PACS image downloading method, equipment and computer program product

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a PACS image downloading method and device and a computer program product, and the PACS image downloading method comprises the following steps: recognizing a focus area image from a target PACS image in a PACS server according to a PACS image reading request of a PACS client; determining a target part category image where the focus area image is located; and downloading a first thickness layer slice sequence image corresponding to the target part category image to the PACS client. The method can effectively solve the problems that in the image reading process of the PACS client side, a large amount of DICOM images need to be downloaded, and waiting time is long.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device and computer program product for downloading images in a PACS image archiving and communication system. Background Art

[0002] DICOM (Digital Imaging and Communications in Medicine) image data in the PACS (picture archiving and communication system) system accounts for more than 80% of the total hospital information data. With the continuous improvement of the resolution of examination equipment such as CT (Computed Tomography) and MR (Magnetic Resonance), such as the development of CT examination equipment from the early 64 rows to 256 rows, the rapid growth of PACS images has brought severe challenges to hospital informationization. The number of images generated by a patient's examination has reached more than 2,000, and the image adjustment time is generally as long as 1 to 2 minutes. Image adjustment jams are the norm, which seriously affects the work efficiency of doctors.

[0003] At present, since the amount of PACS images for each patient is relatively large and cannot be stored in memory, the PACS reading program will first write the downloaded PACS images to the local hard disk and then read them from the local hard disk. The read and write performance of the local disk often becomes a bottleneck. In addition, from a network perspective, conventional data storage is still mainly centralized, that is, all PACS images are stored in the center of the network for the imaging equipment in the hospital to store data and all workstations to read data. During peak working hours, a large number of PACS client software access the PACS image server at the same time, resulting in a large number of concurrent server network accesses and tight network bandwidth resources.

[0004] Therefore, the existing technology does not provide an effective solution to the above-mentioned problems existing in the existing PACS image reading process. Summary of the invention

[0005] The embodiments of the present invention provide a PACS image downloading method, device and computer program product to at least partially solve the above problems.

[0006] In a first aspect, an embodiment of the present invention provides a method for downloading PACS images, the method comprising: According to the PACS image reading request of the PACS client, the image of the lesion area is identified from the target PACS image in the PACS server; Determine the target part category image where the lesion area image is located; The first thickness layer slice sequence images corresponding to the target part category image are downloaded to the PACS client.

[0007] Optionally, identifying the lesion area image from the target PACS image in the PACS server includes: The lesion area image is identified from the second thickness layer slice sequence image of the target PACS image in the PACS server; wherein the first thickness is smaller than the second thickness.

[0008] Optionally, before identifying the lesion area image from the second thickness layer slice sequence image of the target PACS image in the PACS server, the method includes: Annotate the lesion area on the second thickness slice sequence images of the pre-acquired PACS image samples; Train according to the annotations to obtain the target detection model; The lesion area image is identified from the second thick slice sequence image of the target PACS image in the PACS server according to the target detection model.

[0009] Optionally, downloading the first thickness layer slice sequence image corresponding to the target part category image to the PACS client includes: According to the image boundary of the target part category image, determining the first thickness layer slice sequence image covered by the image boundary on the localizer image of the target PACS image; The first thickness layer slice sequence images covered by the image boundary are downloaded to the PACS client.

[0010] Optionally, the determining of the target part category image where the lesion area image is located includes: Determining a positioning line of a second thick slice sequence image including the lesion area image on the positioning slice image of the target PACS image; The target part category image where the lesion area image is located is determined according to the positioning line and the image boundaries of each part category image.

[0011] Optionally, before determining the first thickness layer slice sequence image covered by the image boundary on the localizer image of the target PACS image based on the image boundary of the target part category image, the method includes: Performing part segmentation network training on the localizer image of the PACS image sample to obtain a part recognition and segmentation model; In the localizer image of the target PACS image, the coordinate position of each part category image is identified according to the part identification and segmentation model; The image boundary of each part category image is determined according to the coordinate position of each part category image.

[0012] Optionally, the determining, based on the image boundary of the target part category image, on the localizer image of the target PACS image, a first thickness layer slice sequence image covered by the image boundary comprises: On the localizer image of the target PACS image, along the vertical direction of the localizer line, determine the projection intersection lines corresponding to the boundary points on both sides of the image boundary respectively; The first thickness layer slice sequence images within the range of the two projection intersection lines are used as the first thickness layer slice sequence images covered by the image boundary.

[0013] Optionally, the determining of projection intersection lines corresponding to boundary points on both sides of the image boundary respectively includes: Determine, from the target PACS image, a set of projection intersection lines of each second thickness layer slice sequence image on the localizer image of the target PACS image; In the set of projected intersection lines, the projected intersection lines where the boundary points on both sides are located are determined.

[0014] In a second aspect, an embodiment of the present invention provides an electronic device, the electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor; When the computer program is executed by the processor, the steps of the PACS image downloading method as described in any one of the above items are implemented.

[0015] In a third aspect, an embodiment of the present invention provides a computer program product, wherein a PACS image downloading program is stored on the computer program product, and when the PACS image downloading program is executed by a processor, the steps of the PACS image downloading method as described in any one of the above items are implemented.

[0016] The various embodiments of the present invention identify the lesion area image from the target PACS image in the PACS server according to the PACS image viewing request of the PACS client, thereby determining the target part category image where the lesion area image is located, and downloading the first thickness layer slice sequence image corresponding to the target part category image to the PACS client, thereby effectively solving the problem of needing to download large amounts of DICOM images during the image viewing process of the PACS client, resulting in long waiting times, and the like. Without changing the PACS image storage architecture, only adjusting the download strategy of the PACS client can achieve fast film viewing on the PACS client and effectively reduce the network traffic of the PACS system. The method proposed in this article can effectively shorten the overall time consumption of file transfer, significantly reduce the network occupancy rate of the PACS server, and significantly improve the film viewing efficiency of the PACS film viewing client. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of a method for downloading images in a PACS image archiving and communication system according to an embodiment of the present invention; Figure 2 is a schematic diagram of a positioning line according to an embodiment of the present invention; Figure 3 is a schematic diagram of an image boundary of a target part category image according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0019] Embodiment 1 The embodiment of the present invention provides a method for downloading images in a PACS image archiving and communication system, such as Figure 1 As shown, the PACS image downloading method comprises: S101, according to a PACS image reading request from a PACS client, identifying a lesion area image from a target PACS image in a PACS server; S102, determining a target part category image where the lesion area image is located; S103, downloading the first thickness layer slice sequence image corresponding to the target part category image to the PACS client.

[0020] The target part category images may be internal parts such as the esophagus, lungs, aorta, liver, pancreas, kidneys, intestines, etc.

[0021] The categories of PACS images include CT images, MR images, mainly DICOM format images, including localizer images, thick layer sequence slice images, and thin layer sequence slice images. In this embodiment, the first thickness layer sequence slice image can correspond to the thin layer sequence slice image, and the second thickness layer sequence slice image corresponds to the thick layer sequence slice image, wherein the first thickness is less than the second thickness.

[0022] The PACS image browsing request may carry a preset downloading strategy, for example, the downloading strategy is to download only images related to the lesion area image, and download images of interest.

[0023] The embodiment of the present invention identifies the lesion area image from the target PACS image in the PACS server according to the PACS image viewing request of the PACS client, thereby determining the target part category image where the lesion area image is located, and downloading the first thickness layer slice sequence image corresponding to the target part category image to the PACS client, thereby effectively solving the problem of needing to download large amounts of DICOM images during the image viewing process of the PACS client, resulting in long waiting times, and the like. Without changing the PACS image storage architecture, only adjusting the download strategy of the PACS client can achieve fast film viewing on the PACS client and effectively reduce the network traffic of the PACS system. The method proposed in this article can effectively shorten the overall time consumption of file transfer, significantly reduce the network occupancy rate of the PACS server, and significantly improve the film viewing efficiency of the PACS film viewing client.

[0024] In some embodiments, identifying the lesion area image from the target PACS image in the PACS server includes: identifying the lesion area image from a second thickness layer slice sequence image of the target PACS image in the PACS server; wherein the first thickness is less than the second thickness.

[0025] The embodiment of the present invention can effectively improve the recognition speed by identifying the lesion area image based on the second thickness layer slice sequence image.

[0026] Based on the above embodiments, other improvements of the present invention are described in detail below through a specific embodiment.

[0027] This specific embodiment takes CT images as an example. The DICOM image of a complete CT image includes a localizer image, a thick-layer sequence slice image, and a thin-layer sequence slice image. The layer sequence slice images (thick-layer sequence slice images, thin-layer sequence slice images) can calculate the projection intersection line on the localizer image through the image position and normal vector information, that is, the currently displayed slice image can draw a positioning line on the positioning map that is orthogonal or oblique to the slice image. According to the above-mentioned calculation principle of the projection intersection line, the projection intersection lines of all slice images on the localizer image can be calculated.

[0028] In this specific embodiment, the training part recognition and segmentation model segments different part category images in the localization film image to obtain the contour coordinate sets of the different part category images; the training target detection model identifies and obtains the slice image containing the suspected lesion area in the thick layer sequence slice image; the projection intersection line of the slice image on the localization film image is calculated, and it is determined whether there is overlap with each contour coordinate set, and the part category image where the suspected lesion is located is obtained; according to the part category image contour point coordinate set, the upper and lower boundary points of the current part category image are obtained, that is, the boundary points on both sides of the current part category image in the vertical direction of the positioning line are obtained; the two projection intersection lines where the upper and lower boundary points of the current part are located are found from the projection intersection lines of all thin layer sequence slice images on the localization film image, and the two thin layer slice images corresponding to the two projection intersection lines are obtained as the start and end range of the downloaded area image; the PACS film reading program only downloads and displays the thin layer sequence slice images within the start and end range.

[0029] In detail, the image archiving and communication system PACS image downloading method provided in this specific embodiment includes: Step 1: Determine the projection intersection lines of all slice images in the thin-layer sequence on the localizer image from the target PACS image. PACS images include CT images, MR images, etc. The target PACS image is the PACS image of the inspection target carried in the PACS image reading request.

[0030] Taking CT images as an example, a complete CT image is composed of multiple sequences, each of which contains multiple DICOM image files, including localizer images, thick-layer sequence slice images, and thin-layer sequence slice images. The thickness of the layers in the thick-layer sequence slice images and the thin-layer sequence slice images can be specifically determined according to actual conditions, and is not specifically limited in this specific embodiment.

[0031] In order to help clinicians intuitively obtain the spatial scanning position of each image, the PACS film reading program provides an image positioning line identification function, that is, when displaying a slice image, the positioning line is displayed on the positioning map that is orthogonal or oblique to the slice, such as Figure 2 shown.

[0032] The projection intersection line of the layer sequence slice image on the localization slice image can be calculated through the image position and normal vector information, that is, the localization line is drawn on the localization slice image. The projection intersection line is obtained by solving the intersection line segment of two planes in three-dimensional space, and the TAG value information in the DICOM image of the PACS image is read to obtain the plane of the layer sequence slice image and the plane of the localization slice image. The intersection points of the four edges of the layer sequence slice image and the localization slice image are calculated to obtain the orthogonal or oblique projection intersection lines of all the layer sequence slice images on the localization slice image. Assuming there are 2000 thin layer sequence slice images, 2000 projection intersection lines can be obtained.

[0033] Step 2: Use the localizer images to train the segmentation model.

[0034] Collect positioning film images of whole-body scans and mark internal parts such as the esophagus, lungs, aorta, liver, pancreas, kidneys, intestines, etc. Train the organ part image category segmentation network model to segment different part category images and obtain the corresponding organ part contour coordinate point set.

[0035] Specifically, a part segmentation network is trained on the localization film image of the PACS image sample to obtain a part recognition segmentation model; in the localization film image of the target PACS image, the coordinate position of each part category image is identified according to the part recognition segmentation model; and the image boundary of each part category image is determined according to the coordinate position of each part category image.

[0036] Step 3: Use thick-slice sequence images of PACS image samples to train the target detection model.

[0037] Collect thick-layer serial slice images of whole-body scans and mark the lesion rectangles. Train the target detection model to identify the slice sequence images containing suspected lesions in the thick-layer serial slice images.

[0038] Specifically, the second thick layer slice sequence images of the pre-acquired PACS image samples are annotated with the lesion area; training is performed based on the annotation to obtain a target detection model; and the lesion area image is identified from the second thick layer slice sequence images of the target PACS image in the PACS server according to the target detection model.

[0039] Step 4: Confirm the location category image of the suspected lesion area and obtain the coordinates of the upper and lower boundary points of the organ image contour.

[0040] Specifically, on the positioning film image of the target PACS image, a positioning line of the second thick slice sequence image including the lesion area image is determined; based on the positioning line and the image boundaries of each part category image, the target part category image where the lesion area image is located is determined.

[0041] For example, the projection intersection line of the layer slice sequence image containing the suspected lesion area image identified in step 3 on the positioning film image is calculated using the method of step 1, and whether there is overlap with the coordinate set of each organ part contour is determined to obtain the image category of the organ part where the suspected lesion is located; based on the part contour point coordinate set, the upper and lower boundary points of the current suspected lesion part are obtained. If there are multiple suspected lesion images, the overlapping positions are calculated with the part contour coordinate set respectively. For example, if one image has inflammation in the lungs, and the projection intersection line is within the lung contour area, it is confirmed that the suspected lesion is in the lungs, and the coordinates of the upper and lower boundary points of the lung contour are extracted. If another image has a lesion in the liver, and the projection intersection line is within the liver contour area, it is confirmed that another suspected lesion is in the liver, and the coordinates of the upper and lower boundary points of the liver contour are extracted, such as Figure 3 shown.

[0042] Step 5: According to the image boundary of the target part category image, determine the first thickness layer slice sequence image covered by the image boundary on the localizer image of the target PACS image.

[0043] Optionally, on the positioning film image of the target PACS image, the projection intersection lines corresponding to the boundary points on both sides of the image boundary are determined along the vertical direction of the positioning line; and the first thickness layer slice sequence image within the range of the two projection intersection lines is used as the first thickness layer slice sequence image covered by the image boundary.

[0044] Specifically, determining the projection intersection lines corresponding to the boundary points on both sides of the image boundary respectively includes: A set of projection intersection lines of each second thickness layer slice sequence image on the localizer image of the target PACS image is determined from the target PACS image; and in the set of projection intersection lines, the projection intersection lines where the two side boundary points are located are determined.

[0045] For example, calculate the start and end range of the downloaded area image in the thin-layer sequence image. Find the upper and lower boundary points of the organ part image category obtained in step 4 from the projection intersection lines of all thin-layer sequence slice images on the localizer image, and obtain two thin-layer slice images corresponding to the two projection intersection lines as the start and end range of the image of the region of interest. If there are multiple parts with lesions, there are multiple regions of interest, and the start and end ranges of multiple images of the region of interest are obtained.

[0046] Step 6: Based on the download strategy, the first thickness layer slice sequence images covered by the image boundary are downloaded to the PACS client.

[0047] The PACS client only downloads and displays the thin-layer sequence slice images within the start and end range of the area corresponding to the download strategy, and the images of the non-interested areas of the thin-layer sequence are no longer downloaded from the server to the client. Assuming that there are 2,000 thin-layer sequence images and the area of ​​interest is in the range of image numbers 100-300, only the images within 100-300 need to be downloaded instead of downloading 2,000 images, which can effectively shorten the overall time consumption of file transmission and significantly reduce the network occupancy rate of the PACS server.

[0048] Embodiment 2 An embodiment of the present invention provides an electronic device, the electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor; When the computer program is executed by the processor, the steps of the PACS image downloading method as described in any one of the first embodiments are implemented.

[0049] Embodiment 3 An embodiment of the present invention provides a computer program product, wherein a PACS image downloading program is stored on the computer program product. When the PACS image downloading program is executed by a processor, the steps of the PACS image downloading method as described in any one of the above items are implemented.

[0050] In the specific implementation process of Embodiments 2 and 3, reference may be made to Embodiment 1, and they have corresponding technical effects.

[0051] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A PACS image downloading method, characterized in that: The PACS image downloading method comprises: According to the PACS image reading request of the PACS client, the image of the lesion area is identified from the target PACS image in the PACS server; Determine the target part category image where the lesion area image is located; The first thickness layer slice sequence images corresponding to the target part category image are downloaded to the PACS client.

2. The PACS image downloading method according to claim 1, characterized in that: The step of identifying the lesion area image from the target PACS image in the PACS server includes: The lesion area image is identified from the second thickness layer slice sequence image of the target PACS image in the PACS server; wherein the first thickness is smaller than the second thickness.

3. The PACS image downloading method according to claim 2, characterized in that: Before identifying the lesion area image from the second thickness layer slice sequence image of the target PACS image in the PACS server, the method includes: Annotate the lesion area on the second thickness slice sequence images of the pre-acquired PACS image samples; Train according to the annotations to obtain the target detection model; The lesion area image is identified from the second thick slice sequence image of the target PACS image in the PACS server according to the target detection model.

4. The PACS image downloading method according to any one of claims 1 to 3, characterized in that: The step of downloading the first thickness layer slice sequence image corresponding to the target part category image to the PACS client comprises: According to the image boundary of the target part category image, determining the first thickness layer slice sequence image covered by the image boundary on the localizer image of the target PACS image; The first thickness layer slice sequence images covered by the image boundary are downloaded to the PACS client.

5. The PACS image downloading method according to claim 4, characterized in that: The step of determining the target part category image where the lesion area image is located comprises: Determining a positioning line of a second thick slice sequence image including the lesion area image on the positioning slice image of the target PACS image; The target part category image where the lesion area image is located is determined according to the positioning line and the image boundaries of each part category image.

6. The PACS image downloading method according to claim 4, characterized in that: The method comprises: before determining the first thickness layer slice sequence image covered by the image boundary on the localizer image of the target PACS image according to the image boundary of the target part category image, comprising: Performing part segmentation network training on the localizer image of the PACS image sample to obtain a part recognition and segmentation model; In the localizer image of the target PACS image, the coordinate position of each part category image is identified according to the part identification and segmentation model; The image boundary of each part category image is determined according to the coordinate position of each part category image.

7. The PACS image downloading method according to claim 4, characterized in that: The step of determining, based on the image boundary of the target part category image, a first thickness layer slice sequence image covered by the image boundary on the localizer image of the target PACS image comprises: On the localizer image of the target PACS image, along the vertical direction of the localizer line, determine the projection intersection lines corresponding to the boundary points on both sides of the image boundary respectively; The first thickness layer slice sequence images within the range of the two projection intersection lines are used as the first thickness layer slice sequence images covered by the image boundary.

8. The PACS image downloading method according to claim 7, characterized in that: The step of determining the projection intersection lines corresponding to the boundary points on both sides of the image boundary respectively comprises: Determine, from the target PACS image, a set of projection intersection lines of each second thickness layer slice sequence image on the localizer image of the target PACS image; In the set of projected intersection lines, the projected intersection lines where the boundary points on both sides are located are determined.

9. An electronic device, characterized in that: The electronic device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor; When the computer program is executed by the processor, the steps of the PACS image downloading method according to any one of claims 1 to 8 are implemented.

10. A computer program product, characterized in that The computer program product stores a PACS image downloading program, and when the PACS image downloading program is executed by a processor, the steps of the PACS image downloading method according to any one of claims 1 to 8 are implemented.

Images