CN110782422A - Method for synthesizing X-ray film and marking through CT image - Google Patents
Method for synthesizing X-ray film and marking through CT image Download PDFInfo
- Publication number
- CN110782422A CN110782422A CN201911003131.4A CN201911003131A CN110782422A CN 110782422 A CN110782422 A CN 110782422A CN 201911003131 A CN201911003131 A CN 201911003131A CN 110782422 A CN110782422 A CN 110782422A
- Authority
- CN
- China
- Prior art keywords
- image
- ray
- marked
- forward projection
- marking
- 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
- 238000000034 method Methods 0.000 title claims abstract description 31
- 230000002194 synthesizing effect Effects 0.000 title claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 12
- 239000003550 marker Substances 0.000 claims abstract description 11
- 238000012805 post-processing Methods 0.000 claims description 5
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 230000007797 corrosion Effects 0.000 claims description 2
- 238000005260 corrosion Methods 0.000 claims description 2
- 230000002708 enhancing effect Effects 0.000 claims 1
- 210000001519 tissue Anatomy 0.000 description 34
- 210000004072 lung Anatomy 0.000 description 15
- 230000011218 segmentation Effects 0.000 description 12
- 210000000038 chest Anatomy 0.000 description 9
- 230000003902 lesion Effects 0.000 description 9
- 238000012549 training Methods 0.000 description 9
- 238000013135 deep learning Methods 0.000 description 8
- 238000002372 labelling Methods 0.000 description 7
- 238000003745 diagnosis Methods 0.000 description 5
- 238000003384 imaging method Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 210000000056 organ Anatomy 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 210000000481 breast Anatomy 0.000 description 2
- 238000002591 computed tomography Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 231100000915 pathological change Toxicity 0.000 description 2
- 230000036285 pathological change Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 206010035664 Pneumonia Diseases 0.000 description 1
- 206010056342 Pulmonary mass Diseases 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000011976 chest X-ray Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002746 orthostatic effect Effects 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 210000003516 pericardium Anatomy 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 210000000115 thoracic cavity Anatomy 0.000 description 1
- 201000008827 tuberculosis Diseases 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—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
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10116—X-ray image
-
- 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
- G06T2207/30008—Bone
-
- 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
- G06T2207/30048—Heart; Cardiac
-
- 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
- G06T2207/30061—Lung
-
- 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
- G06T2207/30061—Lung
- G06T2207/30064—Lung nodule
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention provides a method for synthesizing an X-ray film and a mark through a CT image, which comprises the following steps: (1) acquiring a CT image; (2) extracting a marked image of the required tissue from the CT image: marking the voxel points belonging to the required tissue in the contour image as a specific value, and marking the rest voxel points as another value to obtain a marked image; (3) carrying out forward projection processing on the obtained CT image at a specific angle to obtain an X-ray image of the CT image; (4) carrying out forward projection processing on the obtained marked image at a specific angle to obtain an X-ray image of the marked image; (5) and synthesizing the X-ray image of the CT image and the X-ray image of the marker image to obtain a synthesized X-ray image with the required tissue marker. The invention synthesizes the X-ray image through the CT image, synthesizes the label on the X-ray image according to the label obtained on the CT image, and can shorten the time and improve the accuracy compared with the method of directly carrying out the label on the X-ray film.
Description
Technical Field
The invention relates to the technical field of CT image processing, in particular to a method for synthesizing an X-ray film and a mark through a CT image, which is used for constructing a training data set of a lesion tissue segmentation deep learning model and belongs to a training data preprocessing part in the training process of the deep learning model.
Background
X-ray film is one of the most commonly used diagnostic imaging techniques. The X-ray machine has low cost and small volume, and is basically a necessary imaging system for hospitals of all levels. Projection images of different tissues can be obtained due to the different X-ray absorption capabilities of different organ tissues. The X-ray film has the advantages of high imaging speed, low cost, small radiation dose, capability of displaying specific pathological structures and the like, and is usually preferred to be used for disease diagnosis and physical examination screening in clinic. However, since the shots are superimposed images, the influence and interference of other tissue structures objectively exist, and for example, the pathological changes of the paravertebral part, the postcardiac part, and other parts cannot be clearly displayed by a simple orthostatic chest film, so that the pathological changes of the parts objectively have the possibility of missed diagnosis. Especially in the process of large-scale physical examination or disease screening, the huge amount of data is very inefficient if it depends on manual methods.
Therefore, development of auxiliary detection technology suitable for X-ray film has become an active research field in recent years (e.g., auxiliary diagnosis of tuberculosis, pneumonia, etc.). The artificial intelligence method based on deep learning can rapidly and accurately segment the lesion area, has strong repeatability, and is a main technique in auxiliary detection technology. Training of artificial intelligence based fast segmentation models of lesion regions generally comprises the following steps: 1. data preparation and preprocessing, 2, network model design and loss function design, 3, network training, 4 and network model verification. The most important part of this is how to generate large and accurate data for neural networks to learn. Using deep learning on small data sets tends to be easy to overfit, so deep learning requires a large amount of data, whereas labeling images is a time consuming process. Especially for X-ray films, due to the imaging principle of the X-ray film, three-dimensional human body structures are compressed on one X-ray film, and a lot of contrast information and three-dimensional space information are lost. It becomes more challenging how to accurately mark regions of tissue structures and lesions, which is not only time consuming but also prone to error due to the nature of overlapping images. Taking an X-ray chest film as an example, the chest cavity contains many important organs of the human body, so the marking of the X-ray chest film is a time-consuming and labor-consuming process. The current labeling method comprises manual labeling and labeling tens of thousands of pictures based on automatic segmentation. Because of the properties of the chest radiograph such as low contrast, organ overlapping, fuzzy boundary and the like, the manual marking directly on the X-ray chest radiograph is time-consuming and has low accuracy. Based on segmentation methods such as threshold segmentation, feature space clustering, region growing, and the like, and based on edge detection, edge tracking methods, and the like, combinations of these segmentation algorithms and new algorithms formed by improvement have been accumulated to reach thousands of times, however, none of these algorithms has good versatility and the labeling effect is worse. There is a need in the art for a method that provides accurate labeling to improve the accuracy of the assisted detection.
Disclosure of Invention
The purpose of the invention is as follows: in order to solve the technical problems that X-ray film marking as training data is difficult and low in accuracy in the existing lesion tissue segmentation technology based on deep learning, the invention provides a method for synthesizing an X-ray film through a CT image and adding an artificial mark. The method utilizes an imaging device CT, which also utilizes X-rays to image a human body, so that the CT image and the X-ray film have great common points. But because of the adoption of the CT scanning mode, the image quality of the CT scanning system has higher contrast and more accurate pixel value compared with an X-ray film. The invention utilizes the data of the CT image to generate the data set of the X-ray film, and provides a more accurate training data set for a lesion region fast segmentation model based on deep learning and taking the X-ray film as input.
The technical scheme is as follows: the invention provides the following technical scheme:
a method for synthesizing X-ray film and mark by CT image includes the following steps:
(1) acquiring a CT image;
(2) segmenting outline images of different tissues on the CT image;
(3) extracting a mark image of the required tissue from the contour image: marking the voxel points belonging to the required tissue in the contour image as a specific value, and marking the rest voxel points as another value to obtain a marked image;
(4) carrying out forward projection processing of a specific angle on the contour image obtained in the step (2) to obtain an X-ray image of the contour image;
(5) carrying out forward projection processing of a specific angle on the mark image added in the step (2) to obtain an X-ray image of the mark image;
(6) and synthesizing the X-ray image of the contour image and the X-ray image of the mark image to obtain a synthesized X-ray image with the required tissue mark.
Further, the forward projection mode includes cone beam forward projection, fan beam forward projection and parallel beam forward projection.
Further, the step (4) further includes performing enhancement processing on the X-ray image of the obtained contour image.
Specifically, the enhancement processing method is an adaptive image equalization method.
Further, before the step (6) is executed, the method further comprises the steps of: and (4) carrying out post-processing on the X-ray image of the marked image obtained in the step (5) to eliminate burrs in the image, wherein the post-processing comprises corrosion and expansion operations.
Has the advantages that: compared with the prior art, the invention has the following advantages:
the invention firstly synthesizes an X-ray image through a CT image and synthesizes a label on the X-ray image according to the label obtained on the CT image. Compared with the marking on an X-ray image, the marking on the CT image by a doctor can obviously shorten the marking time and improve the accuracy, and the automatic segmentation marking method on the CT image is also obviously superior to the marking method on the X-ray image.
By the method, the training data set with the accurate marks can be provided for the lesion tissue segmentation model based on deep learning, so that the trained lesion tissue segmentation model based on deep learning is higher in accuracy.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention;
FIG. 2 is a CT image of a breast, from left to right, in the order of a cross-sectional, coronal and sagittal image;
FIG. 3 is a chest lung tissue marker image, in left-to-right order, a cross-sectional, coronal and sagittal image;
FIG. 4 is an X-ray image of a CT image of the breast;
FIG. 5 is an enhanced chest X-ray image;
FIG. 6 is an X-ray image of the lungs;
FIG. 7 is an X-ray image of the lungs after post-treatment;
fig. 8 is a composite X-ray image of the chest with lung tissue markers.
Detailed Description
The invention will be further described with reference to the following drawings and specific embodiments.
FIG. 1 is a flow chart of an embodiment of the present invention, comprising the steps of:
(1) data collection and sorting: dicom image data or CT image data in other formats is collected. In this embodiment, a dicom image of a chest CT is taken, as shown in fig. 2, the image of fig. 2 is sequentially a cross-sectional image, a coronal image and a sagittal image from left to right, each image has a size of 1024 × 800, and there are a total of 800 images, and each image has 1024 rows and columns. The pixel size of the image is 0.3125mm by 0.3125mm, and the smaller the pixel size of the image, the higher the resolution, and the better the quality of the synthesized X-ray image and the labeled image.
(2) Performing forward projection processing of a specific angle, such as 0-degree or 90-degree forward projection processing, on the CT image to obtain an X-ray image of the CT image; the forward projection can be performed according to the geometric parameters of cone beam, fan beam or parallel beam, and the embodiment adopts the simplest geometric parameters of parallel beam, and the forward projection obtains 0-degree synthesized X-ray image, as shown in FIG. 4.
(3) The X-ray image of CT is enhanced to increase the contrast of the image. There are many image enhancement methods, and in this embodiment, an adaptive image equalization method is used to enhance an image, and the enhanced image is shown in fig. 5.
(4) The tissue of interest is extracted, that is, usually on the DR flat sheet, the tissue region needs to be seen, such as the outline of the lung, the outline of the pericardium, the rib structure, the lung nodule, etc., and the doctor needs to see the image of the tissue first and then give a diagnosis suggestion in the diagnosis. This tissue of interest may be extracted singly or plurally. We extract the tissue of interest by means of image labeling: and (3) marking the voxel points belonging to the required tissue in the CT image obtained in the step (1) as a specific value, and marking the rest voxel points as another value to obtain a marked image. In this embodiment, a lung tissue is selected as a required tissue to be marked, the lung tissue is extracted from the dicom image of the chest CT shown in fig. 2, a voxel point belonging to the lung tissue is marked as 1, and a voxel point not belonging to the lung tissue is marked as 0, so that a marked image of the lung tissue is obtained, as shown in fig. 3, a cross-section marked image of the lung tissue, a coronal plane marked image of the lung tissue, and a sagittal plane marked image of the lung tissue are sequentially shown from left to right in fig. 3.
(5) The marker image obtained in step (4) is subjected to forward projection processing as in the case of the CT image to obtain an X-ray image of the marker image, i.e., an X-ray image of lung tissue, as shown in fig. 6.
(6) And (3) post-treatment: post-processing operations such as erosion and dilation are performed on the X-ray image of the lung tissue to remove the burred points in the image, and the processed image is shown in fig. 7.
(7) The X-ray image of the CT image is combined with the X-ray image of the marker image to obtain a combined X-ray image with the desired tissue marker, as shown in fig. 8.
When a lesion tissue segmentation model based on deep learning is trained, a neural network is trained using a synthetic X-ray image with a desired tissue marker as training data.
The above description is only of the preferred embodiments of the present invention, and it should be noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.
Claims (5)
1. A method for synthesizing X-ray film and mark by CT image includes steps:
(1) acquiring a CT image;
(2) extracting a marked image of the required tissue from the CT image: marking the voxel points belonging to the required tissue in the contour image as a specific value, and marking the rest voxel points as another value to obtain a marked image;
(3) carrying out forward projection processing of a specific angle on the CT image obtained in the step (1) to obtain an X-ray image of the CT image;
(4) carrying out forward projection processing of a specific angle on the marked image obtained in the step (2) to obtain an X-ray image of the marked image;
(5) and synthesizing the X-ray image of the CT image and the X-ray image of the marker image to obtain a synthesized X-ray image with the required tissue marker.
2. The method of claim 1, wherein the forward projection modes include cone beam forward projection, fan beam forward projection and parallel beam forward projection.
3. The method of claim 1, wherein the step (4) further comprises enhancing the X-ray image of the contour image.
4. The method of claim 1, wherein the enhancement processing is adaptive image equalization.
5. The method for synthesizing X-ray film and marker by CT image as claimed in claim 1, further comprising the step of, before performing step (6): and (4) carrying out post-processing on the X-ray image of the marked image obtained in the step (5) to eliminate burrs in the image, wherein the post-processing comprises corrosion and expansion operations.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911003131.4A CN110782422A (en) | 2019-10-21 | 2019-10-21 | Method for synthesizing X-ray film and marking through CT image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911003131.4A CN110782422A (en) | 2019-10-21 | 2019-10-21 | Method for synthesizing X-ray film and marking through CT image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110782422A true CN110782422A (en) | 2020-02-11 |
Family
ID=69386233
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911003131.4A Pending CN110782422A (en) | 2019-10-21 | 2019-10-21 | Method for synthesizing X-ray film and marking through CT image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110782422A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103315764A (en) * | 2013-07-17 | 2013-09-25 | 沈阳东软医疗系统有限公司 | Method for acquiring CT locating images and CT device |
| CN107545551A (en) * | 2017-09-07 | 2018-01-05 | 广州华端科技有限公司 | The method for reconstructing and system of digital galactophore body layer composograph |
| CN108711177A (en) * | 2018-05-15 | 2018-10-26 | 南方医科大学口腔医院 | The fast automatic extracting method of volume data arch wire after a kind of oral cavity CBCT is rebuild |
-
2019
- 2019-10-21 CN CN201911003131.4A patent/CN110782422A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103315764A (en) * | 2013-07-17 | 2013-09-25 | 沈阳东软医疗系统有限公司 | Method for acquiring CT locating images and CT device |
| CN107545551A (en) * | 2017-09-07 | 2018-01-05 | 广州华端科技有限公司 | The method for reconstructing and system of digital galactophore body layer composograph |
| CN108711177A (en) * | 2018-05-15 | 2018-10-26 | 南方医科大学口腔医院 | The fast automatic extracting method of volume data arch wire after a kind of oral cavity CBCT is rebuild |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Automatic segmentation of individual tooth in dental CBCT images from tooth surface map by a multi-task FCN | |
| CN108898595B (en) | Construction method and application of positioning model of focus region in chest image | |
| CN105105775B (en) | Cardiac motion resolver | |
| CN103294883A (en) | Method and system for intervention planning for transcatheter aortic valve implantation | |
| KR20150045885A (en) | Systems and methods for registration of ultrasound and ct images | |
| CN109801276B (en) | Method and device for calculating heart-chest ratio | |
| JP2017064370A (en) | Image processing device, and method and program for controlling image processing device | |
| CN112862833A (en) | Blood vessel segmentation method, electronic device and storage medium | |
| CN104616289A (en) | Removal method and system for bone tissue in 3D CT (Three Dimensional Computed Tomography) image | |
| CN109934829B (en) | Liver segmentation method based on three-dimensional graph segmentation algorithm | |
| CN111340825A (en) | Method and system for generating mediastinal lymph node segmentation model | |
| EP2689344B1 (en) | Knowledge-based automatic image segmentation | |
| Jimenez-Carretero et al. | Optimal multiresolution 3D level-set method for liver segmentation incorporating local curvature constraints | |
| CN111311626A (en) | Skull fracture automatic detection method based on CT image and electronic medium | |
| CN110866905A (en) | Rib identification and marking method | |
| Tseng et al. | An adaptive thresholding method for automatic lung segmentation in CT images | |
| CN111080676B (en) | Method for tracking endoscope image sequence feature points through online classification | |
| CN110570430B (en) | Orbital bone tissue segmentation method based on volume registration | |
| Karthikeyan et al. | Lungs segmentation using multi-level thresholding in CT images | |
| CN116797612B (en) | Ultrasonic image segmentation method and device based on weak supervision depth activity contour model | |
| CN104915989A (en) | CT image-based blood vessel three-dimensional segmentation method | |
| CN116309647B (en) | Method for constructing craniocerebral lesion image segmentation model, image segmentation method and device | |
| WO2023092124A1 (en) | Machine-learning based segmentation of biological objects in medical images | |
| CN116168097A (en) | Method, device, equipment and medium for constructing CBCT sketching model and sketching CBCT image | |
| CN110782422A (en) | Method for synthesizing X-ray film and marking through CT image |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200211 |