CN116823740A

CN116823740A - Vascular plaque detection system, device, and storage medium

Info

Publication number: CN116823740A
Application number: CN202310658996.4A
Authority: CN
Inventors: 洪凯; 马骏; 郑凌霄; 兰宏志
Original assignee: Shenzhen Raysight Intelligent Medical Technology Co Ltd
Current assignee: Shenzhen Raysight Intelligent Medical Technology Co Ltd
Priority date: 2023-06-05
Filing date: 2023-06-05
Publication date: 2023-09-29

Abstract

The invention discloses a vascular plaque detection system, a vascular plaque detection device and a storage medium, and belongs to the technical field of medical image processing. The system includes a processor configured to perform a vascular plaque detection method comprising: projecting blood vessels in the image to be processed along at least one projection direction to obtain at least one blood vessel projection image; determining plaque segmentation results of each blood vessel projection image, wherein the plaque segmentation results comprise calcified spots and/or non-calcified spots; based on a set rule, carrying out back projection on plaque segmentation results of each blood vessel projection image along a projection direction corresponding to each blood vessel projection image to obtain a back projection result; and determining plaque detection results of the image to be processed according to all the back projection results. The technical effect of improving the plaque detection accuracy is achieved.

Description

Vascular plaque detection system, device, and storage medium

Technical Field

The present invention relates to the field of medical image processing technologies, and in particular, to a vascular plaque detection system, a vascular plaque detection device, and a storage medium.

Background

The prior art generally adopts a deep learning neural network to carry out plaque classification on coronary images, and the biggest defect is in two aspects of data preparation and feature extraction. Correct label data is a necessary requirement of an algorithm, but the center line point-by-point data labeling is obviously unsuitable, and interval section type labeling leads to poor generalization capability of the deep learning neural network; the information provided by the non-calcified plaque compared with the calcified plaque with obvious characteristics is limited, which leads to a large precision error of the classification of the non-calcified plaque, and a certain precision error of the classification of the calcified plaque.

In summary, the prior art has the problems of low segmentation accuracy of the system or the method mainly comprising the classification method.

Disclosure of Invention

The invention provides a vascular plaque detection system, a vascular plaque detection device and a storage medium, which are used for solving the problem of low accuracy of the existing plaque detection system.

According to an aspect of the present invention, there is provided a vascular plaque detection system including a processor configured to perform a vascular plaque detection method comprising:

projecting blood vessels in the image to be processed along at least one projection direction to obtain at least one blood vessel projection image;

determining plaque segmentation results of each blood vessel projection image, wherein the plaque segmentation results comprise calcified spots and/or non-calcified spots;

based on a set rule, carrying out back projection on plaque segmentation results of each blood vessel projection image along a projection direction corresponding to each blood vessel projection image to obtain a back projection result;

and determining plaque detection results of the image to be processed according to all the back projection results.

According to another aspect of the present invention, there is provided a vascular plaque detection apparatus including:

According to another aspect of the present invention, there is provided a computer readable storage medium storing computer instructions for causing a processor to execute the vascular plaque detection method according to any of the embodiments of the present invention.

Compared with the prior art, the plaque segmentation result of the three-dimensional blood vessel image is directly determined, and the plaque segmentation data processing amount is reduced by determining the plaque segmentation result of the blood vessel projection image, so that the plaque segmentation speed and the plaque segmentation accuracy are improved; carrying out back projection on plaque segmentation results of each blood vessel projection image along the projection direction corresponding to each blood vessel projection image to obtain a back projection result; and determining the plaque detection result of the image to be processed according to all the back projection results, thereby realizing the purpose of quickly and indirectly determining the plaque detection result of the image to be processed.

It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the invention or to delineate the scope of the invention. Other features of the present invention will become apparent from the description that follows.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic structural diagram of a vascular plaque detection system provided according to an embodiment of the present invention;

fig. 2A is a flowchart of a method for detecting a vascular plaque according to an embodiment of the present invention;

FIG. 2B is a schematic illustration of an equally spaced projection provided in accordance with an embodiment of the present invention;

FIG. 3A is a flowchart of yet another method for detecting vascular plaque provided in accordance with an embodiment of the present invention;

fig. 3B is a schematic diagram of a rectangular coordinate system according to an embodiment of the present invention;

FIG. 3C is a schematic diagram of a polar coordinate system provided in accordance with an embodiment of the present invention;

FIG. 4 is a flow chart of yet another method for detecting vascular plaque provided in accordance with an embodiment of the present invention;

FIG. 5 is a flow chart of yet another method for detecting vascular plaque provided in accordance with an embodiment of the present invention;

FIG. 6 is a flow chart of yet another method for detecting vascular plaque provided in accordance with an embodiment of the present invention;

FIG. 7A is a flow chart of a back-projection method provided in accordance with an embodiment of the present invention;

FIG. 7B is a schematic diagram of a positional relationship between a target pixel and a center line according to an embodiment of the present invention;

fig. 8 is a schematic structural view of a vascular plaque detection apparatus according to an embodiment of the present invention.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.

It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

Fig. 1 shows a schematic diagram of a vascular plaque detection system 10 that may be used to implement an embodiment of the present invention. Vascular plaque detection systems are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers.

