CN116777962A - Two-dimensional medical image registration method and system based on artificial intelligence - Google Patents
Two-dimensional medical image registration method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN116777962A CN116777962A CN202310828003.3A CN202310828003A CN116777962A CN 116777962 A CN116777962 A CN 116777962A CN 202310828003 A CN202310828003 A CN 202310828003A CN 116777962 A CN116777962 A CN 116777962A
- Authority
- CN
- China
- Prior art keywords
- image
- dimensional medical
- medical image
- feature space
- feature
- 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 58
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 27
- 230000033001 locomotion Effects 0.000 claims abstract description 64
- 238000012545 processing Methods 0.000 claims abstract description 41
- 238000001514 detection method Methods 0.000 claims abstract description 31
- 230000009467 reduction Effects 0.000 claims abstract description 19
- 238000012544 monitoring process Methods 0.000 claims abstract description 17
- 239000008280 blood Substances 0.000 claims description 45
- 210000004369 blood Anatomy 0.000 claims description 45
- 210000004204 blood vessel Anatomy 0.000 claims description 38
- 238000004422 calculation algorithm Methods 0.000 claims description 28
- 239000013598 vector Substances 0.000 claims description 25
- 238000000605 extraction Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 15
- 238000006073 displacement reaction Methods 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 9
- 230000009466 transformation Effects 0.000 claims description 7
- 230000006399 behavior Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 238000006386 neutralization reaction Methods 0.000 claims description 6
- 230000003287 optical effect Effects 0.000 claims description 6
- 210000000056 organ Anatomy 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000006835 compression Effects 0.000 claims description 4
- 238000007906 compression Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000005259 measurement Methods 0.000 abstract description 4
- 238000002591 computed tomography Methods 0.000 description 30
- 230000000875 corresponding effect Effects 0.000 description 26
- 238000003745 diagnosis Methods 0.000 description 13
- 238000004458 analytical method Methods 0.000 description 7
- 238000002595 magnetic resonance imaging Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 4
- 238000010191 image analysis Methods 0.000 description 4
- 238000002405 diagnostic procedure Methods 0.000 description 3
- 238000012937 correction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 206010028980 Neoplasm Diseases 0.000 description 1
- 208000007536 Thrombosis Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000017531 blood circulation Effects 0.000 description 1
- 210000000481 breast Anatomy 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000003702 image correction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000007170 pathology Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 230000029058 respiratory gaseous exchange Effects 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
- 230000002792 vascular Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- 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
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- 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)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention relates to a two-dimensional medical image registration method and a system based on artificial intelligence, wherein the method comprises the steps of obtaining a first two-dimensional medical image, extracting image characteristics of the first two-dimensional medical image to obtain image characteristics, and uniformly converting a plurality of image characteristics to generate corresponding characteristic spaces; performing foreground detection on a moving target in a plurality of image features, and extracting a moving region of one or more image features; randomly calibrating four mutually perpendicular detection points for a plurality of image features in the feature space, and carrying out real-time vertical monitoring on the detection points in the feature space through a motion area; and acquiring a second two-dimensional medical image, loading the second two-dimensional medical image into a feature space to perform artifact reduction processing and non-rigid deformation compensation processing, and outputting the processed second two-dimensional medical image to realize a multi-mode two-dimensional medical image registration method by combining foreground detection, three-dimensional measurement, non-rigid deformation compensation and motion artifact reduction.
Description
Technical Field
The invention relates to the technical field of image data processing, in particular to a two-dimensional medical image registration method and system based on artificial intelligence.
Background
Artificial Intelligence (AI) is a subject in the field of computer science that studies and simulates human intelligence. The computer has the functions of human perception, reasoning, planning, learning, execution, communication and the like, and is mainly applied to the fields of machine learning, deep learning, computer vision, natural language processing, expert decision and the like at present.
The existing two-dimensional medical image technology is an important clinical medical auxiliary diagnosis means, and image equipment in different modes such as breast X-ray, magnetic Resonance Imaging (MRI), computed Tomography (CT), ultrasonic waves and the like is used for collecting body tissue images of a patient to generate two-dimensional slice images, so that the two-dimensional medical image technology is widely applied to the fields of diagnosis support, treatment planning, operation navigation and the like, but a plurality of difficulties and defects exist on the existing two-dimensional medical image registration defense line:
the images with different modes (such as CT and MRI) are faced, and the problems of intensity difference, contrast difference, noise and the like among different types of images need to be processed in the two-dimensional medical image registration, so that the matching measurement is difficult to realize.
Non-rigid deformation, which is difficult to model, adds complexity to the registration task because biological tissue is susceptible to non-rigid deformation (e.g., organ deformation, tumor shrinkage, etc.) at different times or in different poses.
Motion artifacts, during the process of acquiring images, physiological factors (such as heartbeat, respiration, etc.) of a patient may cause the images to appear motion artifacts, which affect alignment accuracy.
Disclosure of Invention
The invention mainly aims to provide a two-dimensional medical image registration method and a system based on artificial intelligence, which are combined with a multi-mode two-dimensional medical image registration method provided by foreground detection, three-dimensional measurement, non-rigid deformation compensation and motion artifact reduction, and the two-dimensional medical image is processed through AI so as to provide technical support for wider medical image analysis and diagnosis.
In order to achieve the above object, the two-dimensional medical image registration method based on artificial intelligence provided by the invention comprises the following steps:
acquiring a first two-dimensional medical image comprising CT images and/or MRI images of different modalities;
extracting image features of the first two-dimensional medical image to obtain a plurality of image features matched with the first two-dimensional medical image, and uniformly converting the plurality of image features to generate corresponding feature spaces;
performing foreground detection on moving targets in a plurality of image features in the feature space, and extracting moving areas of one or more image features;
randomly calibrating four mutually perpendicular detection points for a plurality of image features in the feature space, and carrying out real-time perpendicular monitoring on the detection points in the feature space through the motion area;
acquiring a second two-dimensional medical image, wherein the second two-dimensional medical image is a two-dimensional medical image acquired after the first two-dimensional medical image;
and loading the second two-dimensional medical image into a feature space for artifact reduction processing and non-rigid deformation compensation processing, and outputting the processed second two-dimensional medical image.
Further, the step of loading the second two-dimensional medical image into a feature space for artifact reduction processing includes:
according to the blood vessel CT image in the second two-dimensional image, carrying out gray-scale setting on the characteristic space, and identifying blood information of the blood vessel CT image through a motion area of the characteristic space;
performing ROI (region of interest) compression on the blood information in the characteristic space after the graying setting to acquire a blood vessel wall image and a black blood image positioned in the blood vessel wall;
and performing blood vessel path matching on the blood vessel wall image and the black blood image and the image features in the feature space, and deleting the part, exceeding the image feature blood vessel path, of the blood vessel wall image and the black blood image.
Further, after the step of performing gray-scale setting on the feature space according to the blood vessel CT image in the second two-dimensional image and identifying the blood information of the blood vessel CT image through the motion region of the feature space, the method includes:
determining a plurality of image features matched with blood information in the feature space;
performing interesting settings on a number of image features to determine unimportant blood information in the blood information through the interesting settings;
and deleting the image features corresponding to the unimportant blood information in the feature space.
Further, the step of loading the second two-dimensional medical image into a feature space for artifact reduction processing further includes:
acquiring other CT images in the second two-dimensional medical image, including but not limited to, a skeleton, a heart, and other organs;
inputting the other CT images into a feature space, and aligning the motion areas of the other CT images in the feature space;
and deleting the image parts of the other CT images exceeding the preset displacement by carrying out displacement presetting on the motion area.
Further, the step of loading the second two-dimensional medical image into a feature space for non-rigid deformation compensation processing includes:
performing deformation field calculation on a second two-dimensional medical image in the feature space by using a B-spline transformation algorithm, and generating a point cloud set matched with the second two-dimensional medical image;
obtaining a difference value of the compared image features and the point cloud set through the comparison of the point cloud set and the image features in the feature space;
and correcting the second two-dimensional medical image according to the difference to the corresponding compensation.
Further, the step of extracting image features of the first two-dimensional medical image to obtain a plurality of image features matched with the first two-dimensional medical image, and uniformly converting the plurality of image features to generate a corresponding feature space includes:
performing identification extraction on the first two-dimensional medical image by adopting a SIFT feature extraction algorithm and a SURF feature extraction algorithm, and respectively identifying a corresponding first feature group and a corresponding second feature group;
matching the first feature set and the second feature set by using a FLANN matching method to match a plurality of approximate features in the first feature set and the second feature set;
coordinate conversion and normalization processing are carried out on the approximate features serving as image features, so that a plurality of image features are generated into corresponding feature spaces;
and connecting a plurality of image features of the feature space with a subsequent image transmission interface to acquire subsequently input two-dimensional medical image data.
Further, the step of performing foreground detection on a moving object in a plurality of image features in the feature space and extracting a moving region of one or more image features includes:
identifying motion vectors for a plurality of image feature pairs in the feature space;
identifying moving targets and speeds of a plurality of the motion vectors through a Lucas-Kanade algorithm;
performing optical flow field thresholding on the calculated moving object and velocity to filter smaller motion vectors;
utilizing switching operation to eliminate isolated noise points in a plurality of motion vectors;
and connecting a plurality of motion areas corresponding to the motion vectors.
Further, calibrating four mutually perpendicular detection points randomly for a plurality of image features in the feature space, and performing real-time vertical monitoring on the detection points in the feature space through the motion area, wherein the method comprises the following steps:
continuously acquiring a first two-dimensional medical image and converting the first two-dimensional medical image into image features which are input into a feature space in a one-to-one correspondence manner;
presetting four points which are vertical, and marking the four points on the image features of the feature space;
monitoring whether the angles of connecting lines of the four points change in real time.
Further, after the step of monitoring whether the angles of the connecting lines of the four points change in real time, the method comprises the following steps:
and if the image characteristics in the characteristic space are changed, the image characteristics are determined to be subjected to artifact reduction processing and non-rigid deformation compensation processing, and then corresponding neutralization behaviors are generated, and the generation process of the characteristic space is re-executed according to the determination, wherein the neutralization behaviors are generated by carrying out mean value calibration on the image characteristics after the second two-dimensional medical image is input.
The invention also provides a two-dimensional medical image registration system based on the artificial intelligence, which comprises a computer system, wherein the computer system executes the two-dimensional medical image registration method based on the artificial intelligence.
The two-dimensional medical image registration method and system based on artificial intelligence provided by the invention have the following beneficial effects:
(1) Image quality improvement: by advanced feature extraction algorithms and artifact reduction processing, image quality is improved for more accurate medical analysis.
(2) Non-rigid deformation compensation treatment: the B-spline transformation algorithm is used for carrying out non-rigid deformation compensation on the image, and the image accuracy is improved by considering the possible deformation of the image under different modes.
(3) Flexible feature space: and a plurality of feature extraction algorithms and matching technologies are adopted for unified conversion, so that a strict and flexible feature space is generated, and the method is suitable for different types of medical images.
(4) Real-time moving object detection: the moving target is effectively identified through a foreground detection technology (such as a Lucas-Kanade algorithm), and the reliability of a diagnosis result is improved.
Drawings
FIG. 1 is a schematic illustration of steps of an artificial intelligence based two-dimensional medical image registration method in accordance with an embodiment of the present invention;
the achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1, the two-dimensional medical image registration method based on artificial intelligence provided by the invention comprises the following steps:
s1, acquiring a first two-dimensional medical image, wherein the first two-dimensional medical image comprises CT images and/or MRI images of different modalities;
s2, extracting image features of the first two-dimensional medical image to obtain a plurality of image features matched with the first two-dimensional medical image, and uniformly converting the plurality of image features to generate corresponding feature spaces;
s3, performing foreground detection on moving targets in a plurality of image features in the feature space, and extracting moving areas of one or more image features;
s4, randomly calibrating four mutually perpendicular detection points for a plurality of image features in the feature space, and carrying out real-time vertical monitoring on the detection points in the feature space through the motion area;
s5, acquiring a second two-dimensional medical image,
and S6, loading the second two-dimensional medical image into a feature space for artifact reduction processing and non-rigid deformation compensation processing, and outputting the processed second two-dimensional medical image.
Specifically, the first two-dimensional medical image and the second two-dimensional medical image are the same image, but are different tenses, and when specifically executed: acquiring a first two-dimensional medical image, which may comprise CT images and/or MRI images of different modalities; performing image feature extraction on the first two-dimensional medical image to obtain image features matched with the first two-dimensional medical image, and performing unified conversion on the image features to create a corresponding feature space; performing recognition on a moving target through foreground detection in image features in a feature space, and extracting a moving region of one or more image features; randomly calibrating four mutually perpendicular detection points in the feature space, and perpendicularly monitoring the detection points of the feature space in real time based on the motion area; acquiring a second two-dimensional medical image, loading the second two-dimensional medical image into a feature space, performing artifact reduction processing and non-rigid deformation compensation processing, and finally outputting the processed second two-dimensional medical image; the method can improve the accuracy for medical image analysis, enhance the diagnosis reliability and promote the working efficiency of doctors and medical institutions.
In one embodiment, the step of loading the second two-dimensional medical image into a feature space for artifact reduction processing includes:
according to the blood vessel CT image in the second two-dimensional image, carrying out gray-scale setting on the characteristic space, and identifying blood information of the blood vessel CT image through a motion area of the characteristic space;
performing ROI (region of interest) compression on the blood information in the characteristic space after the graying setting to acquire a blood vessel wall image and a black blood image positioned in the blood vessel wall;
and performing blood vessel path matching on the blood vessel wall image and the black blood image and the image features in the feature space, and deleting the part, exceeding the image feature blood vessel path, of the blood vessel wall image and the black blood image.
The method is characterized in that when in specific implementation: in the feature space, the vessel CT image in the second two-dimensional medical image is subjected to graying processing, namely, the color image is converted into a gray-scale image so as to further identify the blood information of the vessel CT image. Blood information is identified in the vessel CT image using a motion region in the feature space, the motion region representing a blood flow region in the image. And (3) pressing the blood information in the characteristic space after the graying setting into the ROI (region of interest). The aim of the ROI compression is to remove blood information areas which are not concerned from the image, so as to acquire a blood vessel wall image and a black blood image positioned in the blood vessel wall. This step may improve the efficiency and accuracy of medical image processing, facilitating subsequent analysis of the vascular condition. And performing blood vessel path matching on the blood vessel wall image and the black blood image and the image features in the feature space. Vessel path matching means that the vessel wall and black blood images are aligned and integrated with the corresponding vessel portions in the feature space. And removing the part which is far away from the blood vessel path range of the characteristic space in the blood vessel wall image and the black blood image in the matching process. This will make the blood vessels in the image clearer and accurately display the condition of the blood vessels, thereby improving the accuracy of the medical professional during the diagnosis.
In one embodiment, after the step of performing gray-scale setting on the feature space according to the blood vessel CT image in the second two-dimensional image and identifying the blood information of the blood vessel CT image through the motion region of the feature space, the method includes:
determining a plurality of image features matched with blood information in the feature space;
performing interesting settings on a number of image features to determine unimportant blood information in the blood information through the interesting settings;
and deleting the image features corresponding to the unimportant blood information in the feature space.
In the specific implementation process: the blood information is identified in a feature space and a number of image features matching the blood information are determined. These image features will be used to further distinguish between important and non-important blood information by comparison and analysis. The settings of interest are performed on these image features. The purpose of the interesting arrangement is to screen blood information of interest to the medical professional, rejecting non-important parts. For example, a practitioner may be interested in blood clots or other features in the blood, and wish to extract such information for subsequent analysis. Pruning the non-important blood information in the feature space will ensure that the image is more focused on the important information. By deleting image features corresponding to unimportant blood information, feature space will be focused only on areas of interest to the medical professional, improving accuracy and efficiency in the diagnostic process. Through the steps, the medical image registration method can purposefully process the images, focus on important information and remove unnecessary parts, so that the effect of the diagnosis process is improved.
In one embodiment, the step of loading the second two-dimensional medical image into a feature space for artifact reduction processing further comprises:
acquiring other CT images in the second two-dimensional medical image, including but not limited to, a skeleton, a heart, and other organs;
inputting the other CT images into a feature space, and aligning the motion areas of the other CT images in the feature space;
and deleting the image parts of the other CT images exceeding the preset displacement by carrying out displacement presetting on the motion area.
The method is characterized in that when in specific implementation: other CT images in the second two-dimensional medical image are acquired, which may include, but are not limited to, a skeleton, a heart, and other organs. These images provide important information for the doctor to use for diagnosis. These other CT images are input into the feature space and then registered according to the motion regions of these images in the feature space. The motion regions enable important parts in different CT images to be identified in the feature space and adjusted appropriately. And carrying out displacement presetting on the motion area. The displacement preset is to set a range so that only image information of the region of interest is focused on during diagnosis. Image portions outside this range are considered unimportant. And deleting the image parts exceeding the preset displacement range in other CT images. This helps focus on important image information and improves the accuracy and efficiency of the medical image diagnostic process.
In one embodiment, the step of loading the second two-dimensional medical image into a feature space for non-rigid deformation compensation processing comprises:
performing deformation field calculation on a second two-dimensional medical image in the feature space by using a B-spline transformation algorithm, and generating a point cloud set matched with the second two-dimensional medical image;
obtaining a difference value of the compared image features and the point cloud set through the comparison of the point cloud set and the image features in the feature space;
and correcting the second two-dimensional medical image according to the difference to the corresponding compensation.
The method is characterized in that when in specific implementation: and applying a B-spline transformation algorithm to the feature space to calculate a deformation field of the second two-dimensional medical image. The B-spline transformation algorithm is an elastic transformation method widely applied to image registration, and can effectively process non-rigid deformation, so that the accuracy of image processing is improved. And generating a point cloud set matched with the second two-dimensional medical image through deformation field calculation. This point cloud contains key feature points of the medical image, which facilitates the comparison and analysis of the second two-dimensional medical image with the image features in the feature space. In the process of comparing the point cloud set with the image features in the feature space, calculating a difference value between the compared image features and the point cloud set. This difference value reflects the degree of similarity of the second two-dimensional medical image to the image features in the feature space and can be used for image correction and optimization. And carrying out corresponding compensation correction on the second two-dimensional medical image according to the calculated difference value. By compensation and correction, the quality of the image can be improved, and the accuracy and efficiency in the diagnosis process can be improved.
In one embodiment, the step of extracting image features of the first two-dimensional medical image to obtain a plurality of image features matched with the first two-dimensional medical image, and uniformly converting the plurality of image features to generate a corresponding feature space includes:
performing identification extraction on the first two-dimensional medical image by adopting a SIFT feature extraction algorithm and a SURF feature extraction algorithm, and respectively identifying a corresponding first feature group and a corresponding second feature group;
matching the first feature set and the second feature set by using a FLANN matching method to match a plurality of approximate features in the first feature set and the second feature set;
coordinate conversion and normalization processing are carried out on the approximate features serving as image features, so that a plurality of image features are generated into corresponding feature spaces;
and connecting a plurality of image features of the feature space with a subsequent image transmission interface to acquire subsequently input two-dimensional medical image data.
The method is characterized in that when in specific implementation: and performing identification extraction on the first two-dimensional medical image by adopting a SIFT (scale invariant feature transform) feature extraction algorithm and a SURF (speeded up robust feature) feature extraction algorithm. The two algorithms are currently commonly used image feature extraction algorithms, and can effectively identify key features in images. The features identified and extracted by the SIFT algorithm and the SURF algorithm are respectively classified into a first feature group and a second feature group. In this way, each set of features contains key information for the first two-dimensional medical image, facilitating subsequent image matching and fusion. And matching the first feature set and the second feature set by using a FLANN (fast neighbor matching) matching method. The method can efficiently find out similar or approximate features between the two feature groups, and the approximate features are used as a basis for subsequent image processing. The approximate features are subjected to coordinate conversion and normalization processing, and are generated into corresponding feature spaces. Within this feature space, various image features may be further analyzed and processed. The plurality of image features of the feature space are connected with a subsequent image transmission interface, so that subsequently input two-dimensional medical image data can be received and processed in real time. This helps to quickly process image data of different sources, improving the effect and accuracy of image diagnosis.
In one embodiment, the step of performing foreground detection on a moving object in a plurality of image features in the feature space and extracting a moving region of one or more image features includes:
identifying motion vectors for a plurality of image feature pairs in the feature space;
identifying moving targets and speeds of a plurality of the motion vectors through a Lucas-Kanade algorithm;
performing optical flow field thresholding on the calculated moving object and velocity to filter smaller motion vectors;
utilizing switching operation to eliminate isolated noise points in a plurality of motion vectors;
and connecting a plurality of motion areas corresponding to the motion vectors.
In a specific implementation, motion vectors for a number of image features in a feature space are identified. A motion vector is information representing the change in the position of a moving object in an image, which helps to analyze the relationship between features of the image. The Lucas-Kanade algorithm is used to identify moving objects and velocities for several motion vectors. The Lucas-Kanade algorithm is a commonly used method of calculating optical flow (optical flow) that can effectively estimate the displacement of image features between adjacent frames. By this algorithm, the moving direction and speed of the moving object can be obtained. The calculated moving objects and velocities are subjected to optical flow field thresholding to filter smaller motion vectors. This process can eliminate those motion vectors that are not important or tiny, focusing on significant motion information. And eliminating isolated noise points in a plurality of motion vectors by using an opening and closing operation. The open-close operation is a common image morphology operation, which can remove noise in an image and retain important characteristics. Thus, the quality of the motion vector can be improved, and the accuracy of subsequent analysis can be improved. And connecting motion areas corresponding to the motion vectors. By correlating these motion vectors, a motion field can be formed to more clearly identify motion information in the image. Through these steps, motion vectors of image features can be efficiently identified and processed in feature space, thereby achieving more accurate and detailed image analysis during diagnosis
In one embodiment, the step of randomly calibrating four mutually perpendicular detection points for a plurality of image features in the feature space and performing real-time vertical monitoring on the detection points in the feature space through the motion area includes:
continuously acquiring a first two-dimensional medical image and converting the first two-dimensional medical image into image features which are input into a feature space in a one-to-one correspondence manner;
presetting four points which are vertical, and marking the four points on the image features of the feature space;
monitoring whether the angles of connecting lines of the four points change in real time.
In a specific implementation process, continuously acquiring a first two-dimensional medical image: the system will continuously receive the input data of the first two-dimensional medical image and thereby update the image information in real time. Converting the medical image into image features and inputting the image features into a feature space: key features in the first two-dimensional medical image are extracted by feature extraction algorithms such as SIFT, SURF, etc., as previously discussed, and these image features are input into feature space for subsequent processing and analysis. Presetting four points and marking: to monitor the image feature variations in the feature space, four points are preset and ensure that they are in a perpendicular position to each other. And then marking the four points on the image features of the feature space, thereby realizing the tracking and monitoring of the points. Monitoring whether the angles of connecting lines of four points change in real time: the change of the image characteristics in the characteristic space can be captured by monitoring the angle change of connecting lines among the four marking points in real time. This helps to assess changes in medical image quality, movement information, etc., and thus to understand the development of patient pathology and take corresponding action in the diagnostic process.
Further, after the step of monitoring whether the angles of the connecting lines of the four points change in real time, the method comprises the following steps:
and if the image characteristics in the characteristic space are changed, the image characteristics are determined to be subjected to artifact reduction processing and non-rigid deformation compensation processing, and then corresponding neutralization behaviors are generated, and the generation process of the characteristic space is re-executed according to the determination, wherein the neutralization behaviors are generated by carrying out mean value calibration on the image characteristics after the second two-dimensional medical image is input.
The invention provides an artificial intelligence-based two-dimensional medical image registration system, which comprises a computer system, wherein the computer system executes the artificial intelligence-based two-dimensional medical image registration method.
S1, acquiring a first two-dimensional medical image, wherein the first two-dimensional medical image comprises CT images and/or MRI images of different modalities;
s2, extracting image features of the first two-dimensional medical image to obtain a plurality of image features matched with the first two-dimensional medical image, and uniformly converting the plurality of image features to generate corresponding feature spaces;
s3, performing foreground detection on moving targets in a plurality of image features in the feature space, and extracting moving areas of one or more image features;
s4, randomly calibrating four mutually perpendicular detection points for a plurality of image features in the feature space, and carrying out real-time vertical monitoring on the detection points in the feature space through the motion area;
s5, acquiring a second two-dimensional medical image,
and S6, loading the second two-dimensional medical image into a feature space for artifact reduction processing and non-rigid deformation compensation processing, and outputting the processed second two-dimensional medical image.
Therefore, the multi-mode two-dimensional medical image registration method combining foreground detection, three-dimensional measurement and non-rigid deformation compensation and reducing motion artifacts is realized, and the two-dimensional medical image is processed through AI so as to provide technical support for wider medical image analysis and diagnosis.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the invention, and all equivalent structures or equivalent processes using the descriptions and drawings of the present invention or direct or indirect application in other related technical fields are included in the scope of the present invention.
Claims (10)
1. The two-dimensional medical image registration method based on artificial intelligence is characterized by comprising the following steps of:
acquiring a first two-dimensional medical image comprising CT images and/or MRI images of different modalities;
extracting image features of the first two-dimensional medical image to obtain a plurality of image features matched with the first two-dimensional medical image, and uniformly converting the plurality of image features to generate corresponding feature spaces;
performing foreground detection on moving targets in a plurality of image features in the feature space, and extracting moving areas of one or more image features;
randomly calibrating four mutually perpendicular detection points for a plurality of image features in the feature space, and carrying out real-time perpendicular monitoring on the detection points in the feature space through the motion area;
acquiring a second two-dimensional medical image, wherein the second two-dimensional medical image is a two-dimensional medical image acquired after the first two-dimensional medical image;
and loading the second two-dimensional medical image into a feature space for artifact reduction processing and non-rigid deformation compensation processing, and outputting the processed second two-dimensional medical image.
2. The two-dimensional medical image registration method based on artificial intelligence according to claim 1, wherein the step of loading the second two-dimensional medical image into a feature space for artifact reduction processing comprises:
according to the blood vessel CT image in the second two-dimensional image, carrying out gray-scale setting on the characteristic space, and identifying blood information of the blood vessel CT image through a motion area of the characteristic space;
performing ROI (region of interest) compression on the blood information in the characteristic space after the graying setting to acquire a blood vessel wall image and a black blood image positioned in the blood vessel wall;
and performing blood vessel path matching on the blood vessel wall image and the black blood image and the image features in the feature space, and deleting the part, exceeding the image feature blood vessel path, of the blood vessel wall image and the black blood image.
3. The two-dimensional medical image registration method based on artificial intelligence according to claim 2, wherein after the step of performing gray-scale setting on the feature space and recognizing blood information of the blood vessel CT image through a motion region of the feature space according to the blood vessel CT image in the second two-dimensional image, comprising:
determining a plurality of image features matched with blood information from the feature space;
performing an interest setting on a plurality of image features to determine out-of-interest blood information in the blood information through the interest setting;
and deleting the image features corresponding to the blood information outside the interest in the feature space.
4. The two-dimensional medical image registration method based on artificial intelligence of claim 1, wherein the step of loading the second two-dimensional medical image into a feature space for artifact reduction processing further comprises:
acquiring other CT images in the second two-dimensional medical image, including but not limited to, a skeleton, a heart, and other organs;
inputting the other CT images into a feature space, and aligning the motion areas of the other CT images in the feature space;
and deleting the image parts of the other CT images exceeding the preset displacement by carrying out displacement presetting on the motion area.
5. The artificial intelligence based two-dimensional medical image registration method of claim 1, wherein loading the second two-dimensional medical image into a feature space for non-rigid deformation compensation processing comprises:
performing deformation field calculation on a second two-dimensional medical image in the feature space by using a B-spline transformation algorithm, and generating a point cloud set matched with the second two-dimensional medical image;
obtaining a difference value of the compared image features and the point cloud set through the comparison of the point cloud set and the image features in the feature space;
and correspondingly compensating and correcting the second two-dimensional medical image according to the difference value.
6. The two-dimensional medical image registration method based on artificial intelligence according to claim 1, wherein the step of performing image feature extraction on the first two-dimensional medical image to obtain a plurality of image features matched with the first two-dimensional image, and performing unified transformation on the plurality of image features to generate a corresponding feature space comprises:
performing identification extraction on the first two-dimensional medical image by adopting a SIFT feature extraction algorithm and a SURF feature extraction algorithm, and respectively identifying a corresponding first feature group and a corresponding second feature group;
matching the first feature set and the second feature set by using a FLANN matching method to match a plurality of approximate features in the first feature set and the second feature set;
coordinate conversion and normalization processing are carried out on the approximate features serving as image features, so that a plurality of image features are generated into corresponding feature spaces;
and connecting a plurality of image features of the feature space with a subsequent image transmission interface to acquire subsequently input two-dimensional medical image data.
7. The two-dimensional medical image registration method based on artificial intelligence according to claim 1, wherein the step of foreground detecting a moving object in a number of image features in the feature space and extracting a moving region of one or more image features comprises:
identifying motion vectors for a plurality of image feature pairs in the feature space;
identifying moving targets and speeds of a plurality of the motion vectors through a Lucas-Kanade algorithm;
performing optical flow field threshold processing on the calculated moving targets and speeds to filter motion vectors lower than a set threshold;
utilizing switching operation to eliminate isolated noise points in a plurality of motion vectors;
and connecting a plurality of motion areas corresponding to the motion vectors.
8. The two-dimensional medical image registration method based on artificial intelligence according to claim 1, wherein the step of randomly calibrating four mutually perpendicular detection points for a plurality of image features in the feature space and vertically monitoring the detection points in the feature space in real time through the motion area comprises the steps of:
continuously acquiring a first two-dimensional medical image and converting the first two-dimensional medical image into image features which are input into a feature space in a one-to-one correspondence manner;
presetting four points which are vertical, and marking the four points on the image features of the feature space;
monitoring whether the angles of connecting lines of the four points change in real time.
9. The artificial intelligence based two-dimensional medical image registration method according to claim 8, wherein the step of loading the second two-dimensional medical image into a feature space for artifact reduction processing and non-rigid deformation compensation processing, and outputting the processed second two-dimensional medical image comprises:
and if the image characteristics in the characteristic space are changed, the image characteristics are determined to be subjected to artifact reduction processing and non-rigid deformation compensation processing, and then corresponding neutralization behaviors are generated, and the generation process of the characteristic space is re-executed according to the determination, wherein the neutralization behaviors are generated by carrying out mean value calibration on the image characteristics after the second two-dimensional medical image is input.
10. An artificial intelligence based two-dimensional medical image registration system comprising a computer system, the computer device performing the artificial intelligence based two-dimensional medical image registration method of any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310828003.3A CN116777962A (en) | 2023-07-07 | 2023-07-07 | Two-dimensional medical image registration method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310828003.3A CN116777962A (en) | 2023-07-07 | 2023-07-07 | Two-dimensional medical image registration method and system based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116777962A true CN116777962A (en) | 2023-09-19 |
Family
ID=88008070
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310828003.3A Pending CN116777962A (en) | 2023-07-07 | 2023-07-07 | Two-dimensional medical image registration method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116777962A (en) |
-
2023
- 2023-07-07 CN CN202310828003.3A patent/CN116777962A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| AU2017292642B2 (en) | System and method for automatic detection, localization, and semantic segmentation of anatomical objects | |
| US10614573B2 (en) | Method for automatically recognizing liver tumor types in ultrasound images | |
| JP6058093B2 (en) | Computer-aided analysis device for medical images and computer program for medical image analysis | |
| US9980704B2 (en) | Non-invasive image analysis techniques for diagnosing diseases | |
| Chu et al. | Ultrasonic thyroid nodule detection method based on U-Net network | |
| CN108171738B (en) | Multi-modal medical image registration method based on brain function template | |
| US20130223711A1 (en) | Maching Learning Techniques for Pectoral Muscle Equalization and Segmentation in Digital Mammograms | |
| Bourbakis | Detecting abnormal patterns in WCE images | |
| CN113298830A (en) | Acute intracranial ICH region image segmentation method based on self-supervision | |
| Kishore et al. | Automatic stenosis grading system for diagnosing coronary artery disease using coronary angiogram | |
| US20240020823A1 (en) | Assistance diagnosis system for lung disease based on deep learning and assistance diagnosis method thereof | |
| Wijata et al. | Unbiased validation of the algorithms for automatic needle localization in ultrasound-guided breast biopsies | |
| CN114332910A (en) | Human body part segmentation method for similar feature calculation of far infrared image | |
| CN111466952B (en) | Real-time conversion method and system for ultrasonic endoscope and CT three-dimensional image | |
| CN116580033B (en) | Multi-mode medical image registration method based on image block similarity matching | |
| CN116309647B (en) | Method for constructing craniocerebral lesion image segmentation model, image segmentation method and device | |
| Isinkaye et al. | Segmentation of medical X-ray bone image using different image processing techniques | |
| CN116777962A (en) | Two-dimensional medical image registration method and system based on artificial intelligence | |
| CN115294329A (en) | Focus image identification method based on deep learning model | |
| Noviana et al. | Axial segmentation of lungs CT scan images using canny method and morphological operation | |
| JP2022052210A (en) | Information processing device, information processing method, and program | |
| CN113689469A (en) | Method for automatically identifying ultrasonic contrast small liver cancer focus and ultrasonic system | |
| Wang et al. | Improving segmentation of breast arterial calcifications from digital mammography: good annotation is all you need | |
| Salehi et al. | Investigation and simulation of different medical image processing algorithms to improve image quality using simulink matlab | |
| CN111640127A (en) | Accurate clinical diagnosis navigation method for orthopedics department |
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 |