CN117474940A - Method, device, equipment and medium for generating medical image segmentation model - Google Patents
Method, device, equipment and medium for generating medical image segmentation model Download PDFInfo
- Publication number
- CN117474940A CN117474940A CN202311562670.8A CN202311562670A CN117474940A CN 117474940 A CN117474940 A CN 117474940A CN 202311562670 A CN202311562670 A CN 202311562670A CN 117474940 A CN117474940 A CN 117474940A
- Authority
- CN
- China
- Prior art keywords
- contour
- segmentation
- image sequence
- layer
- target
- 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 42
- 238000003709 image segmentation Methods 0.000 title claims abstract description 32
- 230000011218 segmentation Effects 0.000 claims abstract description 123
- 238000005070 sampling Methods 0.000 claims abstract description 32
- 238000001914 filtration Methods 0.000 claims abstract description 15
- 238000012952 Resampling Methods 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims description 19
- 230000002159 abnormal effect Effects 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 8
- 238000009499 grossing Methods 0.000 claims description 7
- 210000002307 prostate Anatomy 0.000 description 10
- 230000009286 beneficial effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 3
- 238000002591 computed tomography Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000003902 lesion Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000000056 organ Anatomy 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000005481 NMR spectroscopy Methods 0.000 description 1
- 230000003187 abdominal effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 210000003734 kidney Anatomy 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 208000017497 prostate disease Diseases 0.000 description 1
- 230000003068 static effect 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
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/181—Segmentation; Edge detection involving edge growing; involving edge linking
-
- 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
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- 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
- G06T2207/30081—Prostate
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method, a device, equipment and a medium for generating a medical image segmentation model, wherein the method comprises the following steps: acquiring a priori segmentation contour; filtering and denoising the original image sequence to obtain a first image sequence; sampling the first image sequence according to the prior segmentation contour to obtain a second image sequence; generating a target candidate frame according to the prior segmentation contour, and cutting out the second image sequence to obtain a third image sequence; generating a segmentation threshold according to the prior segmentation contour to form a segmentation threshold candidate set; acquiring a region of interest candidate set of each layer according to the segmentation threshold candidate set; acquiring an edge candidate set; merging the intersection points of the edges in each layer to form an intersection point set; selecting an automatic segmentation contour which is most similar to the prior segmentation contour as a reference contour of the current layer according to the intersection point set; resampling all the reference contours, and modeling to form a target medical image segmentation model. The method is used for automatically acquiring the high-precision medical image segmentation model.
Description
Technical Field
The present invention relates to the field of medical image processing technologies, and in particular, to a method, an apparatus, a device, and a medium for generating a medical image segmentation model.
Background
At present, prostate disease is a common disorder in men, and involves magnetic resonance (magnetic resonance, MR) images of the prostate during both diagnosis and treatment. The segmentation of the prostate by the MR image not only assists the doctor in judging the lesions, but also guides the protection of the organs around the lesion during the treatment.
Currently, segmentation of the prostate on the MR image of the prostate is mostly based on a doctor's layer-by-layer manual delineation, the accuracy of the segmentation is entirely dependent on the doctor's experience and subjective knowledge, and the layer-by-layer delineation results in higher labor and time costs. Therefore, a new method, apparatus, device and medium for generating a medical image segmentation model are needed to improve the above-mentioned problems.
Disclosure of Invention
The invention aims to provide a method, a device, equipment and a medium for generating a medical image segmentation model, which are used for automatically acquiring a high-precision medical image segmentation model.
In a first aspect, the present invention provides a method for generating a medical image segmentation model, comprising: s0, acquiring a priori segmentation contour containing a target; s1, filtering and denoising an original image sequence to obtain a first image sequence; s2, according to the prior segmentation contour, multidirectional sampling is carried out on the first image sequence, and a second image sequence containing an automatic segmentation contour is obtained; s3, generating a target candidate frame according to the prior segmentation contour, and cutting out the second image sequence according to the target candidate frame to obtain a third image sequence; s4, generating a segmentation threshold value corresponding to each prior segmentation layer in the third image sequence according to the prior segmentation contour so as to form a segmentation threshold value candidate set; s5, acquiring a region of interest candidate set of each layer in the third image sequence according to the segmentation threshold candidate set; s6, carrying out edge extraction and edge smoothing on the region of interest in each layer to obtain edge candidate sets of all the regions of interest in the layer; s7, merging edges in each layer, and obtaining intersection points of the edges to form an intersection point set; s8, selecting an automatic segmentation contour which is most similar to the prior segmentation contour as a reference contour of the current layer according to the intersection point set; s9, resampling all reference contours in the third image sequence to obtain target contours corresponding to all layers in the original image sequence, and modeling to form a target medical image segmentation model.
The method has the beneficial effects that: the invention supports manual sketching of any cut layer as the prior cut contour, and obtains the automatic cut contour of other layers according to the prior cut contour, thereby being beneficial to improving the cut efficiency. And cutting the second image sequence according to the target candidate frame obtained by the prior segmentation contour, thereby being beneficial to improving the target screening efficiency. The segmentation threshold candidate set, the region of interest candidate set, the edge candidate set and the intersection point set are arranged so as to be convenient for selecting the reference contour. The target contour corresponding to the first image sequence can be obtained by resampling all the reference contours, so that the target medical image segmentation model directly corresponds to the original image sequence, and the model segmentation precision is improved.
Optionally, the method further comprises: s10, for layers between adjacent prior segmentation contours, sequentially acquiring two reference contours for each layer along opposite directions, and then fusing the two reference contours to obtain a target contour.
Optionally, the method further comprises: s11, adding at least one layer containing a priori segmentation contour aiming at an abnormal section of the target medical image segmentation model, and repeating S1-S9 until the acquired target medical image segmentation model is abnormal.
Optionally, the S2 includes: sampling the first image sequence in X, Y and Z directions, wherein the sampling rate in the Z direction is set to be larger than the sampling rates in the X direction and the Y direction; after multidirectional sampling is carried out on the first image sequence and the contour is automatically segmented, inserting a layer containing the prior segmentation contour into a corresponding position of the first image sequence, and obtaining a second image sequence containing the prior segmentation contour and the automatic segmentation contour.
Optionally, the S4 includes: and dividing the layer image by using different thresholds to obtain a pre-divided contour, calculating the intersection ratio of the pre-divided contour and the prior divided contour, and taking the threshold with the maximum intersection ratio as the dividing threshold of the layer.
Optionally, the step S8 includes: for each layer in the third image sequence, obtaining all contours passing through at least two intersection points by utilizing an intersection point set of the edges of the layer; selecting a contour which is most similar to the nearest prior segmentation contour from all contours as a target contour of the current layer according to the prior characteristics of the target; and selecting the contour most similar to the contour of the previous layer from all the contours as the target contour of the current layer by using the layers between the adjacent priori segmentation contours.
In a second aspect, the present invention provides a medical image segmentation model generating apparatus, for use in the method of any one of the first aspects, comprising: an input unit for acquiring a priori segmentation contours comprising the object; the filtering unit is used for carrying out filtering denoising processing on the original image sequence to obtain a first image sequence; the sampling unit is used for performing multidirectional sampling on the first image sequence and automatically dividing the outline to obtain a second image sequence containing the automatically divided outline; the clipping unit is used for generating a target candidate frame according to the prior segmentation contour, clipping the second image sequence according to the target candidate frame and obtaining a third image sequence; the processing unit is used for generating a segmentation threshold value of each prior segmentation layer in the third image sequence according to the prior segmentation contour so as to form a segmentation threshold value candidate set; acquiring a region of interest candidate set of each layer in a third image sequence according to the segmentation threshold candidate set; performing edge extraction and edge smoothing on the region of interest in each layer to obtain edge candidate sets of all the regions of interest in the layer; merging edges in each layer, and obtaining intersection points of the edges after merging to form an intersection point set; selecting an automatic segmentation contour most similar to the prior segmentation contour as a reference contour of the current layer according to the intersection point set; resampling all reference contours in the third image sequence to obtain target contours corresponding to all layers in the original image sequence, and modeling to form a target medical image segmentation model.
Optionally, the method further comprises: the processing unit is further used for sequentially acquiring two reference contours of each layer along opposite directions for the layers between adjacent prior segmentation contours, and then fusing the two reference contours to obtain a target contour.
In a third aspect, the present invention provides an electronic device comprising a memory and a processor, the memory having stored thereon a program executable on the processor, which when executed by the processor causes the electronic device to implement the method of any of the first aspects.
In a fourth aspect, the present invention provides a readable storage medium having a program stored therein, wherein the program, when executed, implements the method of any one of the first aspects.
Drawings
FIG. 1 is a schematic flow chart of a method for generating a medical image segmentation model provided by the invention;
fig. 2 is a schematic diagram of a magnetic resonance image including a prostate according to the present invention;
FIG. 3 is a schematic illustration of a magnetic resonance image including an automatic segmentation contour according to the present invention;
FIG. 4 is a schematic diagram of a two-way reference profile acquisition method according to the present invention;
FIG. 5 is a schematic structural diagram of a device for generating a medical image segmentation model according to the present invention;
fig. 6 is a schematic structural diagram of an electronic device according to the present invention.
Reference numerals in the drawings:
1. a first a priori segmentation contour; 2. a second prior segmented contour;
31. an input unit; 32. a filtering unit; 33. a sampling unit; 34. a cutting unit; 35. a processing unit;
41. a processor; 42. a memory; 43. an output interface; 44. a communication interface; 45. an antenna.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention. Unless otherwise defined, technical or scientific terms used herein should be given the ordinary meaning as understood by one of ordinary skill in the art to which this invention belongs. As used herein, the word "comprising" and the like means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof without precluding other elements or items.
In view of the problems in the prior art, as shown in fig. 1, a first embodiment provides a method for generating a medical image segmentation model, including: s0, acquiring a priori segmentation contour containing a target; s1, filtering and denoising an original image sequence to obtain a first image sequence; s2, according to the prior segmentation contour, multidirectional sampling is carried out on the first image sequence, and a second image sequence containing an automatic segmentation contour is obtained; s3, generating a target candidate frame according to the prior segmentation contour, and cutting out the second image sequence according to the target candidate frame to obtain a third image sequence; s4, generating a segmentation threshold value corresponding to each prior segmentation layer in the third image sequence according to the prior segmentation contour so as to form a segmentation threshold value candidate set; s5, acquiring a region of interest candidate set of each layer in the third image sequence according to the segmentation threshold candidate set; s6, carrying out edge extraction and edge smoothing on the region of interest in each layer to obtain edge candidate sets of all the regions of interest in the layer; s7, merging edges in each layer, and obtaining intersection points of the edges to form an intersection point set; s8, selecting an automatic segmentation contour which is most similar to the prior segmentation contour as a reference contour of the current layer according to the intersection point set; s9, resampling all reference contours in the third image sequence to obtain target contours corresponding to all layers in the original image sequence, and modeling to form a target medical image segmentation model.
It is worth to be noted that, the embodiment supports manual sketching of any slicing layer as a priori segmentation contour, and obtains automatic segmentation contours of other layers according to the priori segmentation contour, which is beneficial to improving segmentation efficiency. And cutting the second image sequence according to the target candidate frame obtained by the prior segmentation contour, thereby being beneficial to improving the target screening efficiency. The segmentation threshold candidate set, the region of interest candidate set, the edge candidate set and the intersection point set are arranged so as to be convenient for selecting the reference contour. The target contour corresponding to the first image sequence can be obtained by resampling all the reference contours, so that the target medical image segmentation model directly corresponds to the original image sequence, and the model segmentation precision is improved.
Specifically, the S0 includes: the method comprises the steps that a target is set to be the prostate of a current object, and the prior segmentation contour is set to be an artificial segmentation contour containing the prostate, wherein the artificial segmentation contour is obtained by manual sketching.
In other specific embodiments, the S0 includes: the prior segmentation contour is set as a default segmentation contour containing the kidney, and the default segmentation contour is obtained by scaling, stretching, translating and rotating the segmentation contour of the non-current object. It is worth noting that the target may be set to any abdominal organ of the current subject.
In still other specific embodiments, the S1 includes: the original image sequence is set to a magnetic resonance image sequence comprising the target. One layer of a magnetic resonance image sequence containing the prostate is shown in figure 2. And carrying out Gaussian filtering processing on the magnetic resonance image sequence to reduce noise in the magnetic resonance image acquisition process.
In still other specific embodiments, the S1 includes: the original image sequence is set to an electronic computed tomography image sequence comprising the object. And carrying out median filtering processing on the electronic computer tomography image sequence to reduce noise in the acquisition process of the electronic computer tomography image. It is worth noting that the original image sequence may be any medical 3D image sequence containing a target.