As shown in fig. 1, the vascular plaque detection system 10 includes at least one processor 11, and a memory, such as a Read Only Memory (ROM) 12, a Random Access Memory (RAM) 13, etc., communicatively connected to the at least one processor 11, in which the memory stores a computer program executable by the at least one processor, and the processor 11 may perform various appropriate actions and processes according to the computer program stored in the Read Only Memory (ROM) 12 or the computer program loaded from the storage unit 18 into the Random Access Memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the vascular plaque detection system 10 may also be stored. The processor 11, the ROM 12 and the RAM 13 are connected to each other via a bus 14. An input/output (I/O) interface 15 is also connected to bus 14.

Various components in the vascular plaque detection system 10 are connected to the I/O interface 15, including: an input unit 16 such as a keyboard, a mouse, etc.; an output unit 17 such as various types of displays, speakers, and the like; a storage unit 18 such as a magnetic disk, an optical disk, or the like; and a communication unit 19 such as a network card, modem, wireless communication transceiver, etc. The communication unit 19 allows the vascular plaque detection system 10 to exchange information/data with other devices via a computer network such as the internet and/or various telecommunication networks.

The processor 11 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various processors running machine learning model algorithms, digital Signal Processors (DSPs), and any suitable processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described below, such as a vascular plaque detection method.

Fig. 2A is a flowchart of a blood vessel plaque detection method according to an embodiment of the present invention, where the method may be performed by a blood vessel plaque detection device, and the blood vessel plaque detection device may be implemented in hardware and/or software, and the blood vessel plaque detection device may be configured in a processor of a blood vessel plaque detection system. As shown in fig. 2B, the method includes:

s110, projecting blood vessels in the image to be processed along at least one projection direction to obtain at least one blood vessel projection image.

The image to be processed is a medical image including blood vessels, such as a CT (Computed Tomography, i.e., electronic computed tomography) image or an MRI (Magnetic Resonance Imaging, abbreviated as MRI) image including blood vessels, or the like. In this embodiment, a detailed description of the technical scheme is given by taking a cardiac CT image including a coronary vessel as an example.

In one embodiment, the set direction is taken as a starting projection direction, and one projection direction is determined every 10 degrees based on the starting projection direction, so as to obtain at least one projection direction for image projection. This embodiment determines the setting direction by the following means including: the setting direction is determined in response to a setting direction selection operation by the user. Illustratively, the user selects a target point on the cardiac CT image, and the processor takes a normal vector of the target point as the set direction if the target point selection operation is detected.

In one embodiment, for each projection direction, a projection view angle range is determined, and a blood vessel image in the projection view angle range in the image to be processed is projected to obtain a blood vessel projection image in the corresponding projection direction. For example, the projection direction is the opposite direction of the normal vector of the image to be processed, that is, the opposite direction of the normal vector corresponding to the zero-degree viewing angle, and the projection viewing angle range is 10 degrees, then the blood vessel image in the range from minus 5 degrees to plus 5 degrees in the image to be processed is projected to obtain the corresponding blood vessel projection image.

In one embodiment, the at least one vessel projection image is determined by the steps comprising:

And a step a1 of reconstructing blood vessels in the image to be processed based on a curved surface reconstruction method to obtain a target image.

And a step a2 of projecting blood vessels in the target image along at least one projection direction by adopting an equidistant projection method so as to obtain at least one blood vessel projection image.

The curved surface reconstruction method ((curved planar reformatio, abbreviated as CPR) is a process of resampling and visualizing a dataset from advanced information obtained from a coronal centerline detection process, a method commonly used for visualizing small diameter structures in medical CT images (face-CPR)

The target image comprises a blood vessel after curved surface reconstruction.

The equidistant projection refers to a projection method in which the center line of the blood vessel (blood vessel in the blood vessel projection image) after the projection is equidistant, see fig. 2B.

S120, determining plaque segmentation results of each blood vessel projection image, wherein the plaque segmentation results comprise calcified spots and/or non-calcified spots.

Plaque-segmentation operations are performed on each vessel projection image to obtain plaque-segmentation results for each vessel projection image. Wherein the plaque-dividing operation may be a plaque-dividing operation for only calcified plaque, where the plaque-dividing result includes only calcified plaque; the plaque-dividing operation may be a plaque-dividing operation for only non-calcified plaque, in which case the plaque-dividing result includes only non-calcified plaque; the plaque segmentation operation can also be performed on calcified plaques and non-calcified plaques at the same time, and the plaque segmentation result comprises the calcified plaques and the non-calcified plaques at the same time.

The step is based on a two-dimensional vessel projection image to obtain a region of interest, unlike a simple vessel center line, the segmentation of the vessel projection image is not a strong class defining the region of interest, but is directed at the classification of pixel points on the whole vessel projection image (the segmentation is to classify each pixel of a graph), and the biggest benefit of the method is that if a certain region of interest is not completely segmented, but only the plaque conforming to a threshold area is segmented, the plaque of the region can be segmented as well.

In one embodiment, the plaque segmentation results are the center points of the corresponding image blocks, which are the intermediate points of the vessel centerlines of the corresponding vessel segments.

In one embodiment, for each patch in the patch division result, determining whether a current patch meets a set pixel number condition; if yes, reserving the current plaque; and if not, deleting the current plaque from the plaque segmentation result to update the plaque segmentation result. In this embodiment, for any plaque in the plaque segmentation result corresponding to any vessel projection image, only when the number of pixels included in the plaque segmentation result is greater than the set pixel number threshold, the plaque is considered to be a true plaque, and otherwise, the plaque is considered to be a false plaque. And the accuracy of plaque detection is guaranteed.

In one embodiment, for each patch in the initial patch identification result corresponding to each data block, determining whether the current patch meets a set pixel number condition; if yes, reserving the current plaque; if not, deleting the current plaque from the initial plaque segmentation result to update the initial plaque segmentation result. In this embodiment, for any plaque in the initial plaque segmentation result corresponding to any data block, only when the number of pixels included in the plaque is greater than the set pixel number threshold, the plaque is considered to be a true plaque, otherwise, the plaque is considered to be a false plaque. And the accuracy of plaque detection is guaranteed.

In one embodiment, updating of the vessel segmentation result is accomplished by the steps comprising:

step b1, determining a current unlabeled plaque in a current blood vessel segmentation result based on a set plaque reading sequence aiming at plaque segmentation results of each blood vessel projection image;

and b2, determining whether a blood vessel segmentation result comprising the current unlabeled plaque exists in other blood vessel segmentation results, if not, executing b3, and if so, executing b4.

Determining whether the blood vessel segmentation result including the current unlabeled plaque exists in other blood vessel segmentation results can be understood as determining whether the blood vessel segmentation result including the unlabeled plaque which is in the same blood vessel section as the current unlabeled plaque exists in other blood vessel segmentation results.

Step b3, adding a second mark to the current unlabeled plaque, and returning to the step of determining the current unlabeled plaque in the current blood vessel segmentation result based on the set plaque reading sequence if the current unlabeled plaque is not the last unlabeled plaque in the current blood vessel segmentation result; and if the current unlabeled plaque is the last unlabeled plaque in the current blood vessel segmentation result, executing a step b6.

If the blood vessel segmentation result including the current unlabeled plaque does not exist in the other blood vessel segmentation results, the current unlabeled plaque only appears in the current blood vessel segmentation result and does not accord with the set quantity condition, and therefore a second identifier is added to the current unlabeled plaque. The second indication indicates that for a certain segment of blood vessels only one blood vessel projection image is identified with plaque. In this embodiment, the plaque corresponding to the second identifier is regarded as an invalid plaque.

And b4, adding a first mark to the current unlabeled plaque in the blood vessel segmentation result comprising the current unlabeled plaque, and determining the number of the plaque segmentation results comprising the current unlabeled plaque.

The first identifier indicates that, for a certain vessel segment, a plaque is identified in a plurality of vessel projection images, and the number of the vessel projection images may be smaller than or greater than a set number. I.e. the plaque corresponding to the first identification in this step may be a valid plaque or an invalid plaque.

Step b5, adding effective plaque identification to the current unlabeled plaque in the corresponding plaque segmentation result under the condition that the quantity meets the set quantity condition; and if the number does not meet the set number condition and the current unlabeled plaque is not the last plaque in the current plaque segmentation result, returning to the step of determining the current unlabeled plaque in the current blood vessel segmentation result based on the set plaque reading sequence.

If the number of the plaque segmentation results including the current unlabeled plaque is greater than or equal to the set number, the current unlabeled plaque in all the plaque segmentation results including the current unlabeled plaque is indicated to be an effective plaque, and therefore an effective plaque identification is added to the current unlabeled plaque in the corresponding plaque segmentation result.

It can be appreciated that if the number of plaque segmentation results including the current unlabeled plaque is smaller than the set number, it indicates that the current unlabeled plaque in all plaque segmentation results including the current unlabeled plaque is an invalid plaque.

And b6, deleting non-valid plaques in each plaque segmentation result to update each plaque segmentation result, wherein the non-valid plaques are plaques which are not marked with valid plaque identifications.

And deleting non-valid plaques in each plaque segmentation result to update each plaque segmentation result under the condition that each plaque segmentation result does not comprise unlabeled plaques.

And S130, based on a set rule, carrying out back projection on the plaque segmentation result of each vessel projection image along the projection direction corresponding to each vessel projection image so as to obtain a back projection result.

For each plaque segmentation result, based on a set rule, performing back projection processing on the plaque segmentation result along a projection direction corresponding to a blood vessel projection image corresponding to the plaque segmentation result so as to obtain a back projection result.

The back projection result may be understood as a mapping result obtained by mapping the plaque segmentation result to the image to be processed, so that each plaque in the plaque segmentation result may be displayed in the image to be processed.

In one embodiment, a pixel point identifier of a center point of an image block on a corresponding blood vessel center line in an image to be processed is determined, and the pixel point identifier is used as a plaque position identifier.

And S140, determining plaque detection results of the image to be processed according to all the back projection results.

Each back projection result corresponds to a projection direction, i.e. to a projection view angle. And under the condition that the projection view angle interval is larger than 180 degrees, overlapping parts exist in the back projection results corresponding to the two adjacent projection view angles. For this reason, the present embodiment determines a plaque combination in the same connected domain, and takes the plaque combination as a target plaque; and taking all target plaques and the residual isolated plaques as plaque detection results.

In one embodiment, for each plaque type, a plaque combination in the same connected domain is determined, and the plaque combination is used as a target plaque of the corresponding plaque type; taking all target plaques and the residual isolated plaques as plaque detection results of corresponding plaque types; and taking the plaque detection results of all plaque types as final plaque detection results.

In one embodiment, the plaque detection result of the image to be processed is determined according to all plaque position identifiers in the image to be processed.

Fig. 3A is a flowchart of a blood vessel plaque detection method according to an embodiment of the present invention, where the determination manner of plaque segmentation results in the foregoing embodiment is further refined. As shown in fig. 3A, the method includes:

S210, projecting blood vessels in the image to be processed along at least one projection direction to obtain at least one blood vessel projection image.

S220, inputting the at least one blood vessel projection image into a trained plaque segmentation result to obtain a plaque segmentation result of each blood vessel projection image, wherein the plaque segmentation result comprises calcified spots and/or non-calcified spots.

This step aims at determining the plaque segmentation result of the at least one vessel projection image by means of machine learning. In one embodiment, the implementation steps are as follows:

step c1, inputting the at least one blood vessel projection image into a trained first plaque segmentation model to obtain calcification plaque segmentation results of each blood vessel projection image.

And c2, inputting the at least one blood vessel projection image into a trained second plaque segmentation model to obtain a non-calcified plaque segmentation result of each blood vessel projection image.

In one embodiment, the first plaque segmentation model employs a neural network of U-Net structure. Compared with surrounding pixels, the brightness of the calcified spots is more prominent, and the large brightness difference enables the calcified spot segmentation with higher accuracy to be realized by adopting a classical U-Net with jump connection, a U-shaped structure and dichios.

In one embodiment, the second plaque segmentation model employs a neural network of U-Net++ structure.

In one embodiment, based on a projection angle corresponding to the vascular projection image, converting a first expression of the vascular projection image in a rectangular coordinate system to a second data expression of the vascular projection image in a polar coordinate system; inputting a second data expression of the at least one blood vessel projection image into a trained second plaque segmentation model to obtain non-calcified plaque segmentation results of each blood vessel projection image.

Specifically, a rectangular coordinate system as shown in fig. 3B and a polar coordinate system as shown in fig. 3C. The polar coordinate conversion means converting a point above the rectangular coordinate system to a point in the polar coordinate system. The coordinate points in the polar coordinate system are one more set of angle information relative to the rectangular coordinate system. For the polar coordinate system, the rows and columns are M and N respectively; each row corresponds to each radius on the rectangular coordinate system, and a length scaling factor, namely M/R exists in the radial direction. The circumferential direction is divided into N halves, i.e. the length scale factor delta_t=2pi/N. The addition of angle information can further enhance the extraction of non-calcified plaque features. Such as more refined ranges of segmentation. Since the projection angle (length scale factor) corresponding to each blood vessel projection image is known, the polar coordinate conversion can be performed on each blood vessel projection image based on the projection angle corresponding to the blood vessel projection image and the first expression of the blood vessel projection image in the rectangular coordinate system, so as to obtain the second data expression of each blood vessel projection image in the polar coordinate system. Since the second data expression includes angle information, determining the non-plaque-segmentation result based on the second data expression of each blood vessel projection image can improve the accuracy of the non-plaque-segmentation result.

And S230, based on a set rule, carrying out back projection on the plaque segmentation result of each vessel projection image along the projection direction corresponding to each vessel projection image so as to obtain a back projection result.

S240, determining plaque detection results of the image to be processed according to all the back projection results.

The embodiment of the invention determines the plaque segmentation result of each vessel projection image based on the trained plaque segmentation model, and is simple and quick.

Fig. 4 is a flowchart of a method for detecting a vascular plaque according to an embodiment of the present invention, where the determination manner of the plaque segmentation result in the foregoing embodiment is further refined. As shown in fig. 4, the method includes:

s310, projecting blood vessels in the image to be processed along at least one projection direction to obtain at least one blood vessel projection image.

S3201, based on the pixel identifications of the blood vessel center lines of the blood vessel projection images, cutting the blood vessels of the blood vessel projection images into data blocks with set sizes, wherein the data blocks carry the image identifications of the corresponding blood vessel projection images and the pixel identifications of the corresponding blood vessel center lines.

The pixel identifier of the blood vessel centerline is understood to be the sequential identifier of the pixels on the blood vessel centerline along the set blood vessel direction, for example, pixel 0001, pixel 0002, pixel 0003, etc.

Wherein the set dimensions may be configured to be 24×24, 32×32, 48×48, 64×64, etc. dimensions.

Wherein the pixel identifier of the data block is the pixel identifier of the starting pixel, the pixel identifier of the middle pixel or the pixel identifier of the ending pixel of the blood vessel center line included in the data block. In one embodiment, the length of the data block in the direction of the center of the blood vessel is an odd number of pixels, and the corresponding pixel is identified as the identification of the middle pixel of the center line of the blood vessel included by the corresponding pixel.

It will be appreciated that the longer the length of the vessel centerline in the vessel projection image, the greater the number of data blocks it corresponds to.

The blood vessel projection image corresponding to the data block can be determined according to the image identification carried by the data block, and the position of the data block on the blood vessel central line can be determined according to the pixel identification carried by the data block. The specific location of the data block can be determined from the image identification and pixel identification carried by the data block.

S3202, splicing the data blocks corresponding to the same pixel identification to obtain a data block set.

Determining the current pixel identification and the data blocks corresponding to the current pixel identification, and then splicing all the data blocks corresponding to the current pixel identification together based on the set sequence to obtain a data block set. The setting sequence is selected from the increasing direction of the projection angle or the decreasing direction of the projection angle. This step aims to maintain the locality and integrity of the same vessel segment, and fully utilizes the vessel characteristics of the vessel at each projection angle, because plaque is generally distributed in a partial region of a certain vessel segment, and no dead angle coverage exists in 360 degrees.

By way of example, the projection angle intervals are 10 degrees, i.e. one vessel projection image is generated every 10 degrees, and a total of 36 vessel projection images are generated. Taking the pixel identification as N and the data block size as (h, w) as an example, determining the data block with the pixel identification as N in each vessel projection image, and then splicing all the data blocks with the pixel identification as N together to obtain a data block set, wherein the data block set has the size as (36, h, w), and N is a positive integer.

S3303, inputting all data block sets into the trained plaque segmentation model to obtain plaque segmentation results of each vessel projection image.

Because the number of the blood vessel projection images is limited, the number of all the data block sets is smaller than or equal to the data volume of all the blood vessel projection images, the data volume of the at least two blood vessel projection images is far smaller than the data volume of the three-dimensional blood vessel, so that the data volume of all the data block sets is far smaller than the data volume of the three-dimensional blood vessel, and therefore, the trained plaque segmentation model is adopted to analyze all the data block sets, and the plaque segmentation speed and the plaque segmentation precision are higher.

In one embodiment, plaque segmentation results are determined by the steps comprising:

And d1, inputting all the data blocks into the trained first plaque segmentation model to obtain calcification plaque segmentation results of each blood vessel projection image.

And d2, inputting all the data blocks into a trained second plaque segmentation model to obtain a non-calcified plaque segmentation result of each blood vessel projection image.

This embodiment aims at employing a data block as input data of a plaque segmentation model to increase the data processing speed of the plaque segmentation model, thereby increasing the plaque segmentation speed.

In one embodiment, the non-calcified plaque segmentation process is optimized by the steps comprising:

and e1, performing polar coordinate conversion on all the data blocks to obtain all the updated data blocks.

And e2, inputting all updated data blocks into a trained second plaque segmentation model to obtain a non-calcified plaque segmentation result of each blood vessel projection image.

It should be noted that, the structures of the first plaque-segment model and the second plaque-segment model refer to the foregoing embodiment, and the polar coordinate conversion method of the data block refers to the polar coordinate conversion method of the vessel projection image in the foregoing embodiment.

S330, based on a set rule, carrying out back projection on the plaque segmentation result of each vessel projection image along the projection direction corresponding to each vessel projection image so as to obtain a back projection result.

S340, determining plaque detection results of the image to be processed according to all the back projection results.

The embodiment of the invention combines the projection data of the same blood vessel segment in each projection angle by adopting a data block splicing mode, so that the trained plaque segmentation model can simultaneously analyze the projection data of the corresponding blood vessel segment in each projection direction, thereby improving the plaque segmentation speed of the trained plaque segmentation model.

Fig. 5 is a flowchart of a method for detecting a vascular plaque according to an embodiment of the present invention, where the method for determining a plaque detection result in the foregoing embodiment is further limited. As shown in fig. 5, the method includes:

s410, projecting blood vessels in the image to be processed along at least one projection direction to obtain at least one blood vessel projection image.

S4201, determining a plaque segmentation result of each of the vessel projection images, wherein the plaque segmentation result includes calcified plaque and/or non-calcified plaque.

S4202, aiming at any one of the blood vessel projection images, determining overlapping plaques of the calcification plaque segmentation result and the non-calcification plaque segmentation result, and taking the overlapping plaques as mixed plaques.

Since the mixed plaque has characteristics of both calcified plaque and non-calcified plaque, it may appear simultaneously in the calcified plaque segmentation result and non-calcified plaque segmentation result. For this reason, the present embodiment determines overlapping plaques of the calcified plaque segmentation result and the non-calcified plaque segmentation result, and takes the overlapping plaques as mixed plaques. The purpose of determining the mixed plaque corresponding to each vessel projection image in an indirect mode is achieved.

S4203, deleting the overlapped plaque from the calcification plaque segmentation result to update the calcification plaque segmentation result; deleting the coincident plaque from the non-calcified plaque segmentation results to update the non-calcified plaque segmentation results.

It will be appreciated that the updated calcified plaque segmentation results include only calcified plaque and the updated non-calcified plaque segmentation results include only non-calcified plaque.

S4204, using the updated calcified plaque segmentation result, the updated non-calcified plaque segmentation result, and all the mixed plaque as new plaque segmentation results.

S430, based on a set rule, carrying out back projection on the plaque segmentation result of each vessel projection image along the projection direction corresponding to each vessel projection image so as to obtain a back projection result.

S440, determining plaque detection results of the image to be processed according to all the back projection results.

The plaque detection results include calcification plaque detection results, non-calcification plaque detection results and mixed plaque detection results.

In one embodiment, plaque detection results of the image to be processed are displayed in a visual mode, and each plaque in the detection results is marked with a corresponding plaque type.

According to the embodiment, the mixed plaque corresponding to each blood vessel projection image is determined by determining the calcification plaque segmentation result corresponding to each blood vessel projection image and the overlapping plaque in the non-calcification plaque segmentation result, so that the purpose of determining the mixed plaque corresponding to the blood vessel projection image by an indirect method is achieved.

Fig. 6 is a flowchart of a method for detecting a vascular plaque according to an embodiment of the present invention, where the method for determining a plaque detection result in the foregoing embodiment is further limited. As shown in fig. 6, the method includes:

s510, projecting blood vessels in the image to be processed along at least one projection direction to obtain at least one blood vessel projection image.

S520, determining plaque segmentation results of each blood vessel projection image, wherein the plaque segmentation results comprise calcified spots and/or non-calcified spots.

And S530, based on a set rule, carrying out back projection on the plaque segmentation result of each vessel projection image along the projection direction corresponding to each vessel projection image so as to obtain a back projection result.

S5401, in a case where the projection angle interval is smaller than the set angle threshold, for each plaque type, performs plaque communication recognition processing on each of the back projection results to obtain a plaque communication recognition result.

The set angle threshold in this embodiment may be configured as a modifiable item, set by the user according to actual needs, for example, 10 degrees or 5 degrees.

This step aims at determining which blobs in the vessel projection images are isolated and which are connected by the plaque combining step. It will be appreciated that in the case where the projection angle interval is sufficiently small, any pixel of a blood vessel will be projected to at least two blood vessel projection images, so when a plaque is identified in any one of the blood vessel projection images and no plaque is identified in any of the adjacent images of the blood vessel projection images, the plaque identified in the blood vessel projection image can be considered as a false plaque; conversely, if a patch is identified in both of the adjacent vessel projection images and all or part of the identified patch is located in the overlapping region of the view angles of the adjacent vessel projection images, then the patch identified by the adjacent vessel projection images may be considered valid.

Exemplary, the view angle range corresponding to the patch a is 10 degrees to 15 degrees, the view angle range corresponding to the patch B is 14 degrees to 17 degrees, the view angle range corresponding to the patch C is 17 degrees to 25 degrees, and the view angle range corresponding to the patch D is 30 degrees to 32 degrees. Since plaque a and plaque B have an overlapping viewing angle of 14 degrees to 15 degrees, the two are connected; since the projection view angles of the plaque B and the plaque C comprise the same boundary view angle of 17 degrees, namely the two are connected by the view angles, the two are also connected; since both the plaque a and the plaque C communicate with the plaque B, the plaque a and the plaque B communicate with the plaque C. Since plaque a, plaque B and plaque C are all free of overlapping viewing angles and are not connected to plaque D, plaque D is isolated.

S5402, determining the number of vessel projection image identifiers corresponding to the connected domains in each plaque connected identification result.

Determining whether the plaques are connected or overlapped or not in the back projection result corresponding to each plaque type through plaque communication identification processing; if so, the connected or overlapped plaque is indicated to be substantially corresponding to one plaque, and the determination of the corresponding relation between the real plaque and the blood vessel projection image is simplified through the label of the blood vessel projection image corresponding to the connected domain.

S5403, deleting the connected domains with the number smaller than the set threshold value to update the corresponding plaque connected recognition result.

For any of the connected domains, if the number of its corresponding vessel projection images is smaller than a set threshold, such as 2, it means that its corresponding plaque is projected into 1 vessel projection image. When the projection interval is smaller than the set angle threshold, it can be considered that only the plaque that appears in one blood vessel projection image is a false plaque, that is, a plaque that does not exist. For this reason, the present embodiment deletes connected domains whose number of corresponding vessel projection image identifications is smaller than a set threshold value, so as to update the corresponding plaque connected identification result.

S5404, determining plaque detection results of the image to be processed according to all updated plaque communication recognition results.

It can be understood that the updated plaque connectivity identification result includes plaque that meets the set plaque condition, so that the plaque detection result of the image to be processed is determined according to all the updated plaque connectivity identification results.

According to the embodiment of the invention, through judging the size relation between the number of the vascular projection images corresponding to each plaque in the plaque communication recognition result and the set threshold, the true and false of the judged plaque are deleted, the true plaque is reserved, and the plaque detection result of the image to be processed is flexibly and accurately determined.

Fig. 7A is a flowchart of a blood vessel plaque detection method according to an embodiment of the present invention, where the embodiment is used for explaining the back projection process in the foregoing embodiment in detail. As shown in fig. 7A, the method includes:

s6401, determining coordinate values of target pixels on each plaque segmentation result, wherein the target pixels are pixels on a blood vessel central line.

In one embodiment, the coordinate values of the target pixels on each plaque-segmentation result are determined by:

and f1, determining a central line of the plaque segmentation result in the second direction, wherein the central line is a line segment of an initial central line of the image to be processed in the second direction projected onto the blood vessel projection image, and the number of pixels of the blood vessel central line is the same as that of the pixels of the initial projection central line.

The second coordinate value of the initial midline of the image [ h, w ] to be processed in the second direction is h/2.

In one embodiment, as shown in fig. 7B, the second direction is the ordinate axis direction. The midline is thus parallel to the abscissa axis direction. In the case where the image to be processed can be expressed as [ h, w ], the second coordinate value of the centerline of the blood vessel projection image is h/2.

And f2, determining a second coordinate value of each target pixel according to the distance between each target pixel of each vessel projection image and the central line.

The distance between each target pixel and the center line is determined, and if the distance table is d, the second coordinate value of each target pixel may be expressed as h/2±d.

And f3, determining pixel identifications of target pixels of the blood vessel projection images, and taking the pixel identifications as first coordinate values of the target pixels, wherein the first coordinate directions corresponding to the first coordinate values are perpendicular to the second coordinate directions corresponding to the second coordinate values.

And determining pixel identifiers of target pixels of each vessel projection image according to a preset first coordinate axis direction, wherein the pixel identifiers are sequence identifiers of the pixels. And taking the pixel mark of the target pixel as a first coordinate value of the target pixel, wherein the first coordinate direction corresponding to the first coordinate value is perpendicular to the second coordinate direction corresponding to the second coordinate value.

It will be appreciated that in the case where the second coordinate direction corresponds to the ordinate axis direction, the first coordinate direction corresponds to the abscissa axis direction.

S6402, for any target pixel in each plaque-segmentation result, back-projecting the coordinate value of the target pixel along the projection direction corresponding to the blood vessel projection image to obtain an initial back-projection result.

A back projection process is understood to be the process of mapping the coordinate values of the target pixels back to the image to be processed.

S6403, determining a pixel nearest to the initial back projection result in the image to be processed based on a nearest neighbor principle, and projecting the target pixel to the nearest neighbor pixel.

It will be understood that, for any target pixel, when the coordinate value is mapped to the image to be processed, there may be no corresponding pixel in the mapped coordinate value, and for this embodiment, the coordinate value is mapped to the corresponding nearest pixel based on the nearest neighbor principle. Exemplary embodiments. The coordinate value of the target pixel A mapped to the image to be processed is M, and the pixel point nearest to the coordinate value M in the image to be processed is determined; if the nearest neighbor pixel point is B, mapping the target pixel A to the pixel B of the image to be processed.

S6404, taking all nearest neighbor pixels corresponding to all target pixels of the plaque-segmentation result as a back projection result of the plaque-segmentation result.

In this embodiment, the plaque segmentation result is mapped to the image to be processed based on the nearest neighbor principle, and since the resolution of the image to be processed is relatively high, the Xu Jingdu errors caused by the nearest neighbor principle are clinically acceptable, so that the plaque detection result determined based on the back projection result can be used for clinical diagnosis.

Fig. 8 is a schematic structural diagram of a vascular plaque detection device according to an embodiment of the present invention. As shown in fig. 8, the apparatus includes:

a projection module 710, configured to project a blood vessel in the image to be processed along at least one projection direction, so as to obtain at least one blood vessel projection image;

a segmentation module 720 for determining plaque segmentation results of each of the vessel projection images, the plaque segmentation results including calcification spots and/or non-calcification spots;

the back projection module 730 is configured to back-project the plaque-segmentation result of each of the vessel projection images along the projection direction corresponding to each of the vessel projection images based on a set rule to obtain a back-projection result;

and a detection result module 740, configured to determine a plaque detection result of the image to be processed according to all the back projection results.

In one embodiment, the projection module 710 is specifically configured to:

reconstructing blood vessels in the image to be processed based on a curved surface reconstruction method to obtain a target image;

and adopting an equidistant projection method to project the blood vessels in the target image along at least one projection direction so as to obtain at least one blood vessel projection image.

In one embodiment, the segmentation module is specifically configured to:

Inputting the at least one blood vessel projection image into the trained plaque segmentation results to obtain plaque segmentation results of each blood vessel projection image.

In one embodiment, the segmentation module 720 is configured to:

the clipping unit is used for clipping the blood vessel of each blood vessel projection image into a data block with a set size based on the pixel identification of the blood vessel central line of each blood vessel projection image, wherein the data block carries the image identification of the corresponding blood vessel projection image and the pixel identification of the corresponding blood vessel central line;

the splicing unit is used for splicing the data blocks corresponding to the same pixel identification to obtain a data block set;

and the segmentation unit is used for inputting all the data block sets into the trained plaque segmentation model to obtain plaque segmentation results of each blood vessel projection image.

In one embodiment, the segmentation unit comprises:

the first segmentation subunit is used for inputting all the data blocks into the trained first plaque segmentation model to obtain calcification plaque segmentation results of each blood vessel projection image;

and the second segmentation subunit is used for inputting all the data blocks into the trained second plaque segmentation model to obtain a non-calcified plaque segmentation result of each blood vessel projection image.

In one embodiment, the second segmentation subunit is configured to:

performing polar coordinate conversion on all the data blocks to obtain all the updated data blocks;

and inputting all updated data blocks into a trained second plaque segmentation model to obtain non-calcified plaque segmentation results of each blood vessel projection image.

In one embodiment, the back projection module 730 is configured to:

determining a current unlabeled plaque in the current blood vessel segmentation result based on a set plaque reading sequence aiming at plaque segmentation results of each blood vessel projection image;

determining whether a blood vessel segmentation result comprising the current unlabeled plaque exists in other blood vessel segmentation results;

if so, adding a first mark to the current unlabeled plaque in the blood vessel segmentation results comprising the current unlabeled plaque, and determining the number of plaque segmentation results comprising the current unlabeled plaque;

under the condition that the number meets the set number condition, adding effective plaque identifications to the current unlabeled plaque in the corresponding plaque segmentation result, wherein the non-effective plaque is a plaque not labeled with the effective plaque identifications;

and deleting non-valid plaques in each plaque segmentation result to update each plaque segmentation result.

In one embodiment, the detection result module 740 is configured to:

under the condition that the projection angle interval is smaller than a set angle threshold, aiming at each plaque type, plaque communication recognition processing is carried out on each back projection result so as to obtain plaque communication recognition results;

determining the number of vessel projection image identifications corresponding to all the plaques in each plaque communication identification result;

deleting the plaques with the number smaller than a set threshold value to update the corresponding plaque communication recognition result;

and determining the plaque detection result of the image to be processed according to all updated plaque communication recognition results.

In one embodiment, the back projection module comprises:

a coordinate value determining unit configured to determine a coordinate value of a target pixel on each of the plaque-dividing results, the target pixel being a pixel on a blood vessel center line;

the back projection unit is used for carrying out back projection on the coordinate value of any target pixel in each plaque segmentation result along the projection direction corresponding to the blood vessel projection image so as to obtain an initial back projection result;

the nearest neighbor unit is used for determining a pixel nearest to the initial back projection result in the image to be processed based on a nearest neighbor principle, and projecting the target pixel to the nearest neighbor pixel;

And the back projection result determining module is used for taking all nearest neighbor pixels corresponding to all target pixels of the plaque-segmentation result as a back projection result of the plaque-segmentation result.

In one embodiment, the coordinate value determining unit is specifically configured to:

determining a central line of the blood vessel projection image in a second direction, wherein the central line is a line segment of an initial central line of the image to be processed in the second direction projected onto the blood vessel projection image, and the number of pixels of the blood vessel central line is the same as that of pixels of the projection central line;

determining a second coordinate value of each target pixel according to the distance between each target pixel of each vessel projection image and the central line;

and determining the pixel identification of the target pixel of each vessel projection image, and taking the pixel identification as a first coordinate value of the target pixel, wherein the first coordinate direction corresponding to the first coordinate value is perpendicular to the second coordinate direction corresponding to the second coordinate value.

The vascular plaque detection device provided by the embodiment of the invention can execute the vascular plaque detection method provided by any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

In some embodiments, the vascular plaque detection method may be implemented as a computer program tangibly embodied on a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 10 via the ROM 12 and/or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the vascular plaque detection method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the vascular plaque detection method by any other suitable means (e.g. by means of firmware).

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

A computer program for carrying out methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions/acts specified in the flowchart and/or block diagram block or blocks to be implemented. The computer program may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of the present invention, a computer-readable storage medium may be a tangible medium that can contain, or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer readable storage medium may be a machine readable signal medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which a user can provide input to the electronic device. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), blockchain networks, and the internet.

The computing system may include clients and servers. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service are overcome.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps described in the present invention may be performed in parallel, sequentially, or in a different order, so long as the desired results of the technical solution of the present invention are achieved, and the present invention is not limited herein.

The above embodiments do not limit the scope of the present invention. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of the present invention.

Claims

1. A vascular plaque detection system comprising a processor configured to perform a vascular plaque detection method comprising:

2. The system of claim 1, wherein the projecting the blood vessel in the image to be processed along the at least one projection direction to obtain at least one blood vessel projection image comprises:

3. The system of claim 1, wherein said determining plaque segmentation results for each of said vessel projection images comprises:

4. The system of claim 1, wherein said determining plaque segmentation results for each of said vessel projection images comprises:

based on the pixel identifications of the blood vessel center lines of the blood vessel projection images, cutting blood vessels of the blood vessel projection images into data blocks with set sizes, wherein the data blocks carry the image identifications of the corresponding blood vessel projection images and the pixel identifications of the corresponding blood vessel center lines;

splicing the data blocks corresponding to the same pixel identification to obtain a data block set;

and inputting all the data block sets into a trained plaque segmentation model to obtain plaque segmentation results of each blood vessel projection image.

5. The system of claim 4, wherein said inputting all data blocks into the trained plaque segmentation model to obtain plaque segmentation results for each of said vessel projection images comprises:

Inputting all the data blocks into a trained first plaque segmentation model to obtain calcification plaque segmentation results of each blood vessel projection image;

inputting all the data blocks into a trained second plaque segmentation model to obtain non-calcified plaque segmentation results of each of the vessel projection images.

6. The system of claim 5, wherein said inputting all data blocks into the trained second plaque segmentation model to obtain non-calcified plaque segmentation results for each of the vessel projection images comprises:

7. The system of claim 4, wherein inputting the set of all data blocks into the trained plaque segmentation model to obtain plaque segmentation results for each of the vessel projection images comprises:

determining whether the current plaque accords with a set pixel quantity condition for each plaque in the plaque segmentation result;

if yes, reserving the current plaque;

and if not, deleting the current plaque from the plaque segmentation result to update the plaque segmentation result.

8. The system of claim 1, wherein before back-projecting the plaque-segmentation result of each of the vessel projection images along the projection direction corresponding to each of the vessel projection images based on the set rule to obtain a back-projected result, further comprising:

9. The system according to claim 1, wherein each plaque in the back-projection results carries a corresponding vessel projection image identifier, and the determining the plaque detection result of the image to be processed according to all back-projection results includes:

10. The system of claim 1, wherein back-projecting the plaque-segmentation result of each of the vessel projection images along a projection direction corresponding to each of the vessel projection images based on a set rule to obtain a back-projected result comprises:

determining coordinate values of target pixels on each plaque segmentation result, wherein the target pixels are pixels on a blood vessel central line;

for any target pixel in each plaque-segmentation result, carrying out back projection on the coordinate value of the target pixel along the projection direction corresponding to the blood vessel projection image so as to obtain an initial back projection result;

based on a nearest neighbor principle, determining a pixel nearest to the initial back projection result in the image to be processed, and projecting the target pixel to the nearest neighbor pixel;

And taking all nearest neighbor pixels corresponding to all target pixels of the plaque-segmentation result as a back projection result of the plaque-segmentation result.

11. The system of claim 10, wherein determining the coordinate values of the target pixel on each of the vessel projection images comprises:

determining a central line of the blood vessel projection image in a second direction, wherein the central line is a line segment of an initial central line of the image to be processed in the second direction projected onto the blood vessel projection image, and the number of pixels of the blood vessel central line is the same as that of the initial central line;

12. A vascular plaque detection apparatus, comprising:

the projection module is used for projecting the blood vessels in the image to be processed along at least one projection direction so as to obtain at least one blood vessel projection image;

A segmentation module for determining plaque segmentation results of each of the vessel projection images, the plaque segmentation results including calcified plaques and/or non-calcified plaques;

the back projection module is used for carrying out back projection on plaque segmentation results of the blood vessel projection images along the projection directions corresponding to the blood vessel projection images based on a set rule so as to obtain back projection results;

and the detection result module is used for determining plaque detection results of the image to be processed according to all the back projection results.

13. A computer readable storage medium storing computer instructions for causing a processor to perform the following method: