Disclosure of Invention
In order to solve the above problems in medical CT image processing, embodiments of the present invention creatively provide a medical image detection method, a medical image detection device, and a computer-readable storage medium.
According to a first aspect of the invention, there is provided a medical image detection method, the method comprising: acquiring a lung Computed Tomography (CT) image; according to the lung CT image, carrying out image segmentation to obtain an image segmentation result comprising a lung blood vessel mask and a pneumonia focus area mask; according to the segmentation result, judging the blood vessel existing form of the pulmonary blood vessel corresponding to the pulmonary blood vessel mask and the focus existing form of the pneumonia focus area corresponding to the pneumonia focus area mask; and optimizing the image segmentation result according to the lung CT image, the existing form of the blood vessel and the existing form of the focus.
According to an embodiment of the present invention, an image segmentation is performed according to the lung CT image to obtain an image segmentation result including a lung blood vessel and a pneumonia lesion region, and the image segmentation result includes: and segmenting the lung CT image by adopting a deep learning algorithm to obtain a lung blood vessel mask and a pneumonia focus region mask.
According to an embodiment of the present invention, the determining, based on the segmentation result, a blood vessel existing form of a pulmonary blood vessel corresponding to the pulmonary blood vessel mask and a lesion existing form of a pneumonia lesion region corresponding to the pneumonia lesion region mask includes: judging whether the blood vessel volume of the pulmonary blood vessel corresponding to each pulmonary blood vessel mask is smaller than a set threshold value or not according to the pulmonary blood vessel masks, and judging that the blood vessel form of the pulmonary blood vessel corresponding to the pulmonary blood vessel mask is free when the blood vessel volume is smaller than the set threshold value; and judging whether a continuum exists in the pneumonia focus area corresponding to the pneumonia focus area mask according to the pneumonia focus area mask.
According to an embodiment of the present invention, optimizing the image segmentation result according to the lung CT image, the blood vessel existing form and the lesion existing form includes: when the pneumonia focus area is judged to have a continuum, performing external expansion on the pneumonia focus area according to the lung CT image to obtain an external expansion area; judging whether a blood vessel exists in the external expansion area; when a blood vessel exists in the flaring region, judging whether the blood vessel in the flaring region is connected with the pneumonia focus region; and if the blood vessel in the flaring region is connected with the pneumonia focus region, re-determining the pneumonia focus region in the flaring region by utilizing an optimization model.
According to an embodiment of the present invention, optimizing the image segmentation result according to the lung CT image, the blood vessel existing form and the lesion existing form includes: and when the blood vessel form of the pulmonary blood vessel is judged to be free, determining a region corresponding to the pulmonary blood vessel in the pulmonary CT image as a pneumonia focus region.
According to a second aspect of the present invention, there is also provided a medical image detection apparatus, the apparatus comprising: the acquisition module is used for acquiring a lung Computed Tomography (CT) image; the segmentation module is used for carrying out image segmentation according to the lung CT image to obtain an image segmentation result comprising a lung blood vessel mask and a pneumonia focus area mask; the form judging module is used for judging the blood vessel existing form of the pulmonary blood vessel corresponding to the pulmonary blood vessel mask and the focus existing form of the pneumonia focus area corresponding to the pneumonia focus area mask according to the segmentation result; an optimization module for optimizing the computed tomography scan according to the lung CT image, the blood vessel existing form and the lesion existing form.
According to an embodiment of the present invention, the form determination module includes: the first form judgment submodule is used for judging whether the blood vessel volume of the pulmonary blood vessel corresponding to each pulmonary blood vessel mask is smaller than a set threshold value or not according to the pulmonary blood vessel masks, and judging that the blood vessel form of the pulmonary blood vessel corresponding to the pulmonary blood vessel masks is free when the blood vessel volume is smaller than the set threshold value; and the second form judgment submodule is used for judging whether a continuum exists in the pneumonia focus area corresponding to the pneumonia focus area mask according to the pneumonia focus area mask.
According to an embodiment of the invention, the optimization module comprises: the external expansion module is used for externally expanding the pneumonia focus area according to the lung CT image to obtain an external expansion area when the existence of a continuum in the pneumonia focus area is judged; the blood vessel judging submodule is used for judging whether blood vessels exist in the externally expanded region or not; the connection judgment submodule is used for judging whether the blood vessels in the external expansion region are connected with the pneumonia focus region or not when the blood vessels exist in the external expansion region; and the first optimization submodule is used for re-determining the pneumonia focus area in the external expansion area by utilizing an optimization model if the blood vessel in the external expansion area is connected with the pneumonia focus area.
According to an embodiment of the invention, the optimization module comprises: and the second optimization submodule is used for determining a region corresponding to the pulmonary blood vessel in the pulmonary CT image to be determined as a pneumonia focus region when the blood vessel form of the pulmonary blood vessel is judged to be free.
According to a third aspect of the present invention, there is also provided a computer-readable storage medium comprising a set of computer-executable instructions for performing any of the medical image detection methods described above when executed.
According to the medical image detection method, the medical image detection device and the computer-readable storage medium, the lung CT image is segmented by using a three-classification method, the free-type lung blood vessel is judged to be the pneumonia focus area, when the focus area has a continuum, the segmentation result of the focus area is further optimized, the interference of subsequent steps can be removed, the confusion of the lung blood vessel and the pneumonia focus area is avoided, and the focus segmentation precision is further improved.
It is to be understood that the teachings of the present invention need not achieve all of the above-described benefits, but rather that specific embodiments may achieve specific technical results, and that other embodiments of the present invention may achieve benefits not mentioned above.
Detailed Description
The principles and spirit of the present invention will be described with reference to a number of exemplary embodiments. It is understood that these embodiments are given only to enable those skilled in the art to better understand and to implement the present invention, and do not limit the scope of the present invention in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
The technical solution of the present invention is further elaborated below with reference to the drawings and the specific embodiments.
FIG. 1 is a flow chart illustrating an implementation of a medical image detection method according to an embodiment of the present invention;
referring to fig. 1, a medical image detection method according to an embodiment of the present invention at least includes the following operation flows: operation 101, acquiring a lung Computed Tomography (CT) image; operation 102, performing image segmentation according to the lung CT image to obtain an image segmentation result including a lung blood vessel mask and a pneumonia focus area mask; operation 103, according to the segmentation result, determining a blood vessel existing form of a pulmonary blood vessel corresponding to the pulmonary blood vessel mask and a focus existing form of a pneumonia focus region corresponding to the pneumonia focus region mask; and operation 104, optimizing the image segmentation result according to the lung CT image, the blood vessel existing form and the focus existing form.
In operation 101, a lung Computed Tomography (CT) image is acquired.
CT (Computed Tomography) uses precisely collimated X-ray beams, gamma rays, ultrasonic waves, etc. to scan sections of a human body one after another around a certain part of the human body together with a detector having extremely high sensitivity, has the characteristics of fast scanning time, clear images, etc., and can be used for the examination of various diseases.
The CT image acquired in the embodiment of the invention mainly refers to a lung CT image, and the operations such as image segmentation, identification and the like are carried out according to the acquired lung CT image.
In an embodiment of the present invention, the CT scanner may be directly connected to the CT scanner to acquire CT image data, or perform data sharing through a cloud storage device, or perform processing on the CT image data after storing the CT image data acquired through the CT scanner through a mobile storage device.
In operation 102, image segmentation is performed according to the lung CT image to obtain an image segmentation result including a lung blood vessel mask and a pneumonia lesion region mask.
In one embodiment of the invention, a deep learning algorithm is adopted to segment a lung CT image to obtain a lung blood vessel mask and a pneumonia focus area mask.
In an embodiment of the present invention, when segmenting the lung CT image, the lung lobe is segmented into five regions in addition to the pulmonary blood vessel mask and the pneumonia focus region mask, so as to obtain the lung lobe segmentation mask.
Here, a lung lobe partition mask is obtained, and the lung lobes may be partitioned, so that in a subsequent operation, a pneumonia focus area may be determined according to the lung lobe partition. The method is favorable for accurately positioning the pneumonia focus area, and eliminates the interference of the connection part of different lung lobe areas on image processing in the image acquisition and image segmentation processes, thereby improving the image detection precision.
For example, a deep learning algorithm may be used to segment a lung CT image. Firstly, three-classification segmentation of pulmonary blood vessels, pneumonia focus areas and lung lobes is carried out on a lung CT image. The classification segmentation network adopts a 3D deep neural network, such as: u-net or modified U-net, the Loss function of which can adopt multi-classification Dice coefficients. The Dice coefficient is a set similarity metric function. The results of segmenting the lung CT image with mask _ res of 1-7 respectively represent: the mask _ res value is 1 corresponding to a pulmonary vessel mask, the mask _ res value is 2 corresponding to a pneumonia focus area mask, and the mask _ res values are 3-7 respectively corresponding to five area masks (including upper left, lower left, upper right, middle right and lower right) of lung lobes.
When the lung CT image is processed, the image segmentation of the three classifications is firstly carried out, so that the interference in the subsequent processes of image identification and the like can be removed, and the precision of obtaining a focus region by image segmentation in the image detection process is effectively improved.
In operation 103, determining a blood vessel existing form of a pulmonary blood vessel and a lesion existing form of a pneumonia lesion region according to the segmentation result;
in an embodiment of the present invention, the blood vessel existence form of the pulmonary blood vessel corresponding to the pulmonary blood vessel mask and the lesion existence form of the pneumonia lesion region corresponding to the pneumonia lesion region mask are determined by the following steps: firstly, judging whether the blood vessel volume of the pulmonary blood vessel corresponding to each pulmonary blood vessel mask is smaller than a set threshold value or not according to the pulmonary blood vessel masks, and judging that the blood vessel form of the pulmonary blood vessel corresponding to the pulmonary blood vessel masks is free when the blood vessel volume is smaller than the set threshold value; and judging whether a continuum exists in the pneumonia focus area corresponding to the pneumonia focus area mask according to the pneumonia focus area mask.
The blood vessel form of the pulmonary blood vessel corresponding to the pulmonary blood vessel mask is free, which means that the pulmonary blood vessel corresponding to the pulmonary blood vessel mask has a small volume or is not connected with other pulmonary blood vessels, belongs to an isolated especially small region, and can be regarded as that the pneumonia focus region is mistakenly judged as the pulmonary blood vessel. Therefore, in an embodiment of the present invention, when the blood vessel shape of the pulmonary blood vessel corresponding to the pulmonary blood vessel mask is free, it is determined that a region corresponding to the pulmonary blood vessel in the pulmonary CT image is a pneumonia lesion region.
The pneumonia focus area corresponding to the pneumonia focus area mask has a continuum, which means that the pneumonia focus area corresponding to a certain pneumonia focus area mask has continuity with the pneumonia focus area corresponding to the pneumonia focus area mask adjacent to the pneumonia focus area mask. For example: the obtained lung CT image has a pneumonia focus area with a larger volume, a plurality of pneumonia focus areas are obtained in the process of image segmentation of the lung CT image, wherein two pneumonia focus area masks correspond to two adjacent pneumonia focus areas in the lung CT image, and the pneumonia focus areas corresponding to the two pneumonia focus area masks can be called to have a continuum. For example, in the segmentation process of the lung CT image, confusion is most likely to occur between the blood vessel region of the lung and the pneumonia focus region. Therefore, the process of optimizing the segmentation result in operation 102 first includes: and determining whether the region corresponding to the pulmonary blood vessel mask obtained in the image segmentation process is a pulmonary blood vessel. Since the blood vessel is a continuous and complete blood vessel, when the pulmonary blood vessel mask in the segmentation result in operation 102 is analyzed, if the pulmonary blood vessel mask is in a free form, the pulmonary blood vessel corresponding to the pulmonary blood vessel mask is necessarily abnormal, and should be adjusted to the pneumonia lesion area. Whether the pulmonary vascular mask is free or not can be judged, whether the blood vessel volume of the pulmonary blood vessel corresponding to each pulmonary vascular mask is smaller than a set threshold or not can be judged, and when the blood vessel volume is smaller than the set threshold, the blood vessel shape of the pulmonary blood vessel corresponding to the pulmonary vascular mask is judged to be free. The set threshold may be set according to actual needs, and is not particularly limited in this embodiment of the present invention. For example: the detection volume of the pulmonary blood vessels corresponding to each pulmonary blood vessel mask can be determined according to the image pixels corresponding to the pulmonary blood vessel mask, and the determined detection volume can be set as a set threshold value.
In addition, the determination of whether the pulmonary blood vessel is free can also be made according to whether other pulmonary blood vessels adjacent to the two ends of the pulmonary blood vessel corresponding to the pulmonary blood vessel mask obtained by segmentation in operation 102 are connected together.
In an embodiment of the present invention, whether a continuum exists in the pneumonia lesion area corresponding to the pneumonia lesion area mask is determined mainly according to whether a continuous pneumonia lesion area image exists in an image of each pneumonia lesion area mask and an adjacent area on the lung CT image. Here, the image detection and judgment is performed by using a 3D neural network algorithm.
In operation 104, the image segmentation result is optimized according to the lung CT image, the blood vessel existing form, and the lesion existing form.
In one embodiment of the present invention, when the blood vessel shape of the pulmonary blood vessel is determined to be free, a region corresponding to the pulmonary blood vessel in the pulmonary CT image is determined to be a pneumonia lesion region.
In one embodiment of the invention, when the pneumonia focus area is judged to have a continuum, the pneumonia focus area is subjected to external expansion according to the lung CT image to obtain an external expansion area; judging whether blood vessels exist in the external expansion area or not; when a blood vessel exists in the flaring region, judging whether the blood vessel in the flaring region is connected with the pneumonia focus region; and (4) whether the blood vessel in the flaring region is connected with the pneumonia focus region or not is determined again by utilizing an optimization model.
For example, when the pneumonia focus area is judged to have a continuum, an external expansion area is taken for the pneumonia focus area. For example: the size of the pneumonia focus area is 24 x 34 x 32, and the edge of the pneumonia focus area is expanded by 8 pixels to obtain an expanded area with the size of 32 x 50 x 48. And (3) performing two-classification identification on the extension area by using a neural network algorithm, for example: and performing two-classification identification by adopting a U-net network, wherein a Dice function is adopted. If the pulmonary blood vessel exists in the external expansion area, whether the pulmonary blood vessel in the external expansion area is connected with the space of the pneumonia focus area is further judged.
The judgment of whether the pulmonary blood vessel in the flaring region is connected with the space of the pneumonia focus region can adopt the following method: and performing expansion operation on the pneumonia focus area to determine whether the expanded pneumonia focus area is overlapped with the pulmonary blood vessel in the external expansion area. If the areas are overlapped, the pulmonary blood vessel in the external expansion area is judged to be connected with the space of the pneumonia focus area.
If the pulmonary blood vessel in the external expansion region is judged to be connected with the space of the pneumonia focus region, an optimization model is started, and the pneumonia focus region and the part of the pulmonary blood vessel are correspondingly subdivided. Specifically, CT image data of set voxels at the pneumonia lesion region and pulmonary vessel junction can be extracted, such as: cube data was extracted with a voxel size of 64 x 64 at the pneumonia lesion area and pulmonary vessel junction. And performing two-classification recognition on the extracted cube data by using a neural network algorithm, and determining whether the corresponding position in the CT image corresponding to the cube data is a pulmonary blood vessel or a pneumonia focus area.
Therefore, according to the medical image detection method, the medical image detection device and the computer-readable storage medium provided by the embodiment of the invention, the lung CT image is segmented by using a three-classification method, the free-type lung blood vessel is judged as the pneumonia focus region, and when the focus region has a continuum, the segmentation result of the focus region is further optimized, so that the interference of subsequent steps can be removed, the confusion between the lung blood vessel and the pneumonia focus region is avoided, and the focus segmentation precision is further improved.
Fig. 2 is a schematic flow chart showing an implementation example of a specific application of the medical image detection method according to the embodiment of the present invention.
Referring to fig. 2, a specific application example of the medical image detection method according to the embodiment of the present invention at least includes the following operation flows:
in operation 201, a lung Computed Tomography (CT) image is acquired.
The specific implementation method refers to operation 101, and is not described herein again.
Operation 202, according to the lung CT image, image segmentation is performed to obtain an image segmentation result including a lung blood vessel mask, a pneumonia focus area mask, and five lung lobe area masks.
The specific implementation method refers to operation 102, which is not described herein again.
In operation 203, the image segmentation result is optimized.
The specific implementation method refers to operations 103-104, where the specific optimization of the image segmentation result of the pulmonary blood vessel and the pneumonia lesion region includes the following specific operations 204-205.
And operation 204, judging whether the pulmonary blood vessels corresponding to the pulmonary blood vessel masks in the image segmentation result are discrete.
Referring to the specific embodiment of determining the presence of blood vessels in the pulmonary blood vessels in operation 103, when the pulmonary blood vessels corresponding to the pulmonary blood vessel mask are determined as free type in the image segmentation result, the corresponding region of the pulmonary blood vessel mask in the lung CT image is determined as the pneumonia lesion region.
In operation 205, it is determined whether there is a continuum in the pneumonia lesion area corresponding to the pneumonia lesion area mask.
Referring to operation 103 for determining a specific implementation manner of the lesion existing form of the pneumonia lesion region, and referring to operation 104, when it is determined that a continuum of the lesion existing form of the pneumonia lesion region exists, performing an external dilation on the corresponding pneumonia lesion region according to a lung CT image to obtain an external dilation region; judging whether blood vessels exist in the external expansion area or not; when a blood vessel exists in the flaring region, judging whether the blood vessel in the flaring region is connected with the pneumonia focus region; and (4) whether the blood vessel in the flaring region is connected with the pneumonia focus region or not is determined again by utilizing an optimization model.
According to the scheme, after the lung CT image is subjected to image segmentation, the segmentation result is further optimized, and discrete regions with small volume in the lung blood vessels corresponding to the lung blood vessel mask are determined as pneumonia focus regions. And when the focus existing form of the pneumonia focus area corresponding to the pneumonia focus area mask exists a continuum, thinning the area boundary in the division result of the pneumonia focus area. The image segmentation precision is effectively improved, so that the medical image detection result is more accurate.
Similarly, based on the above medical image detection method, an embodiment of the present invention further provides a computer-readable storage medium, in which a program is stored, and when the program is executed by a processor, the processor is caused to perform at least the following operation steps: operation 101, acquiring a lung Computed Tomography (CT) image; operation 102, performing image segmentation according to the lung CT image to obtain an image segmentation result including a lung blood vessel and a pneumonia lesion region; operation 103, determining the blood vessel existing form of the pulmonary blood vessel and the focus existing form of the pneumonia focus region according to the segmentation result; and operation 104, optimizing the image segmentation result according to the lung CT image, the blood vessel existing form and the focus existing form.
Further, based on the medical image detection method, an embodiment of the present invention further provides a medical image detection apparatus, and fig. 3 shows a schematic structural diagram of the medical image detection apparatus according to the embodiment of the present invention. Referring to fig. 3, the apparatus 30 includes: an obtaining module 301, configured to obtain a lung computed tomography CT image; a segmentation module 302, configured to perform image segmentation according to the lung CT image to obtain an image segmentation result including a lung blood vessel mask and a pneumonia focus area mask; a form judging module 303, configured to judge, according to the segmentation result, a blood vessel form of a pulmonary blood vessel corresponding to the pulmonary blood vessel mask and a focus form of a pneumonia focus region corresponding to the pneumonia focus region mask; and an optimizing module 304, configured to optimize computed tomography according to the lung CT image, the blood vessel existing form, and the lesion existing form.
According to an embodiment of the present invention, the form determining module 303 includes: the first form judgment submodule is used for judging whether the blood vessel volume of the pulmonary blood vessel corresponding to each pulmonary blood vessel mask is smaller than a set threshold value or not according to the pulmonary blood vessel masks, and judging that the blood vessel form of the pulmonary blood vessel corresponding to the pulmonary blood vessel masks is free when the blood vessel volume is smaller than the set threshold value; and the second form judgment submodule is used for judging whether a continuum exists in the pneumonia focus area corresponding to the pneumonia focus area mask according to the pneumonia focus area mask.
According to an embodiment of the present invention, the optimization module 304 includes: the external expansion module is used for externally expanding the pneumonia focus area according to the lung CT image to obtain an external expansion area when the existence of a continuum in the pneumonia focus area is judged; the blood vessel judging submodule is used for judging whether blood vessels exist in the externally expanded region or not; the connection judgment submodule is used for judging whether the blood vessels in the external expansion region are connected with the pneumonia focus region or not when the blood vessels exist in the external expansion region; and the first optimization submodule is used for determining the pneumonia focus area in the flaring area again by utilizing an optimization model if the blood vessel in the flaring area is connected with the pneumonia focus area.
According to an embodiment of the present invention, the optimization module 304 includes: and the second optimization submodule is used for determining a region corresponding to the pulmonary blood vessel in the pulmonary CT image to be determined as a pneumonia focus region when the blood vessel form of the pulmonary blood vessel is judged to be free.
Here, it should be noted that: the above description of the embodiment of the medical image detection apparatus is similar to the description of the embodiment of the method shown in fig. 1 to 2, and has similar beneficial effects to the embodiment of the method shown in fig. 1 to 2, and therefore, the description is omitted. For technical details that are not disclosed in the embodiment of the medical image detection apparatus of the present invention, please refer to the description of the method embodiment shown in fig. 1 to 2 of the present invention for understanding, and therefore, for brevity, will not be described again.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described device embodiments are merely illustrative, for example, the division of a unit is only one logical function division, and there may be other division ways in actual implementation, such as: multiple units or components may be combined, or may be integrated into another system, or some features may be omitted, or not implemented. In addition, the coupling, direct coupling or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or units may be electrical, mechanical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units; can be located in one place or distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, all the functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately regarded as one unit, or two or more units may be integrated into one unit; the integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.
Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: various media that can store program codes, such as a removable Memory device, a Read Only Memory (ROM), a magnetic disk, or an optical disk.
Alternatively, the integrated unit of the present invention may be stored in a computer-readable storage medium if it is implemented in the form of a software functional module and sold or used as a separate product. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or a part contributing to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the methods of the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, a ROM, a magnetic or optical disk, or other various media that can store program code.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.