CN113838068B - Automatic segmentation method, device and storage medium for myocardial segments - Google Patents
Automatic segmentation method, device and storage medium for myocardial segments Download PDFInfo
- Publication number
- CN113838068B CN113838068B CN202111134915.8A CN202111134915A CN113838068B CN 113838068 B CN113838068 B CN 113838068B CN 202111134915 A CN202111134915 A CN 202111134915A CN 113838068 B CN113838068 B CN 113838068B
- Authority
- CN
- China
- Prior art keywords
- left ventricular
- segmentation
- ventricular wall
- wall
- segments
- 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.)
- Active
Links
- 230000011218 segmentation Effects 0.000 title claims abstract description 132
- 238000000034 method Methods 0.000 title claims abstract description 56
- 230000002107 myocardial effect Effects 0.000 title claims abstract description 44
- 230000002861 ventricular Effects 0.000 claims abstract description 109
- 238000005192 partition Methods 0.000 claims description 26
- 210000005240 left ventricle Anatomy 0.000 claims description 17
- 210000005241 right ventricle Anatomy 0.000 claims description 15
- 238000003384 imaging method Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 2
- 230000000747 cardiac effect Effects 0.000 abstract description 11
- 238000012545 processing Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 9
- 238000000638 solvent extraction Methods 0.000 description 9
- 238000004891 communication Methods 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000003759 clinical diagnosis Methods 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 210000004165 myocardium Anatomy 0.000 description 4
- 230000003068 static effect Effects 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 208000031225 myocardial ischemia Diseases 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 208000019622 heart disease Diseases 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 230000003902 lesion Effects 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 208000010125 myocardial infarction Diseases 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000002604 ultrasonography Methods 0.000 description 2
- 208000024172 Cardiovascular disease Diseases 0.000 description 1
- 206010061216 Infarction Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000013170 computed tomography imaging Methods 0.000 description 1
- 208000029078 coronary artery disease Diseases 0.000 description 1
- 210000004351 coronary vessel Anatomy 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000007574 infarction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000008719 thickening Effects 0.000 description 1
- 230000007704 transition 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
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- 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
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The disclosure relates to an automatic segmentation method, device and storage medium for myocardial segments, wherein the automatic segmentation method comprises the steps of obtaining a 3D image of a heart, carrying out 3D segmentation on a left ventricular wall by using a first 3D learning network based on the 3D image of the heart and determining a long axis of the left ventricular wall, dividing the left ventricular wall into four circumferential extension parts along the long axis and sequentially comprising a base part, a middle part, a near cardiac apex and a cardiac apex, obtaining a first slice image and a second slice image from corresponding positions of the 3D image of the heart based on position information of a junction of the base part and the middle part and a junction of the middle part and the near cardiac apex, determining at least one key point in the first slice image and at least one key point in the second slice image by using a second learning network, and respectively dividing four circumferential extension parts of the left ventricular wall as junction points of at least two segments of the base part and the middle part and junction points of at least two segments of the near cardiac apex so as to obtain 17 segments meeting AHA17 segment segmentation standards.
Description
Technical Field
The present disclosure relates to the field of image processing, and more particularly, to an automatic segmentation method, apparatus, and storage medium for myocardial segments.
Background
In medical clinical diagnosis, when myocardial ischemia is caused by coronary heart disease or the like, abnormal segmental motion of the wall of the heart occurs, and the left wall of the heart needs to be artificially segmented so as to judge the myocardial ischemia or the infarct position and range, so that the positioning of myocardial infarction is based on the left wall segmentation. The myocardial partition is very important for clinical diagnosis of heart diseases, doctors can accurately position abnormal parts according to the left ventricular partition, deduce the positions of lesion coronary arteries, and further diagnose and treat lesion areas, which is a precondition for treating heart diseases such as myocardial ischemia, myocardial infarction and the like.
The traditional myocardial partitioning method firstly requires medical image professionals to manually outline myocardial slice boundaries for left ventricular segmentation, and then positions key parts of AHA17 segment partitioning on the basis of the left ventricular segmentation, so that the labor cost is high. Meanwhile, the traditional manual partitioning method has the defects of high time cost, high subjectivity and easy misdiagnosis caused by errors, and the requirements of real-time and accuracy of diagnosis and treatment of patients are hardly met. Moreover, in the current public myocardial partitioning algorithm, no full-automatic partitioning algorithm has been proposed yet.
Disclosure of Invention
The present disclosure is provided to solve the above-mentioned problems occurring in the prior art.
There is a need for an automatic segmentation method of myocardial segments, which can perform complete 3D segmentation on a left ventricular wall of a 3D image of a heart by using a deep learning network and determine a long axis of the left ventricular wall, further determine position information of key points in the first slice image and the second slice image by using the learning network as boundary points among segments of the base, the middle and the near-apex, and perform full-automatic segmentation of 17 segments conforming to an AHA17 segment segmentation standard on four circumferential extensions of the left ventricular wall by using the determined boundary points on the basis of dividing the left ventricular wall into the base, the middle, the near-apex and the apex along the long axis. The method can provide full-automatic myocardial segmentation of the 3D heart image, which meets AHA17 segment segmentation standards and is controllable in process, with higher real-time performance, accuracy and robustness, so that the timeliness and accuracy of clinical diagnosis are greatly improved, and the diagnosis efficiency and treatment effect are improved.
According to a first scheme of the disclosure, an automatic segmentation method of a myocardial segment is provided, the automatic segmentation method comprises the steps of obtaining a 3D image of a heart, obtaining, by a processor, a first slice image and a second slice image of corresponding positions from the 3D image of the heart based on the 3D image of the heart by using a first 3D learning network, determining a long axis of the left ventricular wall based on 3D segmentation information of the left ventricular wall, dividing, by the processor, the left ventricular wall into four circumferentially extending parts along the long axis, sequentially comprising a base part, a middle part, a proximal apex and a apex, dividing, by the processor, the first slice image and the second slice image of corresponding positions from the 3D image of the heart based on first slice position information of a junction of the base part and the middle part and second slice position information of the proximal apex in the 3D segmentation information of the left ventricular wall by using at least one second learning network, determining position information of at least one key point in the first slice image and at least one key point in the second slice image by using the junction of at least one segment as a key point in the first slice image and obtaining, by using the at least one segment of the two segments in the first slice image and the second segment as a key point in the first segment and the second segment, and obtaining a critical point in the first segment and the second segment is segmented by using the at least two key point as a key point in the first segment and the second segment is obtained by the at least the first segment and the second segment is obtained.
According to a second aspect of the present disclosure, an automatic segmentation apparatus of a myocardial segment is provided, the automatic segmentation apparatus comprising an interface configured to receive a 3D image of a heart acquired by an imaging apparatus, and a processor configured to perform an automatic segmentation method of a myocardial segment according to various embodiments of the present disclosure.
According to a third aspect of the present disclosure, there is provided a computer storage medium having stored thereon executable instructions which when executed by a processor implement the steps of the above-described method of automatic segmentation of myocardial segments.
With the automatic segmentation method, apparatus and storage medium of myocardial segments according to various embodiments of the present disclosure, it is possible to perform complete 3D segmentation of the left ventricular wall of a 3D image of a heart and determine the long axis of the left ventricular wall using a deep learning network, further based on a first slice image and a second slice image of the junction of the base and the middle and the junction of the middle and the apex of the heart on the basis of dividing the left ventricular wall into the base, the middle, the apex of the heart along the long axis, determine position information of key points in the first slice image and the second slice image using a learning network as junction points between the segments of the base, the middle and the apex of the heart, and perform full automatic segmentation of 17 segments conforming to the AHA17 segment segmentation standard for four circumferential extensions of the left ventricular wall using the determined junction points. The method can provide full-automatic myocardial segmentation of the 3D heart image which accords with AHA17 segment segmentation standard and has higher real-time performance, accuracy and robustness and controllable whole process, thereby greatly improving the timeliness and accuracy of clinical diagnosis and improving the diagnosis efficiency and treatment effect.
Drawings
In the drawings, which are not necessarily drawn to scale, like numerals may describe similar components in different views. The accompanying drawings illustrate various embodiments by way of example in general and not by way of limitation, and together with the description and claims serve to explain the disclosed embodiments. Wherever possible, the same reference numbers will be used throughout the drawings to refer to the same or like parts. Such embodiments are illustrative and not intended to be exhaustive or exclusive of the present apparatus or method.
Fig. 1 shows a flow chart of a method of automatic segmentation of a myocardial segment in accordance with an embodiment of the present disclosure.
Fig. 2 shows a schematic diagram of segmentation of the long axis of the left ventricular wall based on 3D segmentation information of the left ventricular wall according to an embodiment of the present disclosure.
Fig. 3 shows a schematic diagram of 17-segment partitioning of myocardium according to an AHA 17-segment partitioning criteria, according to an embodiment of the present disclosure.
Fig. 4 shows a schematic diagram of 17-segment segmentation of a left ventricular wall using keypoint location information in accordance with an AHA 17-segment segmentation standard, in accordance with an embodiment of the present disclosure.
Fig. 5 shows a schematic diagram of the composition of an automatic segmentation apparatus of myocardial segments in accordance with an embodiment of the present disclosure.
Detailed Description
In order to better understand the technical solutions of the present disclosure, the following detailed description of the present disclosure is provided with reference to the accompanying drawings and the specific embodiments. Embodiments of the present disclosure will be described in further detail below with reference to the drawings and specific embodiments, but not by way of limitation of the present disclosure. The order in which the steps are described herein by way of example should not be construed as limiting if there is no necessity for a relationship between each other, and it should be understood by those skilled in the art that the steps may be sequentially modified without disrupting the logic of each other so that the overall process is not realized.
Fig. 1 shows a schematic diagram of a flow of an automatic segmentation method of myocardial segments according to an embodiment of the present disclosure.
In step S101 of fig. 1, a 3D image of the heart is first acquired, and in some embodiments, a 3D image of the heart may be acquired, for example, by medical contrast imaging techniques such as CT, MR, radionuclide scanning, and ultrasound. In other embodiments, a 3D image of the heart may also be acquired from a 3D heart image database, without limitation.
In step S102, a 3D segmentation of the left ventricular wall is performed by the processor using the first 3D learning network based on the acquired 3D image of the heart.
In some embodiments, the first 3D learning network may be a 3D-UNet segmentation network. The 3D-UNet deep learning network adopts a convolutional neural network model, so that a better image segmentation effect can be achieved with less training data. In this example, the 3D-UNet segmentation network may be trained using a training set composed of multiple 3D cardiac images of multiple cases, and the image data may be manually labeled, or may be automatically labeled using a Graph Cut or other method, which is not limited herein. After the 3D-UNet segmentation network training is completed, a 3D image of the heart can be input, and the 3D segmentation network is used for carrying out 3D segmentation on the left ventricular wall. In other embodiments, the first 3D learning network may be another type of deep learning network, which is not particularly limited herein.
In step S103, the long axis of the left ventricular wall is determined by the processor based on the 3D segmentation information of the left ventricular wall. The determination of the long axis of the left ventricular wall based on the 3D segmentation information of the left ventricular wall may take a variety of different forms, and in some embodiments, a 3D ellipsoid fitting calculation may be performed using a least squares method to determine coefficients of an ellipsoid equation corresponding to the 3D segmentation of the left ventricular wall, and determine a feature vector of the corresponding ellipsoid based on the coefficients of the ellipsoid equation, thereby determining the direction of the long axis of the left ventricular wall.
In other embodiments, principal Component Analysis (PCA) may also be performed based on the 3D segmentation information of the left ventricular wall to determine a feature vector with the largest feature value, and determine a direction corresponding to the feature vector with the largest feature value as a direction of a long axis of the left ventricular wall. In other embodiments, machine learning or deep learning methods may also be employed to further feature the 3D image of the heart based on the 3D segmentation information of the left ventricular wall to determine the long axis of the left ventricular wall.
In other embodiments, after determining the direction of the long axis of the left ventricular wall, the long axis of the left ventricular wall may be rotated to a position parallel to the specified coordinate axis based on the long axis direction and the position of the center of sphere of the ellipsoid calculated using the ellipsoid fitting algorithm described above, and further segmental division may be performed at that angle.
In the process of 3D segmentation of the left ventricular wall according to the embodiments of the present disclosure as described in steps S101-S103, the 3D segmentation of the left ventricular wall using the first 3D learning network, such as the 3D-UNet deep learning network, for the 3D image of the heart may effectively suppress noise in the CT image, such as the heart, and the left ventricular region may not be significantly affected by the background contrast, etc., as compared to the manual delineating and segmentation of the slice boundary by a doctor, thereby performing a faster and more accurate 3D segmentation of the left ventricular wall.
In step S104, the left ventricular wall is divided by the processor into four circumferential extensions along the long axis, sequentially including a base, a middle, a proximal apex, and a apex.
In some embodiments, the left ventricular wall may be further divided along the direction of the long axis into four circumferential extensions, comprising, in order, a base, a middle, a proximal apex, and a apex. The above-described division into preliminary divisions according to the AHA17 segment division criteria, further segment divisions in the short axis direction will be accomplished in the following steps with a method according to an embodiment of the present disclosure.
In step S105, the processor obtains, from the 3D image of the heart, a first slice image and a second slice image corresponding to the positions based on the first slice position information at the junction between the base and the middle and the second slice position information at the junction between the middle and the apex of the heart in the 3D segmentation information of the left ventricular wall, and determines, using at least one second learning network, the position information of at least one key point in the first slice image and the position information of at least one key point in the second slice image.
In some embodiments, the first slice image and the second slice image may be acquired by first rotating the 3D image of the heart in a manner that includes both the center point and the long axis direction that are used to coincide with the rotation of the 3D segmentation of the left ventricular wall. Next, in some embodiments, according to the division of the base, the middle, the apex and the apex determined in step S104, the number of layers corresponding to the first slice at the junction of the base and the middle and the number of layers corresponding to the second slice at the junction of the middle and the apex may be acquired, and slice images of the corresponding layers may be respectively used as the first slice image and the second slice image. In actual operation, two corresponding 2D slices may also be directly indexed from the three-dimensional matrix corresponding to the rotated 3D image of the heart.
In other embodiments, when any plane slicing can be performed on the 3D image of the heart, slicing can also be performed at exact coordinate positions corresponding to the junction between the base and the middle of the rotated 3D image of the heart and the junction between the middle and the apex of the heart, and the acquired 2D image is used as the first slice image and the second slice image respectively.
In some embodiments, the at least one keypoint in the first slice image comprises a first intersection point and a second intersection point between the right ventricle and the left ventricle. In other embodiments, the at least one keypoint in the second slice image comprises a third and fourth intersection between the right ventricle and the left ventricle.
In some embodiments, the second learning network may include at least one of a probabilistic graph-based UNet network and a coordinate regression-based CNN network, and the second learning network is a single learning network.
In step S106, the processor segments the four circumferential extensions of the left ventricular wall using at least one keypoint of the first slice image as an intersection of at least two segments of the base and middle, respectively, and using at least one keypoint of the second slice image as an intersection of at least two segments of the proximal apex, to obtain 17 segments meeting the AHA17 segment segmentation criteria.
In some embodiments, a first intersection between the right ventricle and the left ventricle is utilized as the intersection of the anterior wall segment and the anterior partition segment of the base, doubling as the intersection of the anterior wall segment and the anterior partition segment of the middle.
In some embodiments, a second intersection between the right ventricle and the left ventricle is utilized as an intersection of the posterior wall segment of the base and the posterior partition segment, doubling as an intersection of the medial posterior wall segment and the posterior partition segment.
In some embodiments, a third junction between the right ventricle and the left ventricle is utilized as the junction of the anterior wall segment and the partition wall segment of the proximal apex, and a fourth junction between the right ventricle and the left ventricle is utilized as the junction of the posterior wall segment and the partition wall segment of the proximal apex.
In some embodiments, the three circumferential extensions of the left ventricular wall other than the apex are segmented based on the respective junctions.
In some embodiments, when dividing the three circumferential extensions of the left ventricular wall other than the apex based on the respective junctions, different manners may be employed, such as, for example, extending the respective junctions to the center point on the corresponding long axis tangent plane of the left ventricular wall for the respective circumferential extensions other than the apex, and equally dividing the sets of segments divided by the respective junction to the center point line extension. In other embodiments, the base and middle may also be extended by connecting each junction with a center point on the corresponding long axis section of the left ventricular wall, dividing the set of anterior and posterior partition segments divided by the connection extension of each junction with the center point to obtain a divided line, sequentially rotating the obtained divided line 120 degrees twice, and extending the obtained divided line with the center point, while for the proximal apex, dividing the set of anterior and lateral wall segments divided by the connection extension of each junction with the center point and the set of partition and posterior wall segments respectively.
In some embodiments, after segmentation of the four circumferential extensions of the left ventricular wall to obtain 17 segments meeting the AHA17 segment segmentation criteria, the resulting segmentation of 17 segments may be rotated back to the original direction of the long axis of the left ventricular wall.
Fig. 2 shows a schematic diagram of segmentation of the long axis of the left ventricular wall based on 3D segmentation information of the left ventricular wall according to an embodiment of the present disclosure. Fig. 3 shows a schematic diagram of 17-segment partitioning of myocardium according to an AHA 17-segment partitioning criteria, according to an embodiment of the present disclosure. A specific method of segmenting the long axis of the left ventricular wall is described below in conjunction with fig. 2 and 3.
In some embodiments, after determining the long axis LA of the left ventricular wall from the 3D segmentation information of the left ventricular wall and rotating the long axis LA of the left ventricular wall to a position parallel to the specified coordinate axis by, for example, an ellipsoid fitting algorithm, the left ventricular wall may then be divided into four circumferential extensions along the direction of the long axis LA, as shown in fig. 2, including a base 201, a middle 202, a proximal Apex 203, and a Apex 204 in order, corresponding to the base (Basal), the middle (Mid-Cavity), the proximal Apex (Apex), and the Apex (Apex), respectively, in fig. 3, in accordance with the AHA17 segment segmentation standard. Segmentation of the apex 204 needs to be positioned at the apex of the left ventricular wall lining, i.e., the transition point along the segmentation long axis from the apex (i.e., the top region of the myocardium) to the left ventricular blood chamber region, corresponding to the left ventricular wall lining apex 301 in the horizontal long axis view (HLA) and vertical long axis View (VLA) in fig. 3.
In some embodiments, the positioning of the left ventricular wall inner layer vertex 301 may be acquired together when the left ventricular wall is 3D segmented using the first 3D learning network, or the long axis of the left ventricular wall is determined based on the 3D segmentation information of the left ventricular wall. In other embodiments, other methods of locating the left ventricular wall inner layer apex 301 may be used, without limitation.
In some embodiments, after the determination of the left ventricular wall inner layer apex 301, the myocardium is bisected with a segmentation plane perpendicular to the LA of the left ventricular wall and passing through the point, and the top-layer segmentation, i.e., the apex 204 in fig. 2, and the remaining region to be segmented is acquired. In some embodiments, the long axis LA of the left ventricular wall may be further taken as a dividing direction, the region to be divided is equally divided into three major segments along the long axis LA with two dividing planes perpendicular to the long axis LA direction, a dividing plane 205 and a dividing plane 206, and the left ventricular wall is divided into three circumferential extensions according to the segments, corresponding to the base 201, the middle 202 and the proximal apex 203 in fig. 2, respectively. Thus, the preliminary division of the left ventricular wall along the long axis LA is completed.
Fig. 4 shows a schematic diagram of 17-segment segmentation of a left ventricular wall using keypoint location information in accordance with an AHA 17-segment segmentation standard, in accordance with an embodiment of the present disclosure.
In fig. 4, the first slice image C1 is a slice image of a boundary between the base and the middle of the 3D segmentation of the left ventricular wall, the second slice image C2 is a slice image of a boundary between the middle and the apex of the near heart of the 3D segmentation of the left ventricular wall, and the position information of at least one key point in the first slice image C1 and the position information of at least one key point in the second slice image C2 are determined by using at least one second learning network. In the example of fig. 4, the first slice image C1 has two keypoints, the keypoint 401 and the keypoint 402 being a first and a second intersection between the right ventricle and the left ventricle, respectively, and similarly, there are two keypoints in the second slice image C2, the keypoint 403 and the keypoint 404 being a third and a fourth intersection between the right ventricle and the left ventricle, respectively. In other embodiments, other keypoints may also be identified based on features of the heart slice images, and the number and location of the keypoints is not limited as long as they can be used for subsequent 3D segmentation of the left ventricular wall.
In some embodiments, the second learning network includes, but is not limited to, at least one of a probabilistic graph-based UNet network and a coordinate regression-based convolutional neural network (CNN network). With CNN networks or UNet networks, high-level abstract features of cardiac 3D images can be automatically extracted without requiring the user to choose or construct potentially useful features based on expertise and long-term experience. In this example, the number of layers of the 3D-UNet segmentation network may be 4 layers, which are commonly used, or may be extended to 8 layers or more, so that deeper 3D image features may be learned, and the segmentation accuracy of the segmentation network may be improved. In some embodiments, location acquisition of key points required for segmentation may also be implemented based on heatMap heat map-assisted methods of UNet networks. The UNet network combined with heatMap heat maps can be used for realizing end-to-end image segmentation, has higher processing resolution in the details of the images, and can further improve the precision of acquiring key points, thereby improving the precision of segment segmentation of the left ventricular wall based on the position information of the key points.
In some embodiments, the second learning network used to determine keypoints in the first slice image C1 and the second slice image C2 may share a single learning network. In other embodiments, for example, to achieve faster key point recognition speed, parallel operations of different learning networks may also be implemented.
In some embodiments, before the second learning network is used to detect the positions of the keypoints, the second learning network may be trained by using a training data set, for example, manually calibrating one or more keypoints to be identified in a cardiac image slice sample or by using a labeling tool, determining whether the probability of the keypoint is determined in a neighborhood around the keypoint according to the distance from the keypoint, wherein the probability of the non-neighborhood is zero, and generating a matrix of dimensions and image peering, wherein the value of each element is the probability of whether the keypoint. Therefore, the trained second learning network detects a probability matrix for each specific image input to be detected, and obtains the position corresponding to the maximum probability value, namely the position of the key point. In other embodiments, the second learning network may be trained using other training methods, without limitation.
After determining the keypoints in the first slice image C1 and the second slice image C2, it is then used for further segmentation of the left ventricular wall. As can be seen from fig. 4, according to the AHA17 segment segmentation criteria, the base PB and the middle PM each include 6 segments and are scaled close, so that in some embodiments, the key point 401 corresponding to the first intersection point between the right ventricle and the left ventricle in the first slice image C1 may be regarded as the intersection point of the base anterior wall segment 1 and the base anterior partition wall segment 2, and at the same time, also as the intersection point of the middle anterior wall segment 7 and the middle anterior partition wall segment 8 of the middle PM. Similarly, the key point 402 corresponding to the second junction between the right ventricle and the left ventricle may be regarded as the junction of the base rear partition section 3 and the base rear wall section 4 of the base PB, and at the same time, as the junction of the middle rear partition section 9 and the middle rear wall section 10 of the middle PM.
In the example of fig. 4, the second slice image C2 is a slice image of the intersection of the middle PM and the apex PA, in which in some embodiments a keypoint 403 corresponding to a third intersection between the right ventricle and the left ventricle and a keypoint 404 corresponding to a fourth intersection may be used for the division of the individual segments in the apex PA, e.g., the keypoint 403 may be taken as the intersection of the apex anterior wall segment 13 and the apex posterior wall segment 14, the keypoint 404 may be taken as the intersection of the apex anterior wall segment 14 and the apex posterior wall segment 15, and thus the boundary of the apex anterior wall segment 14 may be determined.
Next, the three circumferential extensions of the left ventricular wall other than the apex may be segmented according to the respective junctions of the base PB, middle PM and proximal apex PA that have been determined above.
In some embodiments, when dividing the three circumferential extensions of the left ventricular wall other than the apex based on each junction, different manners may be employed, such as, for example, each junction may be wired to a center point (not shown) on a corresponding long-axis tangential plane (not shown) of the left ventricular wall, and each set of segments divided by the wired extension of each junction to the center point may be equally divided. In other embodiments, for the base PB and the middle PM, the respective junctions may also be extended with respect to the central point on the corresponding long axis section of the left ventricular wall, the groups of the anterior and posterior partition segments divided by the extension of the respective junctions with respect to the central point may be equally divided to obtain average lines, the obtained average lines may be sequentially rotated 120 degrees twice, and the obtained average lines may be extended with respect to the central point, while for the proximal apex, the respective junctions may be extended with respect to the central point on the corresponding long axis section of the left ventricular wall, and the groups of the anterior and lateral wall segments and the groups of the partition and the posterior wall segments divided by the extension of the respective junctions with respect to the central point may be equally divided, thereby obtaining 17 segments conforming to the AHA17 segment division standard.
For the sake of clarity, the determination method of the respective segment boundaries in the first slice image C1 and the second slice image C2 is further described below in connection with fig. 4 further from the viewpoint of the cut plane.
The segment division of the base PB and the middle PM will be described taking the middle PM as an example. In some embodiments, when the first slice image C1 has two key points, i.e., the key point 401 and the key point 402, an intersection point between the long axis (not shown in fig. 4) and the plane of the first slice image C1 may be first determined, as a center point CP1 of the slice of the first slice image C1, the key point 401 and the key point 402 are respectively connected with the center point CP1, and the included angles formed by the three points are equally divided, and an intersection point 405 with the left ventricular wall is determined, so that a range of the middle anterior partition wall segment 8 is between the key point 401 and the intersection point 405, and a range of the middle posterior partition wall segment 9 is between the key point 402 and the intersection point 405. Next, the lines of the keypoints 401, 402, and 405 with the center point CP1 may be extended to generate the boundary points 406, 407, and 408 with the edges of the first slice image C1, respectively, whereby the ranges of 6 segments of the middle PM are determined.
In other embodiments, after the junctions 405 and 408 are determined as described above, the extension lines passing through the junctions 405, CP1, and 408 may be sequentially rotated 120 degrees twice to obtain the boundary points of the intermediate back wall segment 10 and the intermediate front side wall segment 11, and the boundary points of the intermediate front wall segment 7 and the intermediate back side wall segment 12. In other embodiments, when there is only one key point, for example, the key point 401, in the first slice image C1, a section (not shown) in the long axis horizontal direction may be made through the key point 401, and the intersection point of the section on the other side of the center point CP1 and the left ventricular wall in the first slice image C1 may be made as the intersection point of the middle back wall section 10 and the middle front side wall section 11, and further, the section is sequentially rotated about the long axis for 120 degrees, for example, the first rotation in the counterclockwise direction for 120 degrees may obtain the intersection point of the middle back wall section 9 and the middle back wall section 10, and the boundary point of the middle front wall section 7 and the middle back side wall section 12, and the second rotation in the counterclockwise direction for 120 degrees may obtain the intersection point of the middle front wall section 11 and the middle back wall section 12, and the boundary point of the middle front wall section 8 and the middle back wall section 9. The determination of the intersection points of the various segments of base PB may take the form of methods including, but not limited to, methods consistent with the examples of midsection PM described above, the particular methods being not limited herein.
In some embodiments, the intersection of the boundary points of the two ends of the apex-proximal partition wall segment 14 and the center point CP2 (the intersection of the long axis and the tangent of the second slice image C2) on the other side of the center point may be used as the intersection 409 of the apex-proximal anterior wall segment 13 and the apex-proximal sidewall segment 16, and the intersection 410 of the apex-proximal posterior wall segment 15 and the apex-proximal sidewall segment 16, in a manner similar to the above-described exemplary method of determining the intermediate PM segment boundary point. In other embodiments, for example, when there is only one key point 403 in the second slice image C2, the line between the key point 403 and the center point CP2 may be extended, and rotated 90, so as to obtain the boundary points of 4 segments. In other embodiments, other ways of determining the interface between the segments of the proximal apex PA may be used, and the particular method is not limited.
According to the AHA 17 segment segmentation criteria, the apex is the independent 17 th segment of the left ventricular wall, and therefore no further segmentation is required.
In some embodiments, after segmentation of the four circumferential extensions of the left ventricular wall to obtain a 17-segment segmentation result according to the AHA 17-segment segmentation criteria as shown in H1 in fig. 4, the resulting segmentation result of the 17-segment according to the AHA 17-segment segmentation criteria may be rotationally redirected to an initial direction of the long axis of the left ventricular wall to obtain a segmentation result H2 consistent with the initial direction.
With the automatic segmentation method of the myocardial segments according to the embodiments of the present disclosure, particularly, by detecting one or more key points of the same type in different slice images, namely, detecting a first intersection point and a second intersection point between a right ventricle and a left ventricle in the slice in a first slice image C1 at the junction of a base and a middle part, and detecting a third intersection point and a fourth intersection point between the right ventricle and the left ventricle in the slice in a second slice image C2 at the junction of the middle part and a proximal apex, it is possible to provide a good key point position for segmenting the myocardial segments in the off-axis direction for each part in the long axis direction, and it is possible to conveniently perform complete myocardial segment segmentation using the detected key point position information.
Therefore, according to the automatic segmentation method of the myocardial segments, the 3D heart image can be subjected to efficient, accurate and automatic segmentation meeting the AHA 17 segment segmentation standard, and on the basis of the accurate segmentation of the myocardial segments, a doctor can perform more accurate quantitative analysis on clinical indexes such as ventricular volume, ejection fraction, left ventricular quality, wall thickening, wall motion abnormality and the like, including more efficient, accurate and reliable assessment on the functions of the whole and part of the heart, so that the diagnosis efficiency and treatment effect of cardiovascular diseases are improved.
Fig. 5 shows a schematic diagram of the composition of an automatic segmentation apparatus of myocardial segments in accordance with an embodiment of the present disclosure. In some embodiments, the automated segmentation device 500 of myocardial segments may be a dedicated smart device or a general purpose smart device. For example, the automatic segmentation apparatus 500 may be a computer customized for the automatic segmentation task of the myocardial segments, or a cloud server. For example, the automatic segmentation apparatus 500 may be integrated into an image processing apparatus.
As an example, in an automatic segmentation apparatus 500 of a myocardial segment, at least an interface 501 and a processor 503 are included, and in some embodiments, a memory 502 may also be included.
In some embodiments, interface 501 is configured to receive cardiac 3D images acquired by imaging devices, for example, interface 501 may receive cardiac 3D images acquired by various imaging devices via a communication cable, a Wireless Local Area Network (WLAN), a Wide Area Network (WAN), a wireless network, such as via radio waves, a cellular or telecommunications network, and/or a local or short range wireless network (e.g., bluetooth TM), or other communication methods.
In some embodiments, interface 501 may include an Integrated Services Digital Network (ISDN) card, a cable modem, a satellite modem, or a modem to provide a data communication connection. In such implementations, interface 501 may send and receive electrical, electromagnetic, and/or optical signals via a direct communication link, which carry analog/digital data streams representing various types of information. In other embodiments, interface 501 may also include a Local Area Network (LAN) card (e.g., an ethernet adapter) to provide a data communication connection to a compatible LAN. As an example, the interface 501 may also include a network interface 5011, via which network interface 5011 the automatic segmentation apparatus 500 may be connected to a network (not shown), such as a local area network or the internet including, but not limited to, in a hospital. The network may connect the automatic segmentation apparatus 500 with external devices such as an image acquisition apparatus (not shown), a 3D heart image database 504, an image data storage 505. The image acquisition means may be any means for acquiring an image of the object, such as an MRI imaging device, a CT imaging device, a radionuclide scan, an ultrasound device or other medical imaging device for acquiring cardiac images of a patient.
In some embodiments, the automatic segmentation apparatus 500 of myocardial segments may additionally include at least one of an input/output 506 and an image display 507.
The processor 503 is a processing device that includes one or more general-purpose processing devices, such as a microprocessor, central Processing Unit (CPU), graphics Processing Unit (GPU), and the like. More specifically, the processor 503 may be a Complex Instruction Set Computing (CISC) microprocessor, a Reduced Instruction Set Computing (RISC) microprocessor, a Very Long Instruction Word (VLIW) microprocessor, a processor executing other instruction sets, or a processor executing a combination of instruction sets. The processor 503 may also be one or more special purpose processing devices such as an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), a Digital Signal Processor (DSP), a system on a chip (SoC), or the like. As will be appreciated by those skilled in the art, in some embodiments, the processor 503 may be a special purpose processor rather than a general purpose processor. The processor 503 may include one or more known processing devices such as a Pentium TM、CoreTM、Xeon TM or Itanium TM series of microprocessors manufactured by Intel corporation, a Turion TM、AthlonTM、SempronTM、OpteronTM、FXTM、PhenomTM series of microprocessors manufactured by AMD corporation, or any of a variety of processors manufactured by Sun Microsystems. The processor 503 may also include a graphics processing unit, such as those manufactured by Nvidia corporationA series of GPUs, a GMA manufactured by intel (TM), an Iris TM series of GPUs, or a Radeon TM series of GPUs manufactured by AMD corporation. The processor 503 may also include an accelerated processing unit such as the desktop A-4 (6, 8) series manufactured by AMD corporation, the Xeon Phi TM series manufactured by Intel corporation. The disclosed embodiments are not limited to any type of processor or processor circuit that is otherwise configured to meet the computational requirements of performing an automated segmentation method, such as a myocardial segment in accordance with embodiments of the present disclosure. In addition, the term "processor" or "image processor" may include more than one processor, for example, a multi-core design or a plurality of processors, each of the plurality of processors having a multi-core design. The processor 503 may execute sequences of computer program instructions stored in the memory 502 to perform the various operations, procedures, methods disclosed herein.
The processor 503 may be communicatively coupled to the memory 502 and configured to execute computer-executable instructions stored therein. Memory 502 may include read-only memory (ROM), flash memory, random-access memory (RAM), dynamic random-access memory (DRAM) such as Synchronous DRAM (SDRAM) or Rambus DRAM, static memory (e.g., flash memory, static random-access memory), etc., upon which computer-executable instructions are stored in any format. In some embodiments, the memory 502 may store computer executable instructions of an automatic segmentation program 5021 of one or more myocardial segments. The computer program instructions may be accessed by the processor 503, read from ROM or any other suitable memory location, and loaded into RAM for execution by the processor 503. For example, memory 502 may store one or more software applications. Software applications stored in memory 502 may include, for example, an operating system (not shown) for a general purpose computer system and a soft control device (not shown). Further, the memory 502 may store the entire software application or only a portion of the software application (e.g., the automatic segmentation program 5021 of the myocardial segment) to be capable of being executed by the processor 503. Additionally, the memory 502 may store a plurality of software modules for implementing the automatic segmentation methods of myocardial segments or the various steps of a process of training a learning network for the automatic segmentation methods of myocardial segments consistent with the present disclosure. Furthermore, the memory 502 may store data generated/cached when executing the computer program, such as 3D heart image data 5022, which comprises medical images sent from an image acquisition device, 3D heart image database 504, image data storage device 505, etc.
In some embodiments, a learning network for automatic segmentation of myocardial segments may be stored in memory 502. In other embodiments, the learning network for automatic segmentation of the myocardial segments may be stored in a remote device, a separate database (such as 3D cardiac image database 504), a distributed device, and may be used by the program 5021 for automatic segmentation of myocardial segments.
The input/output 506 may be configured to allow the automatic segmentation apparatus 500 of the myocardial segment to receive and/or transmit data. The input/output 506 may include one or more digital and/or analog communication devices that allow the auto-segmentation apparatus 500 to communicate with a user or other machine and apparatus. For example, input/output 506 may include a keyboard and mouse that allow a user to provide input.
The network interface 5011 may include network adapters, cable connectors, serial connectors, USB connectors, parallel connectors, high speed data transmission adapters such as fiber optics, USB 3.0, lightning, wireless network adapters such as WiFi adapters, telecommunications (3G, 4G/LTE, etc.) adapters. The automatic segmentation apparatus 500 may be connected to a network through a network interface 5011. The network may provide the functionality of a Local Area Network (LAN), a wireless network, a cloud computing environment (e.g., software as a service, a platform as a service, an infrastructure as a service, etc.), a client server, a Wide Area Network (WAN), etc.
In addition to the automatically segmented image of the myocardial segments, the image display 507 may also display other information, such as relevant contrast size information for the individual segments, and the like. For example, the image display 507 may be an LCD, CRT, or LED display.
Embodiments of the present disclosure also provide a computer storage medium having stored thereon computer executable instructions that, when executed by a processor, implement a method of automatic segmentation of myocardial segments in accordance with the foregoing. The storage medium may include read-only memory (ROM), flash memory, random Access Memory (RAM), dynamic Random Access Memory (DRAM) such as Synchronous DRAM (SDRAM) or Rambus DRAM, static memory (e.g., flash memory, static random access memory), etc., upon which computer-executable instructions may be stored in any format.
Furthermore, although exemplary embodiments have been described herein, the scope thereof includes any and all embodiments having equivalent elements, modifications, omissions, combinations (e.g., of the various embodiments across schemes), adaptations or alterations based on the present disclosure. The elements in the claims are to be construed broadly based on the language employed in the claims and are not limited to examples described in the present specification or during the practice of the application, which examples are to be construed as non-exclusive. It is intended, therefore, that the specification and examples be considered as exemplary only, with a true scope and spirit being indicated by the following claims and their full scope of equivalents.
The above description is intended to be illustrative and not restrictive. For example, the above-described examples (or one or more aspects thereof) may be used in combination with each other. For example, other embodiments may be used by those of ordinary skill in the art upon reading the above description. In addition, in the above detailed description, various features may be grouped together to streamline the disclosure. This is not to be interpreted as an intention that the disclosed features not being claimed are essential to any claim. Rather, the disclosed subject matter may include less than all of the features of a particular disclosed embodiment. Thus, the following claims are hereby incorporated into the detailed description as examples or embodiments, with each claim standing on its own as a separate embodiment, and it is contemplated that these embodiments may be combined with one another in various combinations or permutations. The scope of the invention should be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.
The above embodiments are only exemplary embodiments of the present disclosure, and are not intended to limit the present invention, the scope of which is defined by the claims. Various modifications and equivalent arrangements of parts may be made by those skilled in the art, which modifications and equivalents are intended to be within the spirit and scope of the present disclosure.
Claims (8)
1. A method of automatic segmentation of a myocardial segment, the method comprising:
acquiring a 3D image of the heart;
Performing, by a processor, 3D segmentation of a left ventricular wall using a first 3D learning network based on a 3D image of the heart;
determining, by the processor, a long axis of a left ventricular wall based on the 3D segmentation information of the left ventricular wall;
Dividing, by the processor, a left ventricular wall into four circumferential extensions along the long axis, comprising, in order, a base, a middle, a proximal apex, and a apex;
Acquiring, by the processor, a first slice image and a second slice image of corresponding positions from the 3D image of the heart based on first slice position information of a base-middle junction and second slice position information of a middle-near-apex junction in the 3D segmentation information of the left ventricular wall, and determining position information of at least one key point in the first slice image and position information of at least one key point in the second slice image using at least one second learning network;
Dividing, by the processor, four circumferential extensions of the left ventricular wall using at least one keypoint in the first slice image as an intersection of at least two segments of the base and middle, respectively, and using at least one keypoint in the second slice image as an intersection of at least two segments of the proximal apex, to obtain 17 segments meeting an AHA17 segment division criterion;
Wherein the at least one keypoint in the first slice image comprises a first and a second intersection between the right and left ventricles and the at least one keypoint in the second slice image comprises a third and a fourth intersection between the right and left ventricles, and
The dividing the four circumferential extensions of the left ventricular wall by using at least one key point in the first slice image as a junction point of at least two sections of the base and the middle part, using at least one key point in the second slice image as a junction point of at least two sections of the proximal apex part specifically comprises dividing three circumferential extensions of the left ventricular wall except the apex part by using the first junction point as a junction point of a front wall section and a front partition wall section of the base, doubling as a junction point of a front wall section and a front partition wall section of the middle part, using the second junction point as a junction point of a rear wall section and a rear partition wall section of the base, doubling as a junction point of a rear wall section and a rear partition wall section of the middle part, using the third junction point as a junction point of a front wall section and a partition wall section of the proximal apex part, and using the fourth junction point as a junction point of a rear wall section and a partition wall section of the proximal apex part, and dividing the three circumferential extensions of the left ventricular wall except the apex part based on the respective junction points.
2. The automatic segmentation method according to claim 1, wherein determining the long axis of the left ventricular wall based on the 3D segmentation information of the left ventricular wall specifically comprises:
based on the 3D segmentation information of the left ventricle wall, performing 3D ellipsoid fitting calculation by using a least square method to determine coefficients of an ellipsoid equation corresponding to the 3D segmentation of the left ventricle wall;
based on the eigenvector represented by the coefficient, the direction of the long axis of the left ventricular wall is determined.
3. The automatic segmentation method as set forth in claim 1, further comprising:
Rotating the long axis of the left ventricular wall to a position parallel to a designated coordinate axis before dividing the left ventricular wall into four circumferential extensions along the long axis;
After segmenting the four circumferential extensions of the left ventricular wall to obtain 17 segments meeting the AHA17 segment segmentation criteria, the resulting segmentation results of the 17 segments are rotated back to the original direction of the long axis of the left ventricular wall.
4. The automatic segmentation method according to claim 1, wherein the segmentation of the three circumferential extensions of the left ventricular wall except for the apex based on the respective junctions specifically comprises:
For each circumferential extension except for the apex, connecting and extending each junction point with the central point on the corresponding long axis section of the left ventricular wall, and equally dividing each group of segments divided by the connecting and extending between each junction point and the central point, or
For the base part and the middle part, connecting each junction point with a central point on a corresponding long axis section of the left ventricular wall for extending, equally dividing groups of front dividing wall segments and rear dividing wall segments which are divided by connecting each junction point with the central point for extending to obtain average lines, sequentially rotating the obtained average lines for 120 degrees twice, and extending the obtained average lines with the connecting lines of the central points;
For the proximal apex, each junction point is extended with a central point on the corresponding long axis section of the left ventricular wall, and the groups of the anterior wall section and the lateral wall section and the groups of the partition wall section and the posterior wall section, which are divided by the extension of the lines of each junction point with the central point, are equally divided.
5. The automatic segmentation method according to claim 1, characterized in that,
The first 3D learning network comprises a 3D-UNet segmentation network, and/or
The second learning network includes at least one of a probabilistic graph-based UNet network and a coordinate regression-based CNN network.
6. The automatic segmentation method according to claim 5, wherein the second learning network is single.
7. An automatic segmentation apparatus for myocardial segments, the apparatus comprising:
An interface configured to receive a 3D image of the heart acquired by the imaging device;
a processor configured to perform the method of automatic segmentation of myocardial segments according to any one of claims 1-6.
8. A computer storage medium having stored thereon executable instructions, which when executed by a processor, implement the method of automatic segmentation of myocardial segments as claimed in any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111134915.8A CN113838068B (en) | 2021-09-27 | 2021-09-27 | Automatic segmentation method, device and storage medium for myocardial segments |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111134915.8A CN113838068B (en) | 2021-09-27 | 2021-09-27 | Automatic segmentation method, device and storage medium for myocardial segments |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113838068A CN113838068A (en) | 2021-12-24 |
| CN113838068B true CN113838068B (en) | 2024-12-03 |
Family
ID=78970598
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111134915.8A Active CN113838068B (en) | 2021-09-27 | 2021-09-27 | Automatic segmentation method, device and storage medium for myocardial segments |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113838068B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112529919A (en) * | 2019-12-11 | 2021-03-19 | 上海联影智能医疗科技有限公司 | System and method for generating a bull's eye map generation of a subject's heart |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7864997B2 (en) * | 2006-04-28 | 2011-01-04 | Pie Medical Imaging B.V. | Method, apparatus and computer program product for automatic segmenting of cardiac chambers |
| CN101506841B (en) * | 2006-06-26 | 2016-06-01 | 皇家飞利浦电子股份有限公司 | The semi-automatic of myocardial region of interest is defined |
| CN104978440B (en) * | 2014-04-03 | 2020-02-07 | 上海联影医疗科技有限公司 | Method for establishing and registering heart model and multi-plane reconstruction |
| US9949643B2 (en) * | 2014-10-18 | 2018-04-24 | International Business Machines Corporation | Automatic visualization of regional functional parameters of left ventricle from cardiac imaging |
| WO2018140596A2 (en) * | 2017-01-27 | 2018-08-02 | Arterys Inc. | Automated segmentation utilizing fully convolutional networks |
| CN109584254B (en) * | 2019-01-07 | 2022-12-20 | 浙江大学 | Heart left ventricle segmentation method based on deep full convolution neural network |
| CN110335236B (en) * | 2019-04-29 | 2021-06-11 | 上海依智医疗技术有限公司 | Method, apparatus, system and medium for processing cardiac medical image |
| CN110335235B (en) * | 2019-04-29 | 2020-12-01 | 上海依智医疗技术有限公司 | Processing device, processing system and medium for cardiac medical image |
| CN110120051A (en) * | 2019-05-10 | 2019-08-13 | 上海理工大学 | A kind of right ventricle automatic division method based on deep learning |
| CN110197492A (en) * | 2019-05-23 | 2019-09-03 | 山东师范大学 | A kind of cardiac MRI left ventricle dividing method and system |
| CN112508949B (en) * | 2021-02-01 | 2021-05-11 | 之江实验室 | Method for automatically segmenting left ventricle of SPECT three-dimensional reconstruction image |
-
2021
- 2021-09-27 CN CN202111134915.8A patent/CN113838068B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112529919A (en) * | 2019-12-11 | 2021-03-19 | 上海联影智能医疗科技有限公司 | System and method for generating a bull's eye map generation of a subject's heart |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113838068A (en) | 2021-12-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11847547B2 (en) | Method and system for generating a centerline for an object, and computer readable medium | |
| CN113902741B (en) | Method, device and medium for performing blood vessel segmentation on medical image | |
| US8824755B2 (en) | Method and computer system for automatically generating a statistical model | |
| JP4512586B2 (en) | Volume measurement in 3D datasets | |
| EP2554120A1 (en) | Projection image generation device, projection image generation programme, and projection image generation method | |
| CN110335236B (en) | Method, apparatus, system and medium for processing cardiac medical image | |
| CN111476791B (en) | Image processing method, image processing apparatus, and non-transitory computer readable medium | |
| CN110619635B (en) | Hepatocellular carcinoma magnetic resonance image segmentation system and method based on deep learning | |
| CN109712163B (en) | Coronary artery extraction method, device, image processing workstation and readable storage medium | |
| Zheng et al. | MDCC-Net: multiscale double-channel convolution U-Net framework for colorectal tumor segmentation | |
| CN113205473A (en) | Method, apparatus and storage medium for data enhancement for medical image detection | |
| CN113192069A (en) | Semantic segmentation method and device for tree structure in three-dimensional tomography image | |
| CN110335235A (en) | Processing unit, processing system and the medium of cardiologic medical image | |
| CN114596311A (en) | Blood vessel function evaluation method and blood vessel function evaluation device based on blood vessel image | |
| CN115035020A (en) | Method, apparatus and storage medium for object analysis of medical images | |
| CN110390667B (en) | Focus extraction method, device, equipment and storage medium based on fundus OCT image | |
| WO2022247573A1 (en) | Model training method and apparatus, image processing method and apparatus, device, and storage medium | |
| CN113129297B (en) | Diameter automatic measurement method and system based on multi-phase tumor image | |
| Tang et al. | Automatic classification and segmentation of atherosclerotic plaques in the intravascular optical coherence tomography (IVOCT) | |
| CN113838068B (en) | Automatic segmentation method, device and storage medium for myocardial segments | |
| CN115018825B (en) | Coronary artery dominant type classification method, classification device and storage medium | |
| CN114419032B (en) | Method and device for segmenting the endocardium and/or the epicardium of the left ventricle of the heart | |
| CN117765081A (en) | Preoperative target plane positioning method and device and medical scanning imaging equipment | |
| CN114004835B (en) | Method, apparatus and storage medium for object analysis of medical images | |
| Chairat et al. | Non-contact chronic wound analysis using deep learning |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |