmentation model, and performing iterative training on the to-be-trained blood vessel guide wire segmentation model to obtain the trained blood vessel guide wire segmentation model.
The image data after the preprocessing operation comprises a historical medical image after the preprocessing operation, a blood vessel segmentation and annotation image after the preprocessing operation corresponding to the historical medical image after the preprocessing operation and a guide wire segmentation and annotation image after the preprocessing operation corresponding to the historical medical image after the preprocessing operation.
One example is: converting the historical medical image, the blood vessel segmentation annotation image corresponding to the historical medical image and the guide wire segmentation annotation image corresponding to the historical medical image into fixed sizes, then performing image data augmentation preprocessing (the image data augmentation preprocessing method comprises one or more of horizontal overturning, vertical overturning, random scaling, random brightness, random contrast, random noise and random elastic transformation) to augment the image data, and then performing gray level binarization processing to convert gray level pixels of the processed image into 0-1, so as to reduce the scale of input features. Marking and masking the processed historical medical image to obtain a final historical medical image after preprocessing operation; 1 in the historical medical image after the preprocessing operation represents a vascular structure region, and 0 represents a background region.
And taking one part of the obtained image data after the preprocessing operation as a training set and the other part of the obtained image data as a test set, and extracting one part of the obtained image data from the test set for verification.
The vessel guidewire segmentation model described in this embodiment is a network model built by adopting Python language and TensorFlow framework based on a U-Net network architecture; aiming at the structural characteristics, a dense connecting block is used for replacing a convolution layer in a U-Net network architecture, namely the position of the dense connecting block in the blood vessel guide wire segmentation model is the same as the position of the convolution layer in the U-Net network architecture, so that the network training speed and precision are improved, and the requirement on real-time performance can be met; the blood vessel guide wire segmentation model comprises four down-sampling processes and four up-sampling processes, and also comprises jump connection to combine network low-level features with high-level features. The vessel guide wire segmentation model also performs zero-padding operation on the input image.
One example is: the trained blood vessel guide wire segmentation model inputs two same historical medical images after preprocessing operation at the same time, namely two same mask images, and outputs two probability graphs with pixel values of 0 to 1, wherein one probability graph is a blood vessel segmentation actual labeling graph, and the closer to 1 in the graph represents that the probability that the pixel is a blood vessel is higher; another histogram is the actual annotation for guidewire segmentation, and a closer 1 in the histogram indicates a higher probability that the pixel is a guidewire.
Further, the loss function of the blood vessel guide wire segmentation model is a binary cross entropy loss function. The blood vessel guide wire segmentation model adopts a learning rate attenuation method, and after a plurality of epochs are trained, corresponding network model training parameters are obtained.
Inputting the test set into the trained vessel guide wire segmentation model to obtain a segmentation graph predicted by the model, and evaluating the performance of the obtained segmentation graph by adopting Accuracy and Dice values so as to evaluate the trained vessel guide wire segmentation model.
Inputting the current frame medical image into a trained blood vessel guide wire segmentation model to acquire blood vessel position information and guide wire position information in the current frame medical image, and the method specifically comprises the following steps:
and carrying out preprocessing operation on the current frame medical image.
Inputting the current medical image after the preprocessing operation into a trained blood vessel guide wire segmentation model to obtain a blood vessel segmentation annotation image and a guide wire segmentation annotation image corresponding to the current medical image after the preprocessing operation;
and processing the blood vessel segmentation annotation image and the guide wire segmentation annotation image corresponding to the current medical image after the preprocessing operation by adopting a morphological processing algorithm, and determining the blood vessel position information and the guide wire position information in the current medical image.
Example two
As shown in fig. 2, the present embodiment provides a system for extracting a vascular guidewire from a medical image, including:
a current frame medical image obtaining module 201, configured to obtain a current frame medical image to be processed; the current frame medical image is a scanning image acquired by a special device, and the scanning image comprises a blood vessel which is visualized by contrast in an interventional operation and a guide wire for guiding the placement of a stent into the blood vessel.
A blood vessel position information and guide wire position information determining module 202, configured to input the current frame medical image into a trained blood vessel guide wire segmentation model, so as to obtain blood vessel position information and guide wire position information in the current frame medical image.
The trained blood vessel guide wire segmentation model is obtained based on a deep learning algorithm and training of multiple groups of historical images; the history image includes: the method comprises the following steps that a historical medical image, a blood vessel segmentation annotation image corresponding to the historical medical image and a guide wire segmentation annotation image corresponding to the historical medical image are obtained; the blood vessel segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation, and the guide wire segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation; the vessel subtraction image is determined by differencing the first frame of historical medical image and the other frames of historical medical images.
EXAMPLE III
The embodiment provides an electronic device, including:
one or more processors;
storage means for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement a method for extracting a guidewire from a blood vessel in a medical image as described in embodiment one.
Example four
The present embodiment provides a storage medium containing computer executable instructions, which when executed by a computer processor, are used to perform the method for extracting a vascular guidewire from a medical image according to the first embodiment.
The method can obtain a large amount of labeled data without manual labeling, so that model training is performed, guide wires and blood vessels in the medical image are accurately and quickly segmented according to the trained model, and the segmentation efficiency is greatly improved.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (10)
1. A method for extracting a vascular guide wire in a medical image is characterized by comprising the following steps:
acquiring a current frame medical image to be processed; the current frame medical image is a scanned image acquired by special equipment, and the scanned image comprises a blood vessel which appears due to contrast in an interventional operation and a guide wire for guiding a stent to be placed into the blood vessel;
inputting the current frame medical image into a trained blood vessel guide wire segmentation model to acquire blood vessel position information and guide wire position information in the current frame medical image;
the trained blood vessel guide wire segmentation model is obtained based on a deep learning algorithm and training of multiple groups of historical images; the history image includes: the method comprises the following steps that a historical medical image, a blood vessel segmentation annotation image corresponding to the historical medical image and a guide wire segmentation annotation image corresponding to the historical medical image are obtained;
the blood vessel segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation, and the guide wire segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation; the vessel subtraction image is determined by differencing the first frame of historical medical image and the other frames of historical medical images.
2. The method for extracting the vascular guide wire from the medical image according to claim 1, wherein the training method of the vascular guide wire segmentation model specifically comprises:
acquiring a plurality of groups of historical medical images, blood vessel segmentation annotation images corresponding to the historical medical images and guide wire segmentation annotation images corresponding to the historical medical images;
and inputting a plurality of groups of the historical medical images, the blood vessel segmentation annotation images corresponding to the historical medical images and the guide wire segmentation annotation images corresponding to the historical medical images into a blood vessel guide wire segmentation model to be trained, and performing iterative training on the blood vessel guide wire segmentation model to be trained to obtain the trained blood vessel guide wire segmentation model.
3. The method according to claim 2, wherein the acquiring a plurality of sets of historical medical images, blood vessel segmentation labeling images corresponding to the historical medical images, and guide wire segmentation labeling images corresponding to the historical medical images specifically comprises:
acquiring a plurality of groups of historical medical images; the total frame number of the historical medical images is N;
sequencing the historical medical images according to the time sequence to obtain a first frame of historical medical image;
performing one-to-one difference between the first frame of historical medical image and other frames of historical medical images to obtain N-1 frames of blood vessel subtraction images;
performing morphological operation on each frame of the blood vessel subtraction image to determine a blood vessel segmentation annotation image;
and performing morphological operation on each frame of the blood vessel subtraction image to determine a guide wire segmentation annotation image.
4. The method according to claim 2, wherein the step of inputting a plurality of sets of the historical medical images, the blood vessel segmentation labeling images corresponding to the historical medical images, and the guide wire segmentation labeling images corresponding to the historical medical images into the blood vessel guide wire segmentation model to be trained, and the step of performing iterative training on the blood vessel guide wire segmentation model to be trained to obtain the trained blood vessel guide wire segmentation model specifically comprises:
preprocessing the historical medical image, the blood vessel segmentation annotation image corresponding to the historical medical image and the guide wire segmentation annotation image corresponding to the historical medical image; the preprocessing operation comprises size transformation, image data augmentation and gray level binarization processing;
inputting a plurality of groups of image data after preprocessing operation into a to-be-trained blood vessel guide wire segmentation model, and performing iterative training on the to-be-trained blood vessel guide wire segmentation model to obtain a trained blood vessel guide wire segmentation model;
the image data after the preprocessing operation comprises a historical medical image after the preprocessing operation, a blood vessel segmentation and annotation image after the preprocessing operation corresponding to the historical medical image after the preprocessing operation and a guide wire segmentation and annotation image after the preprocessing operation corresponding to the historical medical image after the preprocessing operation.
5. The method for extracting the vascular guidewire from the medical image according to claim 2, wherein the vascular guidewire segmentation model is a network model constructed by adopting a Python language and a TensorFlow framework on the basis of a U-Net network architecture; the position of a dense connecting block in the blood vessel guide wire segmentation model is the same as the position of a convolution layer in the U-Net network architecture; the vascular guide wire segmentation model comprises four down-sampling processes and four up-sampling processes.
6. The method for extracting the vascular guidewire from the medical image according to claim 2, wherein the loss function of the vascular guidewire segmentation model is a binary cross entropy loss function.
7. The method according to claim 1, wherein the current frame medical image is input into a trained vessel guidewire segmentation model to obtain vessel position information and guidewire position information in the current frame medical image, and the method specifically comprises:
preprocessing the current frame medical image;
inputting the current medical image after the preprocessing operation into a trained blood vessel guide wire segmentation model to obtain a blood vessel segmentation annotation image and a guide wire segmentation annotation image corresponding to the current medical image after the preprocessing operation;
and processing the blood vessel segmentation annotation image and the guide wire segmentation annotation image corresponding to the current medical image after the preprocessing operation by adopting a morphological processing algorithm, and determining the blood vessel position information and the guide wire position information in the current medical image.
8. A system for extracting a vascular guide wire in a medical image is characterized by comprising:
the current frame medical image acquisition module is used for acquiring a current frame medical image to be processed; the current frame medical image is a scanned image acquired by special equipment, and the scanned image comprises a blood vessel which appears due to contrast in an interventional operation and a guide wire for guiding a stent to be placed into the blood vessel;
the blood vessel position information and guide wire position information determining module is used for inputting the current frame medical image into a trained blood vessel guide wire segmentation model so as to acquire blood vessel position information and guide wire position information in the current frame medical image;
the trained blood vessel guide wire segmentation model is obtained based on a deep learning algorithm and training of multiple groups of historical images; the history image includes: the method comprises the following steps that a historical medical image, a blood vessel segmentation annotation image corresponding to the historical medical image and a guide wire segmentation annotation image corresponding to the historical medical image are obtained;
the blood vessel segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation, and the guide wire segmentation annotation image is determined after the blood vessel subtraction image is subjected to morphological operation; the vessel subtraction image is determined by differencing the first frame of historical medical image and the other frames of historical medical images.
9. An electronic device, characterized in that the electronic device comprises:
one or more processors;
storage means for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement a method of vascular guidewire extraction in medical images as claimed in any one of claims 1-7.
10. A storage medium containing computer executable instructions for performing the method of vascular guidewire extraction in medical images as claimed in any one of claims 1-7 when executed by a computer processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111513444.1A CN113902746A (en) | 2021-12-13 | 2021-12-13 | Method and system for extracting blood vessel guide wire in medical image, electronic device and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111513444.1A CN113902746A (en) | 2021-12-13 | 2021-12-13 | Method and system for extracting blood vessel guide wire in medical image, electronic device and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113902746A true CN113902746A (en) | 2022-01-07 |
Family
ID=79026087
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111513444.1A Pending CN113902746A (en) | 2021-12-13 | 2021-12-13 | Method and system for extracting blood vessel guide wire in medical image, electronic device and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113902746A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115145453A (en) * | 2022-09-02 | 2022-10-04 | 北京唯迈医疗设备有限公司 | Method, system and storage medium for adjusting display visual angle of medical image |
| CN115546233A (en) * | 2022-10-12 | 2022-12-30 | 上海奥朋医疗科技有限公司 | Method and system for extracting interest region of interventional medical device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109087306A (en) * | 2018-06-28 | 2018-12-25 | 众安信息技术服务有限公司 | Arteries iconic model training method, dividing method, device and electronic equipment |
| CN110163809A (en) * | 2019-03-31 | 2019-08-23 | 东南大学 | Confrontation network DSA imaging method and device are generated based on U-net |
| EP3719744A1 (en) * | 2019-04-06 | 2020-10-07 | Kardiolytics Inc. | Method and system for machine learning based segmentation of contrast filled coronary artery vessels on medical images |
| CN111798451A (en) * | 2020-06-16 | 2020-10-20 | 北京理工大学 | 3D guidewire tracking method and device based on blood vessel 3D/2D matching |
| CN111915538A (en) * | 2020-08-19 | 2020-11-10 | 南京普爱医疗设备股份有限公司 | Image enhancement method and system for digital blood vessel subtraction |
| CN112862850A (en) * | 2021-02-03 | 2021-05-28 | 北京阅影科技有限公司 | Digital silhouette angiography image processing method and device |
-
2021
- 2021-12-13 CN CN202111513444.1A patent/CN113902746A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109087306A (en) * | 2018-06-28 | 2018-12-25 | 众安信息技术服务有限公司 | Arteries iconic model training method, dividing method, device and electronic equipment |
| CN110163809A (en) * | 2019-03-31 | 2019-08-23 | 东南大学 | Confrontation network DSA imaging method and device are generated based on U-net |
| EP3719744A1 (en) * | 2019-04-06 | 2020-10-07 | Kardiolytics Inc. | Method and system for machine learning based segmentation of contrast filled coronary artery vessels on medical images |
| CN111798451A (en) * | 2020-06-16 | 2020-10-20 | 北京理工大学 | 3D guidewire tracking method and device based on blood vessel 3D/2D matching |
| CN111915538A (en) * | 2020-08-19 | 2020-11-10 | 南京普爱医疗设备股份有限公司 | Image enhancement method and system for digital blood vessel subtraction |
| CN112862850A (en) * | 2021-02-03 | 2021-05-28 | 北京阅影科技有限公司 | Digital silhouette angiography image processing method and device |
Non-Patent Citations (1)
| Title |
|---|
| 吴杰 等: "《医学电子学基础与医学影像物理学》", 31 March 2014 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115145453A (en) * | 2022-09-02 | 2022-10-04 | 北京唯迈医疗设备有限公司 | Method, system and storage medium for adjusting display visual angle of medical image |
| CN115145453B (en) * | 2022-09-02 | 2022-12-16 | 北京唯迈医疗设备有限公司 | Method, system and storage medium for adjusting display visual angle of medical image |
| CN115546233A (en) * | 2022-10-12 | 2022-12-30 | 上海奥朋医疗科技有限公司 | Method and system for extracting interest region of interventional medical device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110232383B (en) | Focus image recognition method and focus image recognition system based on deep learning model | |
| CN110287956B (en) | Blood vessel centerline automatic matching method and device | |
| CN110110723A (en) | A kind of method and device that objective area in image automatically extracts | |
| CN113554665A (en) | A kind of blood vessel segmentation method and device | |
| CN113539402B (en) | Multi-mode image automatic sketching model migration method | |
| CN113902746A (en) | Method and system for extracting blood vessel guide wire in medical image, electronic device and medium | |
| CN112862805B (en) | Acoustic neuroma image automatic segmentation method and system | |
| CN113920132B (en) | Coronary CTA segmentation method and system based on multi-scale feature learning network | |
| CN117409002A (en) | Visual identification detection system for wounds and detection method thereof | |
| CN116385467B (en) | Cerebral blood vessel segmentation method and related equipment based on self-supervised learning | |
| CN112785581A (en) | Training method and device for extracting and training large blood vessel CTA (computed tomography angiography) imaging based on deep learning | |
| CN112085696B (en) | Training method and segmentation method for medical image segmentation network model and related equipment | |
| CN116977748A (en) | Colitis image classification method and system | |
| CN113780444B (en) | Training method of tongue fur image classification model based on progressive learning | |
| CN109785325A (en) | A method for multimodal medical imaging based on deep learning | |
| CN116739890B (en) | Method and equipment for training generation model for generating healthy blood vessel image | |
| CN118212411A (en) | A pulmonary embolism segmentation method based on deep learning | |
| CN116758098A (en) | Hypothalamic nucleus segmentation method and model construction method of magnetic resonance image | |
| US20230206428A1 (en) | Tubular structure segmentation | |
| CN116109872A (en) | Blood vessel naming method and device, electronic equipment and storage medium | |
| CN113223003A (en) | Bile duct image segmentation method based on deep learning | |
| CN114049315A (en) | Joint identification method, electronic device, storage medium and computer program product | |
| CN112132850A (en) | A method, system and device for blood vessel boundary detection based on modal learning | |
| Wu et al. | Coarse-to-fine tranformer for articular disc of the temporomandibular joint segmentation | |
| CN119888233B (en) | Pseudo tag selection method, system, device and medium for computer tomography image segmentation based on multi-modal model |
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 |