In some embodiments, the S2 comprises: sampling the first image sequence in X, Y and Z directions, wherein the sampling rate in the Z direction is set to be larger than the sampling rates in the X direction and the Y direction; after multidirectional sampling is carried out on the first image sequence and the contour is automatically segmented, inserting a layer containing the prior segmentation contour into a corresponding position of the first image sequence, and obtaining a second image sequence containing the prior segmentation contour and the automatic segmentation contour. The automatic segmentation profile obtained is shown in fig. 3.
Specifically, the X direction is set to be the body front direction, the Y direction is set to be the body side direction, and the Z direction is set to be the head top direction. The sampling rate in the direction X, Y, Z is set to (0.57,0.57,4.00) mm. In other embodiments, the sampling rate in the X, Y, Z direction is set to (1.20,1.20,8.00) mm. In yet another specific embodiment, the sampling rate in the direction X, Y, Z is set to (0.20,0.20,2.00) mm. It should be noted that the sampling rate in the direction X, Y, Z may be set to any value, but it is required that the sampling rate in the direction X, Y be the same and smaller than the sampling rate in the Z direction.
In some embodiments, the S4 comprises: and dividing the layer image by using different thresholds to obtain a pre-divided contour, calculating the intersection ratio of the pre-divided contour and the prior divided contour, and taking the threshold with the maximum intersection ratio as the dividing threshold of the layer. The pre-segmented contours are calculated from a priori segmented contours by an adaptive algorithm.
Specifically, the self-adaptive algorithm is an algorithm based on region growth, an image is divided into a plurality of regions according to texture features of a nuclear magnetic image, and then contour extraction is performed according to features in the regions. The region growing algorithm may select seed points based on the a priori segmented contours and then adaptively expand the region to generate pre-segmented contours.
In other specific embodiments, the adaptive algorithm is a level set theory-based algorithm that represents the shape in the nuclear magnetic image as a minimum of a level set function. The algorithm based on the level set theory can update the level set function through iteration, so that the shape gradually approximates to the prior segmentation contour, and finally the pre-segmentation contour is generated.
In still other specific embodiments, the adaptive algorithm is a deep learning based algorithm, and the deep learning algorithm may adaptively learn the features and patterns that generate the pre-segmented contours under the direction of the pre-segmented contours.
In some embodiments, the S5 comprises: and for each layer which does not contain the prior segmentation contour in the third image sequence, sequentially segmenting by using each threshold value in the candidate threshold value sets to obtain a region-of-interest candidate set corresponding to the layer.
In some embodiments, the S6 includes: and for each layer in the third image sequence, sequentially carrying out edge extraction and edge smoothing on each region of interest in the third image sequence, wherein edges corresponding to all the regions of interest of each layer form an edge candidate set.
Illustratively, for a layer in the third image sequence, the S7 includes: the layer extracts 2 regions of interest with overlapping portions, traverses the edges of the overlapping portions, and can determine the intersection point of the 2 regions of interest.
In some embodiments, the S8 comprises: for each layer in the third image sequence, obtaining all contours passing through at least two intersection points by utilizing an intersection point set of the edge of the layer; selecting a contour which is most similar to the nearest prior segmentation contour from all contours as a target contour of the current layer according to the prior characteristics of the target; and selecting the contour most similar to the contour of the previous layer from all the contours as the target contour of the current layer by using the layers between the adjacent priori segmentation contours.
In particular, when the target is the prostate, its a priori features include: the prostate gland becomes larger and then smaller in the Z direction. The similarity between contours is measured by the hausdorff distance, and the smaller the hausdorff distance value, the higher the similarity.
In some embodiments, the S9 includes: when resampling is performed, the sampling rate is set to coincide with the original image sequence. Illustratively, the original image sequence and the first image sequence are each set to 30 layers, and the second image sequence and the third image sequence are each set to 10 layers. The fourth image sequence obtained after resampling is set to 30 layers. The 30 layers in the fourth image sequence correspond one-to-one with the 30 layers in the original image sequence.
In some specific embodiments, the third image sequence set to 10 layers may be expanded up to form a fourth image sequence of 30 layers by performing interpolation processing on the third image sequence. In the upward expansion, interpolation methods can be used to estimate new pixel values.
As shown in fig. 4, in some embodiments, further comprising: s10, for layers between adjacent prior segmentation contours, sequentially acquiring two reference contours for each layer along opposite directions, and then fusing the two reference contours to obtain a target contour.
Specifically, taking n layers between the first prior segmentation contour 1 and the second prior segmentation contour 2 as an example, n is a positive integer. When the reference contour is acquired along the Z direction, the forward reference contour of the initial layer L1 is acquired according to the first priori segmentation contour 1, the forward reference contour of the intermediate layer L2 is acquired according to the initial layer L1, and the forward reference contour of the ending layer Ln is acquired according to the intermediate layer Ln-1.
In other specific embodiments, when the reference contour is obtained in the Z-direction, the backward reference contour of the ending layer Ln is obtained according to the second prior segmentation contour 2, the backward reference contour of the middle layer Ln-1 is obtained according to the ending layer Ln, and the backward reference contour of the starting layer L1 is obtained according to the middle layer L2.
In still other embodiments, the target profile is obtained by fusing the forward reference profile and the backward reference profile of the same layer.
In some embodiments, further comprising: s11, adding at least one layer containing a priori segmentation contour aiming at an abnormal section of the target medical image segmentation model, and repeating S1-S9 until the acquired target medical image segmentation model is abnormal.
Specifically, the layer containing the prior segmentation contour is a layer obtained by manually sketching the abnormal layer corresponding to the original image sequence. In other specific embodiments, the layer containing the a priori segmented contours is a layer obtained based on a non-outlier layer iterative calculation. In some embodiments, S11 further comprises determining whether each layer in the target medical image segmentation model is abnormal based on the prior features. When the current layer is confirmed to be abnormal, M layers adjacent to the current layer are confirmed to be abnormal slice intervals.
As shown in fig. 5, a second embodiment provides a medical image segmentation model generating apparatus, which is configured to be used in the method according to any one of the foregoing embodiments, including: an input unit 31 for acquiring a priori segmented contours comprising the object; a filtering unit 32, configured to perform filtering denoising processing on the original image sequence, so as to obtain a first image sequence; a sampling unit 33, configured to perform multidirectional sampling on the first image sequence and automatically segment the contour, so as to obtain a second image sequence including the automatically segmented contour; a cropping unit 34, configured to generate a target candidate frame according to the prior segmentation contour, crop the second image sequence according to the target candidate frame, and obtain a third image sequence; a processing unit 35, configured to generate a segmentation threshold value of each a priori segmentation layer in the third image sequence according to the a priori segmentation contour, so as to form a segmentation threshold candidate set; acquiring a region of interest candidate set of each layer in a third image sequence according to the segmentation threshold candidate set; performing edge extraction and edge smoothing on the region of interest in each layer to obtain edge candidate sets of all the regions of interest in the layer; merging edges in each layer, and obtaining intersection points of the edges after merging to form an intersection point set; selecting an automatic segmentation contour most similar to the prior segmentation contour as a reference contour of the current layer according to the intersection point set; resampling all reference contours in the third image sequence to obtain target contours corresponding to all layers in the original image sequence, and modeling to form a target medical image segmentation model.
Specifically, the input unit 31 is configured as a nuclear magnetic resonance spectrometer, the filtering unit 32 is configured as a filter, the sampling unit 33 is configured as a radio frequency receiver, and the clipping unit 34 and the processing unit 35 are configured as processors. In other specific embodiments, the input unit 31, the filtering unit 32, the sampling unit 33, the clipping unit 34, and the processing unit 35 are disposed on the same image processing chip.
In some embodiments, further comprising: the processing unit is further used for sequentially acquiring two reference contours of each layer along opposite directions for the layers between adjacent prior segmentation contours, and then fusing the two reference contours to obtain a target contour.
As shown in fig. 6, a third embodiment provides an electronic device comprising a memory 42 and a processor 41, the memory 42 having stored thereon a program executable on the processor 41, which when executed by the processor 41 causes the electronic device to implement the method of any of the above embodiments.
In one possible embodiment, the electronic device further comprises: an output interface 43 for outputting a result; a communication interface 44 for communicating the transmission signal; an antenna 45 for transmitting or receiving signals.
It should be noted that the processor 41 in the present embodiment may be an image processing chip or an integrated circuit chip having processing capability for image signals. In implementation, the steps of the above method embodiments may be implemented by integrated logic circuits of hardware in a processor or instructions in software form. The processor may be a general purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), an off-the-shelf programmable gate array (Field Programmable Gate Array, FPGA), or other programmable logic device. The methods, steps and logic blocks disclosed in the present embodiment may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the present embodiment may be embodied directly in hardware, in a decoded processor, or in a combination of hardware and software modules in a decoded processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory, and the processor reads the information in the memory and, in combination with its hardware, performs the steps of the above method.
It will be appreciated that the memory 42 in this embodiment may be either volatile memory or nonvolatile memory, or may include both volatile and nonvolatile memory. The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable EPROM (EEPROM), or a flash Memory. The volatile memory may be random access memory (Random Access Memory, RAM) which acts as an external cache. By way of example, and not limitation, many forms of RAM are available, such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (Double Data Rate SDRAM), enhanced SDRAM (ESDRAM), synchronous DRAM (SLDRAM), and Direct RAM (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to comprise, without being limited to, these and any other suitable types of memory.
A fourth embodiment provides a readable storage medium having a program stored therein, characterized in that the program, when executed, implements the method of any one of the first aspects.
It is noted that the method may be stored in a readable storage medium if implemented in the form of a software functional unit and sold or used as a stand-alone product. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in the form of a software product stored in a storage medium, comprising several instructions for causing an electronic device to perform all or part of the steps of the method described in the various embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a U disk, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk.
While embodiments of the present invention have been described in detail hereinabove, it will be apparent to those skilled in the art that various modifications and variations can be made to these embodiments. It is to be understood that such modifications and variations are within the scope and spirit of the present invention as set forth in the following claims. Moreover, the invention described herein is capable of other embodiments and of being practiced or of being carried out in various ways.
Claims (10)
1. A method of generating a medical image segmentation model, comprising:
s0, acquiring a priori segmentation contour containing a target;
s1, filtering and denoising an original image sequence to obtain a first image sequence;
s2, according to the prior segmentation contour, multidirectional sampling is carried out on the first image sequence, and a second image sequence containing an automatic segmentation contour is obtained;
s3, generating a target candidate frame according to the prior segmentation contour, and cutting out the second image sequence according to the target candidate frame to obtain a third image sequence;
s4, generating a segmentation threshold value corresponding to each prior segmentation layer in the third image sequence according to the prior segmentation contour so as to form a segmentation threshold value candidate set;
s5, acquiring a region of interest candidate set of each layer in the third image sequence according to the segmentation threshold candidate set;
s6, carrying out edge extraction and edge smoothing on the region of interest in each layer to obtain edge candidate sets of all the regions of interest in the layer;
s7, merging edges in each layer, and obtaining intersection points of the edges to form an intersection point set;
s8, selecting an automatic segmentation contour which is most similar to the prior segmentation contour as a reference contour of the current layer according to the intersection point set;
s9, resampling all reference contours in the third image sequence to obtain target contours corresponding to all layers in the original image sequence, and modeling to form a target medical image segmentation model.
2. The method as recited in claim 1, further comprising:
s10, for layers between adjacent prior segmentation contours, sequentially acquiring two reference contours for each layer along opposite directions, and then fusing the two reference contours to obtain a target contour.
3. The method as recited in claim 1, further comprising:
s11, adding at least one layer containing a priori segmentation contour aiming at an abnormal section of the target medical image segmentation model, and repeating S1-S9 until the acquired target medical image segmentation model is abnormal.
4. The method according to claim 1, wherein S2 comprises:
sampling the first image sequence in X, Y and Z directions, wherein the sampling rate in the Z direction is set to be larger than the sampling rates in the X direction and the Y direction;
after multidirectional sampling is carried out on the first image sequence and the contour is automatically segmented, inserting a layer containing the prior segmentation contour into a corresponding position of the first image sequence, and obtaining a second image sequence containing the prior segmentation contour and the automatic segmentation contour.
5. The method according to claim 1, wherein S4 comprises:
and dividing the layer image by using different thresholds to obtain a pre-divided contour, calculating the intersection ratio of the pre-divided contour and the prior divided contour, and taking the threshold with the maximum intersection ratio as the dividing threshold of the layer.
6. The method according to claim 1, wherein said S8 comprises:
for each layer in the third image sequence, obtaining all contours passing through at least two intersection points by utilizing an intersection point set of the edges of the layer; selecting a contour which is most similar to the nearest prior segmentation contour from all contours as a target contour of the current layer according to the prior characteristics of the target; and selecting the contour most similar to the contour of the previous layer from all the contours as the target contour of the current layer by using the layers between the adjacent priori segmentation contours.
7. A medical image segmentation model generation apparatus for use in the method of any one of claims 1 to 6, comprising:
an input unit for acquiring a priori segmentation contours comprising the object;
the filtering unit is used for carrying out filtering denoising processing on the original image sequence to obtain a first image sequence;
the sampling unit is used for performing multidirectional sampling on the first image sequence and automatically dividing the outline to obtain a second image sequence containing the automatically divided outline;
the clipping unit is used for generating a target candidate frame according to the prior segmentation contour, clipping the second image sequence according to the target candidate frame and obtaining a third image sequence;
the processing unit is used for generating a segmentation threshold value of each prior segmentation layer in the third image sequence according to the prior segmentation contour so as to form a segmentation threshold value candidate set; acquiring a region of interest candidate set of each layer in a third image sequence according to the segmentation threshold candidate set; performing edge extraction and edge smoothing on the region of interest in each layer to obtain edge candidate sets of all the regions of interest in the layer; merging edges in each layer, and obtaining intersection points of the edges after merging to form an intersection point set; selecting an automatic segmentation contour most similar to the prior segmentation contour as a reference contour of the current layer according to the intersection point set; resampling all reference contours in the third image sequence to obtain target contours corresponding to all layers in the original image sequence, and modeling to form a target medical image segmentation model.
8. The apparatus as recited in claim 7, further comprising:
the processing unit is further used for sequentially acquiring two reference contours of each layer along opposite directions for the layers between adjacent prior segmentation contours, and then fusing the two reference contours to obtain a target contour.
9. An electronic device comprising a memory and a processor, the memory having stored thereon a program executable on the processor, which when executed by the processor, causes the electronic device to implement the method of any of claims 1 to 6.
10. A readable storage medium having a program stored therein, characterized in that the program, when executed, implements the method of any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311562670.8A CN117474940A (en) | 2023-11-21 | 2023-11-21 | Method, device, equipment and medium for generating medical image segmentation model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311562670.8A CN117474940A (en) | 2023-11-21 | 2023-11-21 | Method, device, equipment and medium for generating medical image segmentation model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117474940A true CN117474940A (en) | 2024-01-30 |
Family
ID=89639602
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311562670.8A Pending CN117474940A (en) | 2023-11-21 | 2023-11-21 | Method, device, equipment and medium for generating medical image segmentation model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117474940A (en) |
-
2023
- 2023-11-21 CN CN202311562670.8A patent/CN117474940A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111798462B (en) | Automatic delineation method of nasopharyngeal carcinoma radiotherapy target area based on CT image | |
CN109035197B (en) | CT radiography image kidney tumor segmentation method and system based on three-dimensional convolution neural network | |
US5319551A (en) | Region extracting method and three-dimensional display method | |
CN106600609B (en) | Spine segmentation method and system in medical image | |
EP2036037B1 (en) | Methods and systems for segmentation using boundary reparameterization | |
CN111161279B (en) | Medical image segmentation method, device and server | |
CN107545579B (en) | Heart segmentation method, device and storage medium | |
CN110310287A (en) | It is neural network based to jeopardize the automatic delineation method of organ, equipment and storage medium | |
CN112435263A (en) | Medical image segmentation method, device, equipment, system and computer storage medium | |
US8131075B2 (en) | Fast 4D segmentation of large datasets using graph cuts | |
CN111784701B (en) | Ultrasonic image segmentation method and system combining boundary feature enhancement and multi-scale information | |
CN109886973B (en) | Blood vessel extraction method and device and computer readable storage medium | |
CN112348769A (en) | Intelligent kidney tumor segmentation method and device in CT (computed tomography) image based on U-Net depth network model | |
CN109712161A (en) | A kind of image partition method, device, equipment and storage medium | |
CN108765445B (en) | Lung trachea segmentation method and device | |
Bischoff et al. | Isosurface reconstruction with topology control | |
CN111932495B (en) | Medical image detection method, device and storage medium | |
CN115564782A (en) | 3D blood vessel and trachea segmentation method and system | |
CN116258784A (en) | CT image metal artifact correction and ablation visualization method and system | |
CN116503607A (en) | CT image segmentation method and system based on deep learning | |
CN114862873A (en) | CT image segmentation processing method and device | |
CN117197345B (en) | Intelligent bone joint three-dimensional reconstruction method, device and equipment based on polynomial fitting | |
CN107240140A (en) | A kind of method and device of image reconstruction | |
CN117474940A (en) | Method, device, equipment and medium for generating medical image segmentation model | |
CN116468103A (en) | Training method, application method and system for lung nodule benign and malignant recognition model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |