Liver segmentation quality assessment method, device, and medium

Abstract

The application relates to a liver segmentation quality evaluation method, device and medium. The method comprises the following steps: acquiring a liver segmentation result to be evaluated; the liver segmentation result comprises orientation information and morphological information of at least one anatomical structure; inputting the liver segmentation result into a preset quantification model for quantification processing to determine a quantification value corresponding to the liver segmentation result; wherein the quantification model is determined based on the liver segmentation result and a quantification mode of the liver segmentation result, the quantification mode is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result; determining a quality category corresponding to the liver segmentation result according to the quantification value corresponding to the liver segmentation result and a preset quantification value range; wherein each quantification value range corresponds to a quality category. The method can ensure the accuracy of the liver segmentation result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a liver segmentation quality evaluation method, device and medium. BACKGROUND

[0002] With the continuous development of three-dimensional reconstruction technology and segmentation technology, after the key region segmentation and three-dimensional reconstruction of human liver and other parts are performed, the segmentation result obtained can be further analyzed to obtain an accurate analysis result. As can be seen, the quality of the segmentation result directly affects the correctness of the subsequent image analysis result, so it is particularly important to obtain an accurate segmentation result.

[0003] In the related art, the liver image is first segmented to obtain a liver mask and a blood vessel mask, and a hepatic vein mask is further calculated based on the blood vessel mask. Then, the overlap between the hepatic vein mask and the liver mask is calculated to evaluate the segmentation quality of the blood vessels in the liver image based on the size of the overlap.

[0004] However, the above-mentioned method for evaluating the segmentation result has poor reliability of the evaluation result, so it is difficult to ensure the accuracy of the segmentation result. SUMMARY

[0005] Therefore, it is necessary to provide a liver segmentation quality evaluation method, device and medium capable of ensuring the accuracy of the liver segmentation result.

[0006] In a first aspect, the present application provides a liver segmentation quality evaluation method, which comprises:

[0007] obtaining a liver segmentation result to be evaluated; the liver segmentation result comprises orientation information and morphological information of at least one anatomical structure;

[0008] inputting the liver segmentation result into a preset quantization model for quantization processing to determine a quantization value corresponding to the liver segmentation result; wherein the quantization model is determined based on the liver segmentation result and a quantization manner of the liver segmentation result, and the quantization manner is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result;

[0009] determining a quality category corresponding to the liver segmentation result according to the quantization value corresponding to the liver segmentation result and a preset quantization value range; wherein each quantization value range corresponds to a quality category.

[0010] In one embodiment, the quantization model comprises a plurality of quantization manners of the liver segmentation result, and each quantization manner corresponds to a segmentation result of different dimensions of the liver segmentation result.

[0011] In one of the embodiments, the inputting the liver segmentation result into the preset quantification model for quantification processing to determine the quantification value corresponding to the liver segmentation result comprises:

[0012] According to the orientation information and the morphological information of each anatomical structure in the liver segmentation result and the corresponding quantification mode, the plurality of initial quantification values corresponding to the liver segmentation result are calculated.

[0013] The plurality of initial quantification values corresponding to the liver segmentation result are integrated to obtain the quantification value corresponding to the liver segmentation result.

[0014] In one of the embodiments, the calculating the first quantification value corresponding to the liver segmentation result according to the orientation information and the morphological information of each anatomical structure in the liver segmentation result and the corresponding first quantification mode comprises:

[0015] According to the orientation information and the morphological information of each anatomical structure in the liver segmentation result and the corresponding first quantification mode, the first quantification value corresponding to the liver segmentation result is calculated.

[0016] According to the orientation information of each anatomical structure in the liver segmentation result, the relative positional relationship between the anatomical structures is obtained, and according to the relative positional relationship between the anatomical structures and the corresponding second quantification mode, the second quantification value corresponding to the liver segmentation result is calculated.

[0017] According to the orientation information and the morphological information of each anatomical structure in the liver segmentation result, the image information of the whole liver in the liver segmentation result is obtained, and according to the image information of the whole liver and the corresponding third quantification mode, the third quantification value corresponding to the liver segmentation result is calculated.

[0018] At least two of the first quantification value, the second quantification value and the third quantification value are taken as the initial quantification value.

[0019] In one of the embodiments, the calculating the first quantification value corresponding to the liver segmentation result according to the orientation information and the morphological information of each anatomical structure in the liver segmentation result and the corresponding first quantification mode comprises:

[0020] According to the orientation information of each anatomical structure and the corresponding first orientation quantification mode, the first orientation quantification value corresponding to each anatomical structure is calculated; the first orientation quantification mode is used to quantize whether the corresponding anatomical structure exists in the liver segmentation result through the actual orientation information of the anatomical structure.

[0021] According to the morphological information of each anatomical structure and the corresponding first morphological quantification mode, the first morphological quantification value corresponding to each anatomical structure is calculated; the first morphological quantification mode is used to quantize the morphology of the anatomical structure in the liver segmentation result through the actual morphological information of the anatomical structure.

[0022] The first orientation quantization value and the first shape quantization value are integrated to obtain a first quantization value.

[0023] In one embodiment, the relative positional relationship between the anatomical structures includes a first relative positional relationship between a segmentation surface of a liver segment and a blood vessel and a second relative positional relationship between an internal part of the liver segment and a portal vein in the blood vessel.

[0024] The second quantization value corresponding to the liver segmentation result is calculated according to the relative positional relationship between the anatomical structures and a corresponding second quantization mode, including:

[0025] The first relative positional relationship is quantitatively calculated according to the first relative positional relationship and a corresponding second quantization mode to obtain a second quantization value corresponding to the first relative positional relationship.

[0026] The second relative positional relationship is quantitatively calculated according to the second relative positional relationship and a corresponding second quantization mode to obtain a second quantization value corresponding to the second relative positional relationship.

[0027] The second quantization value corresponding to the liver segmentation result is determined according to the second quantization value corresponding to the first relative positional relationship and the second quantization value corresponding to the second relative positional relationship.

[0028] In one embodiment, the image information of the whole liver includes data quality of original data corresponding to the liver segmentation result and contour fitting degree of an anatomical structure in the liver segmentation result to the original data.

[0029] The third quantization value corresponding to the liver segmentation result is calculated according to the image information of the whole liver and a corresponding third quantization mode, including:

[0030] The data quality of the original data is quantitatively processed according to the data quality of the original data and a corresponding third quantization mode to obtain a third quantization value corresponding to the data quality of the original data.

[0031] The contour fitting degree is quantitatively processed according to the contour fitting degree of the anatomical structure in the liver segmentation result to the original data and a corresponding third quantization mode to obtain a third quantization value corresponding to the contour fitting degree.

[0032] The third quantization value corresponding to the liver segmentation result is determined according to the third quantization value corresponding to the data quality of the original data and the third quantization value corresponding to the contour fitting degree.

[0033] In one embodiment, the anatomical structure includes at least one of a liver, a liver segment, and a blood vessel.

[0034] In a second aspect, the present application also provides a liver segmentation quality evaluation device, which comprises:

[0035] an acquisition module, configured to acquire a liver segmentation result to be evaluated; the liver segmentation result comprises orientation information and morphological information of at least one anatomical structure;

[0036] a quantification module, configured to input the liver segmentation result into a preset quantification model for quantification processing, so as to determine a quantification value corresponding to the liver segmentation result; the quantification model is determined based on the liver segmentation result and a quantification manner of the liver segmentation result, and the quantification manner is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result;

[0037] an evaluation module, configured to determine a quality category corresponding to the liver segmentation result according to the quantification value corresponding to the liver segmentation result and a preset quantification value range; each quantification value range corresponds to one quality category.

[0038] In a third aspect, the present application also provides a computer device, which comprises a memory and a processor; the memory stores a computer program; and the processor implements the following steps when executing the computer program:

[0039] acquiring a liver segmentation result to be evaluated; the liver segmentation result comprises orientation information and morphological information of at least one anatomical structure;

[0040] inputting the liver segmentation result into a preset quantification model for quantification processing, so as to determine a quantification value corresponding to the liver segmentation result; the quantification model is determined based on the liver segmentation result and a quantification manner of the liver segmentation result, and the quantification manner is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result;

[0041] determining a quality category corresponding to the liver segmentation result according to the quantification value corresponding to the liver segmentation result and a preset quantification value range; each quantification value range corresponds to one quality category.

[0042] In a fourth aspect, the present application also provides a computer readable storage medium, which stores a computer program; the computer program is executed by a processor to implement the following steps:

[0043] acquiring a liver segmentation result to be evaluated; the liver segmentation result comprises orientation information and morphological information of at least one anatomical structure;

[0044] inputting the liver segmentation result into a preset quantification model to perform quantification processing, and determining a quantification value corresponding to the liver segmentation result; wherein the quantification model is determined based on the liver segmentation result and a quantification manner of the liver segmentation result, and the quantification manner is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result;

[0045] determining a quality category corresponding to the liver segmentation result according to the quantification value corresponding to the liver segmentation result and a preset quantification value range; wherein each quantification value range corresponds to a quality category.

[0046] In a fifth aspect, the present application further provides a computer program product. The computer program product comprises a computer program which, when executed by a processor, implements the following steps:

[0047] obtaining a liver segmentation result to be evaluated; the liver segmentation result comprises orientation information and morphological information of at least one anatomical structure;

[0048] inputting the liver segmentation result into a preset quantification model to perform quantification processing, and determining a quantification value corresponding to the liver segmentation result; wherein the quantification model is determined based on the liver segmentation result and a quantification manner of the liver segmentation result, and the quantification manner is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result;

[0049] determining a quality category corresponding to the liver segmentation result according to the quantification value corresponding to the liver segmentation result and a preset quantification value range; wherein each quantification value range corresponds to a quality category.

[0050] The liver segmentation quality evaluation method, device and medium described above, by obtaining a liver segmentation result comprising orientation information and morphological information of an anatomical structure, inputting the liver segmentation result into a preset quantification model to perform quantification, determining a corresponding quantification value, and determining a quality category corresponding to the liver segmentation result according to the quantification value and a preset quantification value range, wherein the quantification model is determined based on the liver segmentation result and a quantification manner corresponding thereto, and the quantification manner is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result. In this method, the quality of the liver segmentation result can be evaluated by the actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result, which takes into account the actual orientation and actual morphology of the anatomical structure, so the quality evaluation of the liver segmentation result is more accurate, i.e. the quality evaluation result of the liver segmentation result obtained is more reliable, thus the accuracy of the liver segmentation result obtained can be ensured. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 is an internal structure diagram of a computer device in one embodiment;

[0052] Figure 2 Flowchart of a liver segmentation quality evaluation method in an embodiment;

[0053] Figure 3 Flowchart of a liver segmentation quality evaluation method in another embodiment;

[0054] Figure 4 Flowchart of a liver segmentation quality evaluation method in another embodiment;

[0055] Figure 5 Flowchart of a process of quantifying each blood vessel in another embodiment;

[0056] Figure 6 Flowchart of a process of quantifying each blood vessel and segmentation surface and liver segment in another embodiment;

[0057] Figure 7 Flowchart of a process of quantifying overall liver image information in another embodiment;

[0058] Figure 8 Structural block diagram of a liver segmentation quality evaluation device in an embodiment. DETAILED DESCRIPTION

[0059] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not intended to limit the present application.

[0060] The liver segmentation quality evaluation method provided by the embodiments of the present application can be applied to a computer device, which can be a terminal or a server. Taking the terminal as an example, its internal structure diagram can be as shown in Figure 1As shown in the figure. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. Wireless mode can be achieved through WIFI, mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by the processor to implement a liver segmentation quality evaluation method. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad provided on the shell of the computer device. It can also be an external keyboard, touchpad or mouse, etc.

[0061] Those skilled in the art can understand that, Figure 1 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0062] In one embodiment, as Figure 2 shown, a liver segmentation quality evaluation method is provided. The method is applied to Figure 1 the computer device in the figure for illustration. The method can include the following steps:

[0063] S202, obtaining the liver segmentation result to be evaluated; the liver segmentation result includes the orientation information and the morphological information of at least one anatomical structure.

[0064] In this step, the liver region of the test object can be scanned in advance to obtain scanning data, and the three-dimensional image of the liver can be obtained by image reconstruction on the scanning data. Then, the liver segmentation result can be obtained by using artificial or deep learning algorithm model for liver segmentation of the three-dimensional image of the liver. Or, the original data of the liver region of the test object can also be obtained in advance, and the liver segmentation result can be obtained by 3D reconstruction after segmentation of the original data.

[0065] The liver segmentation result includes each anatomical structure obtained by segmentation, and orientation information and morphological information of each anatomical structure. The orientation information is used to represent the position of the anatomical structure in the liver segmentation result and the relative positional relationship with other anatomical structures, and the morphological information is used to represent the posture and shape of the anatomical structure in the liver segmentation result. The anatomical structure includes but is not limited to liver, liver segment, blood vessel, etc., and the blood vessel can include portal vein, hepatic vein, hepatic proper artery, etc.

[0066] In S204, the liver segmentation result is input into a preset quantization model for quantization processing to determine a quantization value corresponding to the liver segmentation result. The quantization model is determined based on the liver segmentation result and a quantization manner of the liver segmentation result. The quantization manner is related to the actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result.

[0067] In this step, the quantization model can be a machine learning model (for example, a neural network model), a model representing the correlation between the liver segmentation result and the quantization manner, or other types of models, which are not limited in the embodiment.

[0068] The quantization manner can be determined by quantitatively analyzing the specific orientation information of each anatomical structure in the liver segmentation result and the corresponding actual orientation information, or by quantitatively analyzing the specific morphological information of each anatomical structure in the liver segmentation result and the corresponding actual morphological information, or by combining the two.

[0069] Specifically, after obtaining the liver segmentation result, the liver segmentation result can be input into the quantization model. In the quantization model, the orientation information and the morphological information of each anatomical structure are quantitatively processed by each quantization manner to obtain a plurality of quantization values. Then, the sum of the plurality of quantization values or one of the plurality of quantization values is obtained to obtain the quantization value corresponding to the liver segmentation result.

[0070] In S206, a quality category corresponding to the liver segmentation result is determined according to the quantization value corresponding to the liver segmentation result and a preset quantization value range. Each quantization value range corresponds to a quality category.

[0071] In this step, a plurality of quantization value ranges can be preset, and each quantization value range is respectively set with a corresponding quality category. The quality category is used to represent the accuracy of the liver segmentation result, for example, a high quality category represents a high accuracy of the liver segmentation result. For example, a high quality category corresponds to a high quantization value, and a low quality category corresponds to a low quantization value.

[0072] Specifically, after obtaining the quantitative value corresponding to the liver segmentation result, the quantitative value can be matched with the plurality of preset quantitative value ranges. If the quantitative value matches a certain quantitative value range successfully, the quality category corresponding to the matched quantitative value range is determined as the quality category corresponding to the liver segmentation result. The quality category can be used to obtain the quality of the liver segmentation result, i.e., to evaluate the quality of the liver segmentation result, and thus a more accurate data basis can be provided for subsequent post-processing using the liver segmentation result. For example, if the quality category indicates that the accuracy of the liver segmentation result is high, the further processing result obtained by using the liver segmentation result is also more accurate and has higher accuracy. For another example, if the quality category indicates that the accuracy of the liver segmentation result is low, the liver segmentation result cannot be directly used, and the manual segmentation method or the parameters of the deep learning algorithm model can be adjusted to obtain a liver segmentation result with higher accuracy.

[0073] In the above liver segmentation quality evaluation method, the liver segmentation result including the orientation information and the shape information of the anatomical structure is obtained, the liver segmentation result is input into a preset quantitative model for quantization to determine a corresponding quantitative value, and a quality category corresponding to the liver segmentation result is determined according to the quantitative value and a preset quantitative value range. The quantitative model is determined based on the liver segmentation result and a corresponding quantitative manner, and the quantitative manner is related to the actual orientation information and the actual shape information of the anatomical structure in the liver segmentation result. In this method, the quality of the liver segmentation result can be evaluated by the actual orientation information and the actual shape information of the anatomical structure in the liver segmentation result. The actual orientation and the actual shape of the anatomical structure are considered, so the quality evaluation of the liver segmentation result is more accurate, i.e., the quality evaluation result of the liver segmentation result is more reliable, and thus the accuracy of the obtained liver segmentation result can be ensured.

[0074] In another embodiment, the above quantitative model includes a plurality of quantitative manners of the liver segmentation result, and each quantitative manner corresponds to a segmentation result of different dimensions of the liver segmentation result.

[0075] The different dimensions herein can be dimensions related to the orientation information and the shape information of each anatomical structure in the liver segmentation result, for example, the orientation information and the shape information can be directly taken as one dimension of data, or the orientation information and the shape information can be combined and directly taken as one dimension of data, or the processed results of the orientation information and the shape information can be taken as one dimension of data, or other manners can be used, which are not limited here.

[0076] Specifically, a corresponding quantization manner can be set for the data of each dimension in advance, the quantization manner being used to represent the correlation / association degree of the data of each dimension with the data of the dimension in the actual situation, and then the data of each dimension is quantized respectively through the quantization manner of each dimension, and a quantization value of the liver segmentation result is obtained through the obtained quantization result.

[0077] In addition, the quantization model in the embodiment can also be trained in advance through the gold standard quantization results of the data of each dimension in the liver segmentation result under each quantization manner.

[0078] In the embodiment, the quantization model includes multiple quantization manners of the liver segmentation result, each quantization manner corresponding to a segmentation result of a different dimension of the liver segmentation result. Here, the liver segmentation result is evaluated through multiple dimensions, and the dimensions are set to be more and more fine, so that the evaluation result of the liver segmentation result is more accurate and reliable.

[0079] In another embodiment, another liver segmentation quality evaluation method is provided, as shown in Figure 3 As shown in the above embodiment, the S204 can include the following steps:

[0080] S302, according to the orientation information and the morphological information of each anatomical structure in the liver segmentation result and the corresponding quantization manner, calculating multiple initial quantization values corresponding to the liver segmentation result.

[0081] In this step, as mentioned above, the quantization model includes multiple quantization manners, which respectively correspond to the data of the orientation information and the morphological information of each anatomical structure in multiple dimensions. The orientation information and / or the morphological information on the corresponding dimension can be processed through each quantization manner to obtain the corresponding quantization value, which is recorded as an initial quantization value, that is, multiple initial quantization values corresponding to the liver segmentation result are obtained.

[0082] S304, synthesizing the multiple initial quantization values corresponding to the liver segmentation result to obtain a quantization value corresponding to the liver segmentation result.

[0083] In this step, after obtaining the multiple initial quantization values corresponding to the liver segmentation result, the initial quantization values can be directly summed to obtain the quantization value corresponding to the liver segmentation result, or the initial quantization values can be weighted and summed to obtain the quantization value corresponding to the liver segmentation result, or one or more of the initial quantization values can be taken as the quantization value corresponding to the liver segmentation result, which is not specifically limited here.

[0084] In this embodiment, a plurality of initial quantization values are calculated through the orientation information and the shape information of each anatomical structure in the liver segmentation result and the corresponding quantization mode, and the quantization value corresponding to the liver segmentation result is obtained by summing the plurality of initial quantization values. The quantization value of the liver segmentation result is obtained through a plurality of initial quantization values obtained from a plurality of dimensions, and the obtained quantization value of the liver segmentation result is more comprehensive and accurate, so that the accuracy of subsequent matching can be improved, and the accuracy of the liver segmentation result is ensured.

[0085] In another embodiment, another liver segmentation quality evaluation method is provided, as shown in Figure 4 On the basis of the above-mentioned embodiment, the S302 can include the following steps:

[0086] S402, according to the orientation information and the shape information of each anatomical structure in the liver segmentation result and the corresponding first quantization mode, calculating the first quantization value corresponding to the liver segmentation result.

[0087] Optionally, the anatomical structure includes at least one of the liver, the liver segment, and the blood vessel. The first quantization mode can be to pre-set a standard anatomical structure through prior knowledge, compare the orientation information of the anatomical structure in the liver segmentation result with the orientation information of the standard anatomical structure, compare the shape information of the anatomical structure in the liver segmentation result with the shape information of the standard anatomical structure, and obtain the first quantization value corresponding to the anatomical structure through the comparison result.

[0088] Optionally, the process of calculating the first quantization value can include: calculating the first orientation quantization value corresponding to each anatomical structure according to the orientation information of each anatomical structure and the corresponding first orientation quantization mode; the first orientation quantization mode is used to quantify whether the corresponding anatomical structure exists in the liver segmentation result through the actual orientation information of the anatomical structure; calculating the first shape quantization value corresponding to each anatomical structure according to the shape information of each anatomical structure and the corresponding first shape quantization mode; the first shape quantization mode is used to quantify the shape of the anatomical structure in the liver segmentation result through the actual shape information of the anatomical structure; and the first orientation quantization value and the first shape quantization value are comprehensively calculated to obtain the first quantization value.

[0089] For example, taking the liver as an example, a standard liver can be pre-set by prior knowledge, which includes the position information of the standard liver and the morphological information of the standard liver. The morphology of the standard and normal liver segment should be wedge-shaped, with smooth and flat edges, and sharp corners. Then, whether the corresponding liver structure exists in the liver segmentation result and is in the correct position can be determined by comparing the position information of the liver in the liver segmentation result with the position information of the standard liver. If the comparison is successful, it is determined that the liver in the liver segmentation result exists and is in the correct position. Then, a quantitative value of 1 can be set for this item. Alternatively, if the comparison fails, it is determined that the liver in the liver segmentation result does not exist or exists but is not in the correct position. Then, a quantitative value of 0 can be set for this item, i.e., the position quantization value of the liver is obtained.

[0090] When the quantitative value of this item is 1, the morphological information of the liver in the liver segmentation result can be compared with the morphological information of the standard liver, the overlap degree between them is calculated, and the calculated overlap degree is matched with a plurality of overlap degree ranges. Each overlap degree range corresponds to a morphological quantization value, and different overlap degree ranges correspond to different morphological quantization values. Thus, the matched overlap degree range and its corresponding morphological quantization value can be obtained. For example, the following gives several examples of morphological quantization values:

[0091] The morphological quantization value is 0: The liver structure is completely absent in the liver segmentation result, i.e., it does not exist. Therefore, when the liver does not exist in the previous liver segmentation result, the morphological quantization value of this item is also 0.

[0092] The morphological quantization value is 1: The liver structure in the liver segmentation result is very poor in morphology, for example, there is severe deformation and error in the morphology, including incorrect segmentation, mutual invasion between different liver segments, and non-adhesion between liver segments.

[0093] The morphological quantization value is 2: The liver structure in the liver segmentation result is poor in morphology, for example, there are some defects in the morphology, such as non-smooth and continuous segmentation surface between liver segments, and discontinuity.

[0094] The morphological quantization value is 3: The liver structure in the liver segmentation result is acceptable in morphology, for example, the morphology, surface, and edge angle are generally correct.

[0095] The morphological quantization value is 4: The liver structure in the liver segmentation result is very good in morphology, for example, the morphology is basically wedge-shaped, the surfaces are basically smooth and continuous, and the edge corners are basically sharp, with a small amount of redundant segmentation and non-smooth surface.

[0096] The morphological quantization value is 5: The liver structure in the liver segmentation result is very perfect in morphology, for example, the morphology conforms to the wedge shape of the normal liver segment, the surfaces are very smooth and continuous, and the edge corners are sharp without redundant segmentation.

[0097] After obtaining the orientation quantification value and the shape quantification value of the liver, the first quantification value corresponding to the liver can be obtained by summing the two.

[0098] For the case where the anatomical structure is a liver segment, the liver segment can be divided into 8 liver segments S1-S8. The standard liver set as described above can be segmented by prior knowledge to obtain 8 liver segments of the standard liver segment, and different liver segments are closely fitted. The quantification method of each liver segment can be the same as the quantification method of the liver as described above, and the shape information or the orientation information can be adjusted for the liver, which will not be described here. It should be noted that each liver segment can obtain a first quantification value, and the first quantification value of each liver segment can be summed to obtain the first quantification value corresponding to all liver segments.

[0099] For the case where the anatomical structure is a blood vessel, the blood vessel can include the portal vein, the hepatic vein, the hepatic artery, etc., and of course can also include multiple branches of each blood vessel. The process of quantifying each blood vessel can be referred to in the table shown in Figure 5 Prior knowledge can be used to pre-set the corresponding standard blood vessel, which includes the orientation information of the standard blood vessel and the shape information of the standard blood vessel, and of course can also include the intersection information of each blood vessel, etc. Then, it can be determined whether the corresponding blood vessel structure exists in the liver segmentation result and is in the correct position by comparing the orientation information of the blood vessel in the liver segmentation result with the orientation information of the standard blood vessel. If the comparison is successful, it is determined that the blood vessel in the liver segmentation result exists and is in the correct position, and a quantification value of 1 can be set for this item. Alternatively, if the comparison fails, it is determined that the blood vessel does not exist or exists but is not in the correct position, and a quantification value of 0 can be set for this item, i.e. the orientation quantification value of the blood vessel is obtained.

[0100] When the quantification value of this item is 1, the shape information of the blood vessel in the liver segmentation result can be compared with the shape information of the standard blood vessel, the overlap degree between the two can be calculated, and the calculated overlap degree can be matched with a plurality of overlap degree ranges, each of which corresponds to a shape quantification value. Different overlap degree ranges correspond to different shape quantification values, so that the matched overlap degree range and its corresponding shape quantification value can be obtained. For example, the following gives several examples of shape quantification values:

[0101] The shape quantification value is 0: the shape of the blood vessel structure in the liver segmentation result is completely missing or incorrect, i.e. the blood vessel is not reconstructed or the blood vessel is incorrectly reconstructed into other parts; therefore, when the blood vessel does not exist in the previous liver segmentation result, the shape quantification value of this item is also 0.

[0102] Morphology quantization value is 1: the blood vessel structure in the liver segmentation result is very poor in morphology, for example, the blood vessel has only a small amount of main stem or branch, the blood vessel is discontinuous, there are many broken places, the shape is not a tube structure, the bifurcation is obviously adhered, and there are many grafting errors between different blood vessels.

[0103] Morphology quantization value is 2: the blood vessel structure in the liver segmentation result is poor in morphology, for example, the blood vessel has a main stem and a small amount of branch, there are a small amount of discontinuous broken places, the shape is like a noodle, the bifurcation is adhered, and there are grafting errors.

[0104] Morphology quantization value is 3: the blood vessel structure in the liver segmentation result is acceptable in morphology, for example, the blood vessel has a main stem and a small amount of important branch, there are a small amount of broken places, the shape is a tube type, the bifurcation is a small amount of adhered, and there are no grafting errors.

[0105] Morphology quantization value is 4: the blood vessel structure in the liver segmentation result is very good in morphology, for example, the blood vessel has a main stem and many important branches, there are no broken places, the shape is a tube type, the bifurcation is a small amount of adhered, and there are no grafting errors.

[0106] Morphology quantization value is 5: the blood vessel structure in the liver segmentation result is very perfect in morphology, for example, the blood vessel has a main stem and a large amount of important branches, there are no broken places, the shape is a tube type from thick to thin, the bifurcation is not adhered, and there are no grafting errors.

[0107] After obtaining the orientation quantization value and the morphology quantization value of the blood vessel, the sum of the two values is obtained, that is, the first quantization value corresponding to the blood vessel is obtained. For the first quantization value of the whole blood vessel, the first quantization values of each blood vessel can be added to obtain the first quantization value corresponding to all blood vessels.

[0108] In summary, by quantizing different anatomical structures in the liver segmentation result as described above, the first quantization value corresponding to each anatomical structure can be obtained, that is, a plurality of first quantization values can be obtained.

[0109] S404, according to the orientation information of each anatomical structure in the liver segmentation result, the relative position relationship between each anatomical structure is obtained, and according to the relative position relationship between each anatomical structure and the corresponding second quantization mode, the second quantization value corresponding to the liver segmentation result is calculated.

[0110] In this step, the general relative position relationship mainly refers to the relative position relationship between the liver segments and the blood vessels, and mainly evaluates whether the segmentation of the liver segments is reasonable. Optionally, the relative position relationship between the anatomical structures can include a first relative position relationship between the segmentation surface of the liver segments and the blood vessels and a second relative position relationship between the interior of the liver segments and the portal vein in the blood vessels. Among them, the segmentation of the liver segments in the liver is anatomically marked, and the anatomical markers include the branches of the blood vessels, the liver fissures, etc. By evaluating the relative position relationship between the segmentation surface and the blood vessels, it can be evaluated whether the segmentation of the liver is accurate. In addition, the portal vein is a blood vessel collection system of liver function, which carries rich nutrients to the liver, and the liver segment is a functional unit structure of the liver, which is closely related to the portal vein. Different branches of the portal vein are responsible for delivering blood to different liver segments, so by evaluating the relative position relationship between the branches of the portal vein and the liver segments, it can be evaluated whether the segmentation of the liver is accurate.

[0111] Further, when quantifying the first relative position relationship and the second relative position relationship, optionally, the first relative position relationship and the second relative position relationship can each have a corresponding second quantization manner, and the corresponding second quantization value is calculated by the second quantization manner. For example, the first relative position relationship can be quantitatively calculated according to the first relative position relationship and the corresponding second quantization manner to obtain the second quantization value corresponding to the first relative position relationship; the second relative position relationship can be quantitatively calculated according to the second relative position relationship and the corresponding second quantization manner to obtain the second quantization value corresponding to the second relative position relationship; and the second quantization value corresponding to the liver segmentation result can be determined according to the second quantization value corresponding to the first relative position relationship and the second quantization value corresponding to the second relative position relationship. The process of quantifying each blood vessel and segmentation surface and liver segment can be referred to the table shown in Figure 6 Generally, the normal liver segment segmentation surface should pass through the corresponding anatomical structure, such as the branch of the liver blood vessel, the liver segment, etc., and the normal liver segment should contain the corresponding branch of the portal vein.

[0112] For the first relative position relationship, the above first relative position relationship is the relative position relationship between the surface and the body, which can be pre-set by prior knowledge. A plurality of different coincidence degree ranges are set, each coincidence degree range corresponds to a corresponding quantization value, that is, a plurality of second quantization manners are set for the first relative position relationship. Specifically, in the quantization process, the coincidence degree between the segmentation surface of the liver segment and the corresponding blood vessel (for example, the coincidence degree between the segmentation surface of the S1 liver segment and the S2-S8 liver segment and the natural arc line between the right margin of the vena cava and the venous cord in the Figure 6 ), and the calculated coincidence degree is matched with the plurality of coincidence degree ranges to obtain the matched coincidence degree range and the corresponding quantization value, that is, the quantization value corresponding to the first relative position relationship between the segmentation surface and the blood vessel is obtained. For theFigure 6 The other segmented planes and blood vessels shown can all be assigned corresponding second quantization methods and quantized values ​​using the same approach. Finally, the quantization values ​​obtained for all the segmented planes and blood vessels' first relative positional relationships are summed to obtain the second quantization value corresponding to the first relative positional relationship.

[0113] Regarding the second relative positional relationship, which is the relative positional relationship between two bodies, multiple different overlap ranges can be pre-defined using prior knowledge. Each overlap range corresponds to a specific quantization value, thus setting multiple second quantization methods for the second relative positional relationship. Specifically, during quantization, the overlap between the interior of the liver segment and the portal vein can be calculated (e.g., Figure 6 The overlap between the S2 hepatic segment and the left superior lateral branch of the portal vein was calculated, and the calculated overlap was matched with multiple overlap ranges to obtain the matched overlap ranges and their corresponding quantitative values. This yielded the quantitative value corresponding to the second relative positional relationship between the interior of the hepatic segment and the portal vein. Figure 6 The second relative positional relationships between the interior of other liver segments and the portal vein shown can all be configured using the same second quantization method to obtain corresponding quantization values. Finally, the quantization values ​​obtained for all the second relative positional relationships between the interior of liver segments and the portal vein are summed to obtain the second quantization value corresponding to the second relative positional relationship.

[0114] Finally, the second quantization value corresponding to the first relative position relationship and the second quantization value corresponding to the second relative position relationship can be summed to obtain the second quantization value corresponding to the liver segmentation result.

[0115] S406. Based on the morphological and orientational information of each anatomical structure in the liver segmentation result, obtain the overall image information of the liver in the liver segmentation result, and calculate the third quantization value corresponding to the liver segmentation result based on the overall image information of the liver and the corresponding third quantization method.

[0116] In this step, the liver segmentation result mainly refers to the three-dimensional liver segmentation result. This is generated by segmenting the original DICOM data (Digital Imaging and Communications in Medicine) of the liver region of the subject and then performing 3D reconstruction. Therefore, the quality of the 3D reconstruction result of the liver segmentation is strongly correlated with the image quality of the original DICOM data. Furthermore, the matching of the contour edges of each anatomical structure generated during liver region segmentation with the actual anatomical structures in the DICOM data is also relevant. Thus, the quality of the liver segmentation result can be evaluated from these two aspects.

[0117] Therefore, the image information of the whole liver can be calculated by the shape information and the orientation information of each anatomical structure in the liver segmentation result, and the image information of the whole liver includes the data quality of the original data corresponding to the liver segmentation result and the contour fitting degree of the anatomical structure in the liver segmentation result and the original data.

[0118] It should be noted that the original Dicom data image in the embodiment of the present application is a liver enhanced CT (Computed Tomography, electronic computed tomography), which includes enhanced period images of the hepatic arterial phase and the portal venous phase, that is, images of two phases, so the original Dicom data image quality of the two phases needs to be considered in the evaluation process. In addition, whether the contour edge matches (i.e., the contour fitting degree) can be whether the segmented contour and the enhanced image development part (the hepatic artery development in the arterial phase and the portal vein development in the portal phase) in the original Dicom match the edge.

[0119] Further, when quantitatively processing the image information of the whole liver, the data quality of the original data and the contour fitting degree of the anatomical structure in the liver segmentation result and the original data can each have a corresponding third quantization method, and the corresponding third quantization value is calculated by the third quantization method. For example, the data quality of the original data can be quantitatively processed according to the data quality of the original data and the corresponding third quantization method to obtain the third quantization value corresponding to the data quality of the original data; the contour fitting degree can be quantitatively processed according to the contour fitting degree and the corresponding third quantization method to obtain the third quantization value corresponding to the contour fitting degree; and the third quantization value corresponding to the liver segmentation result can be determined according to the third quantization value corresponding to the data quality of the original data and the third quantization value corresponding to the contour fitting degree. The specific quantization item can be referred to in the description of the first embodiment. Figure 7

[0120] Specifically, a plurality of quality ranges can be set in advance, each quality range corresponding to a quantization value, that is, the data quality of the original data is set with a corresponding third quantization method. Then, after obtaining the data quality of the original data here, the data quality of the original data can be matched with the plurality of quality ranges to obtain the matched quality range and the corresponding quantization value, which is denoted as the third quantization value. The way to obtain the data quality of the original data here can be through manual comparison or neural network comparison, which is not limited here.

[0121] ​Meanwhile, a plurality of adhesion degree ranges can be set in advance, each of which corresponds to a quantization value, i.e., a third quantization mode corresponding to the contour adhesion degree of the anatomical structure in the liver segmentation result to the original data. Then, after obtaining the contour adhesion degree of the anatomical structure in the liver segmentation result to the original data, the obtained adhesion degree and the plurality of adhesion degree ranges can be matched to obtain the matched adhesion degree range and the corresponding quantization value, i.e., the third quantization value. Here, the way of obtaining the adhesion degree can be through manual comparison or neural network comparison, which is not limited here.

[0122] Finally, the third quantization value corresponding to the data quality of the original data and the third quantization value corresponding to the adhesion degree can be summed to obtain the third quantization value corresponding to the liver segmentation result.

[0123] S408, at least two of the first quantization value, the second quantization value, and the third quantization value are taken as initial quantization values.

[0124] In this step, after obtaining the first quantization value, the second quantization value, and the third quantization value corresponding to the liver segmentation result through the above calculation, at least two of them can be taken as initial quantization values, for example, all three quantization values can be taken as initial quantization values, and the sum of the three can be obtained, i.e., the quantization value corresponding to the liver segmentation result. In turn, the quality category corresponding to the liver segmentation result can be obtained through the quantization value corresponding to the liver segmentation result.

[0125] In this embodiment, the quantization value corresponding to each part is obtained through the orientation information and the morphological information of each anatomical structure in the liver segmentation result, the relative positional relationship between each anatomical structure, the image information of the whole liver, and the corresponding quantization mode. In turn, the quantization value of the liver segmentation result is obtained. Here, the quantization value of the liver segmentation result is calculated from multiple dimensions through multiple quantization modes, and the data dimensions considered are more, so the result of quality evaluation of the liver segmentation result is more accurate and reliable.

[0126] It should be understood that although each step in the flowchart involved in each of the above embodiments is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in each of the above embodiments can include a plurality of steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0127] Based on the same inventive concept, the embodiments of the present application also provide a liver segmentation quality evaluation device for implementing the above-mentioned liver segmentation quality evaluation method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more liver segmentation quality evaluation device embodiments provided below can be referred to the limitations of the liver segmentation quality evaluation method in the foregoing, which will not be described here again.

[0128] In one embodiment, as shown in Figure 8 A liver segmentation quality evaluation device is provided, comprising: an acquisition module 11, a quantification module 12, and an evaluation module 13, wherein:

[0129] The acquisition module 11 is configured to acquire a liver segmentation result to be evaluated, wherein the liver segmentation result comprises orientation information and morphological information of at least one anatomical structure.

[0130] The quantification module 12 is configured to input the liver segmentation result into a preset quantification model for quantification processing to determine a quantification value corresponding to the liver segmentation result, wherein the quantification model is determined based on the liver segmentation result and a quantification manner of the liver segmentation result, and the quantification manner is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result.

[0131] The evaluation module 13 is configured to determine a quality category corresponding to the liver segmentation result according to the quantification value corresponding to the liver segmentation result and a preset quantification value range, wherein each quantification value range corresponds to a quality category.

[0132] In another embodiment, the quantification model comprises a plurality of quantification manners of the liver segmentation result, and each quantification manner corresponds to a segmentation result of a different dimension of the liver segmentation result.

[0133] In another embodiment, another liver segmentation quality evaluation device is provided, and based on the above-mentioned embodiment, the quantification module 12 can comprise:

[0134] A calculation unit is configured to calculate a plurality of initial quantification values corresponding to the liver segmentation result according to the orientation information and the morphological information of each anatomical structure in the liver segmentation result and the corresponding quantification manner.

[0135] A synthesis unit is configured to synthesize the plurality of initial quantification values corresponding to the liver segmentation result to obtain the quantification value corresponding to the liver segmentation result.

[0136] In another embodiment, another liver segmentation quality evaluation device is provided, and based on the above-mentioned embodiment, the calculation unit can comprise:

[0137] The first calculation subunit is configured to calculate a first quantization value corresponding to the liver segmentation result according to the orientation information and the shape information of each anatomical structure in the liver segmentation result and a corresponding first quantization mode.

[0138] The second calculation subunit is configured to obtain a relative positional relationship between each anatomical structure according to the orientation information of each anatomical structure in the liver segmentation result, and calculate a second quantization value corresponding to the liver segmentation result according to the relative positional relationship between each anatomical structure and a corresponding second quantization mode.

[0139] The third calculation subunit is configured to obtain image information of the whole liver in the liver segmentation result according to the shape information and the orientation information of each anatomical structure in the liver segmentation result, and calculate a third quantization value corresponding to the liver segmentation result according to the image information of the whole liver and a corresponding third quantization mode.

[0140] The determination subunit is configured to take at least two of the first quantization value, the second quantization value and the third quantization value as an initial quantization value.

[0141] Optionally, the anatomical structures include at least one of a liver, a liver segment and a blood vessel; the first calculation subunit is specifically configured to calculate a first orientation quantization value corresponding to each anatomical structure according to the orientation information of each anatomical structure and a corresponding first orientation quantization mode; the first orientation quantization mode is used to quantize whether the corresponding anatomical structure exists in the liver segmentation result through the actual orientation information of the anatomical structure; calculate a first shape quantization value corresponding to each anatomical structure according to the shape information of each anatomical structure and a corresponding first shape quantization mode; the first shape quantization mode is used to quantize the shape of the anatomical structure in the liver segmentation result through the actual shape information of the anatomical structure; and the first orientation quantization value and the first shape quantization value are comprehensively calculated to obtain the first quantization value.

[0142] Optionally, the relative positional relationship between each anatomical structure includes a first relative positional relationship between a segmentation surface of a liver segment and a blood vessel and a second relative positional relationship between an internal part of the liver segment and a portal vein in the blood vessel; the second calculation subunit is specifically configured to quantitatively calculate the first relative positional relationship according to the first relative positional relationship and a corresponding second quantization mode to obtain a second quantization value corresponding to the first relative positional relationship; quantitatively calculate the second relative positional relationship according to the second relative positional relationship and a corresponding second quantization mode to obtain a second quantization value corresponding to the second relative positional relationship; and determine the second quantization value corresponding to the liver segmentation result according to the second quantization value corresponding to the first relative positional relationship and the second quantization value corresponding to the second relative positional relationship.

[0143] Optionally, the image information of the whole liver includes data quality of original data corresponding to the liver segmentation result and contour fitting degree of the anatomical structure in the liver segmentation result to the original data; the third calculation subunit is specifically configured to quantize the data quality of the original data according to the data quality of the original data and a third quantization mode corresponding to the data quality, to obtain a third quantization value corresponding to the data quality of the original data; quantize the contour fitting degree according to the contour fitting degree of the anatomical structure in the liver segmentation result to the original data and a third quantization mode corresponding to the contour fitting degree, to obtain a third quantization value corresponding to the contour fitting degree; and determine a third quantization value corresponding to the liver segmentation result according to the third quantization value corresponding to the data quality of the original data and the third quantization value corresponding to the contour fitting degree.

[0144] The modules in the liver segmentation quality evaluation device can be realized by software, hardware, or a combination thereof. The modules can be embedded in or independent of a processor in a computer device in hardware form, or stored in a memory in the computer device in software form, so as to be called and executed by the processor.

[0145] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the following steps when executing the computer program:

[0146] obtaining a liver segmentation result to be evaluated; the liver segmentation result including orientation information and morphological information of at least one anatomical structure; inputting the liver segmentation result into a preset quantization model for quantization processing to determine a quantization value corresponding to the liver segmentation result; wherein the quantization model is determined based on the liver segmentation result and a quantization mode of the liver segmentation result, the quantization mode being related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result; determining a quality category corresponding to the liver segmentation result according to the quantization value corresponding to the liver segmentation result and a preset quantization value range; wherein each quantization value range corresponds to a quality category.

[0147] In one embodiment, the quantization model includes multiple quantization modes of the liver segmentation result, each quantization mode corresponding to a segmentation result of different dimensions of the liver segmentation result.

[0148] In one embodiment, the processor further implements the following steps when executing the computer program:

[0149] calculating multiple initial quantization values corresponding to the liver segmentation result according to the orientation information and morphological information of each anatomical structure in the liver segmentation result and the corresponding quantization mode; and synthesizing the multiple initial quantization values corresponding to the liver segmentation result to obtain the quantization value corresponding to the liver segmentation result.

[0150] In one embodiment, the processor, when executing the computer program, also implements the following steps:

[0151] According to the orientation information and the shape information of each anatomical structure in the liver segmentation result and the corresponding first quantification manner, a first quantification value corresponding to the liver segmentation result is calculated; according to the orientation information of each anatomical structure in the liver segmentation result, a relative positional relationship between the anatomical structures is obtained, and according to the relative positional relationship between the anatomical structures and the corresponding second quantification manner, a second quantification value corresponding to the liver segmentation result is calculated; according to the shape information and the orientation information of each anatomical structure in the liver segmentation result, image information of the whole liver in the liver segmentation result is obtained, and according to the image information of the whole liver and the corresponding third quantification manner, a third quantification value corresponding to the liver segmentation result is calculated; at least two of the first quantification value, the second quantification value and the third quantification value are taken as initial quantification values.

[0152] In one embodiment, the processor, when executing the computer program, also implements the following steps:

[0153] According to the orientation information of each anatomical structure and the corresponding first orientation quantification manner, a first orientation quantification value corresponding to each anatomical structure is calculated; the first orientation quantification manner is used to quantize whether the corresponding anatomical structure exists in the liver segmentation result through the actual orientation information of the anatomical structure; according to the shape information of each anatomical structure and the corresponding first shape quantification manner, a first shape quantification value corresponding to each anatomical structure is calculated; the first shape quantification manner is used to quantize the shape of the anatomical structure in the liver segmentation result through the actual shape information of the anatomical structure; the first orientation quantification value and the first shape quantification value are comprehensively calculated to obtain a first quantification value.

[0154] In one embodiment, the processor, when executing the computer program, also implements the following steps:

[0155] According to the first relative positional relationship and the corresponding second quantification manner, the first relative positional relationship is quantitatively calculated to obtain a second quantification value corresponding to the first relative positional relationship; according to the second relative positional relationship and the corresponding second quantification manner, the second relative positional relationship is quantitatively calculated to obtain a second quantification value corresponding to the second relative positional relationship; according to the second quantification value corresponding to the first relative positional relationship and the second quantification value corresponding to the second relative positional relationship, a second quantification value corresponding to the liver segmentation result is determined.

[0156] In one embodiment, the processor, when executing the computer program, also implements the following steps:

[0157] According to the data quality of the original data and the third quantization mode corresponding thereto, the data quality of the original data is quantized to obtain a third quantization value corresponding to the data quality of the original data; according to the contour adhesion degree of the anatomical structure in the liver segmentation result and the original data and the third quantization mode corresponding thereto, the contour adhesion degree is quantized to obtain a third quantization value corresponding to the contour adhesion degree; and according to the third quantization value corresponding to the data quality of the original data and the third quantization value corresponding to the contour adhesion degree, a third quantization value corresponding to the liver segmentation result is determined.

[0158] In one embodiment, the anatomical structure includes at least one of a liver, a liver segment, and a blood vessel.

[0159] In one embodiment, a computer readable storage medium is provided, and a computer program is stored on the computer readable storage medium. When the computer program is executed by a processor, the following steps are implemented:

[0160] A liver segmentation result to be evaluated is obtained, and the liver segmentation result includes orientation information and morphological information of at least one anatomical structure. The liver segmentation result is input into a preset quantization model for quantization processing to determine a quantization value corresponding to the liver segmentation result. The quantization model is determined based on the liver segmentation result and a quantization mode of the liver segmentation result. The quantization mode is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result. According to the quantization value corresponding to the liver segmentation result and a preset quantization value range, a quality category corresponding to the liver segmentation result is determined. Each quantization value range corresponds to a quality category.

[0161] In one embodiment, the quantization model includes a plurality of quantization modes of the liver segmentation result, and each quantization mode corresponds to a segmentation result of a different dimension of the liver segmentation result.

[0162] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0163] According to the orientation information and morphological information of each anatomical structure in the liver segmentation result and the corresponding quantization mode, a plurality of initial quantization values corresponding to the liver segmentation result are calculated. The plurality of initial quantization values corresponding to the liver segmentation result are integrated to obtain a quantization value corresponding to the liver segmentation result.

[0164] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0165] According to the orientation information and the shape information of each anatomical structure in the liver segmentation result and the corresponding first quantification mode, a first quantification value corresponding to the liver segmentation result is calculated; according to the orientation information of each anatomical structure in the liver segmentation result, a relative position relationship between the anatomical structures is obtained, and according to the relative position relationship between the anatomical structures and the corresponding second quantification mode, a second quantification value corresponding to the liver segmentation result is calculated; according to the shape information and the orientation information of each anatomical structure in the liver segmentation result, image information of the whole liver in the liver segmentation result is obtained, and according to the image information of the whole liver and the corresponding third quantification mode, a third quantification value corresponding to the liver segmentation result is calculated; at least two of the first quantification value, the second quantification value and the third quantification value are taken as initial quantification values.

[0166] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0167] According to the orientation information of each anatomical structure and the corresponding first orientation quantification mode, a first orientation quantification value corresponding to each anatomical structure is calculated; the first orientation quantification mode is used to quantize whether the corresponding anatomical structure exists in the liver segmentation result through the actual orientation information of the anatomical structure; according to the shape information of each anatomical structure and the corresponding first shape quantification mode, a first shape quantification value corresponding to each anatomical structure is calculated; the first shape quantification mode is used to quantize the shape of the anatomical structure in the liver segmentation result through the actual shape information of the anatomical structure; the first orientation quantification value and the first shape quantification value are comprehensively calculated to obtain the first quantification value.

[0168] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0169] According to the first relative position relationship and the corresponding second quantification mode, the first relative position relationship is quantitatively calculated to obtain a second quantification value corresponding to the first relative position relationship; according to the second relative position relationship and the corresponding second quantification mode, the second relative position relationship is quantitatively calculated to obtain a second quantification value corresponding to the second relative position relationship; according to the second quantification value corresponding to the first relative position relationship and the second quantification value corresponding to the second relative position relationship, the second quantification value corresponding to the liver segmentation result is determined.

[0170] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0171] According to the data quality of the original data and the third quantization mode corresponding thereto, the data quality of the original data is quantized to obtain a third quantization value corresponding to the data quality of the original data; according to the contour adhesion degree of the anatomical structure in the liver segmentation result and the original data and the third quantization mode corresponding thereto, the contour adhesion degree is quantized to obtain a third quantization value corresponding to the contour adhesion degree; and according to the third quantization value corresponding to the data quality of the original data and the third quantization value corresponding to the contour adhesion degree, a third quantization value corresponding to the liver segmentation result is determined.

[0172] In one embodiment, the anatomical structure includes at least one of a liver, a liver segment, and a blood vessel.

[0173] In one embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the following steps:

[0174] A liver segmentation result to be evaluated is obtained; the liver segmentation result includes orientation information and morphological information of at least one anatomical structure; the liver segmentation result is input into a preset quantization model for quantization processing to determine a quantization value corresponding to the liver segmentation result; the quantization model is determined based on the liver segmentation result and a quantization mode of the liver segmentation result, and the quantization mode is related to actual orientation information and actual morphological information of the anatomical structure in the liver segmentation result; a quality category corresponding to the liver segmentation result is determined according to the quantization value corresponding to the liver segmentation result and a preset quantization value range; each quantization value range corresponds to a quality category.

[0175] In one embodiment, the quantization model includes a plurality of quantization modes of the liver segmentation result, and each quantization mode corresponds to a segmentation result of a different dimension of the liver segmentation result.

[0176] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0177] According to the orientation information and morphological information of each anatomical structure in the liver segmentation result and the corresponding quantization mode, a plurality of initial quantization values corresponding to the liver segmentation result are calculated; and the plurality of initial quantization values corresponding to the liver segmentation result are integrated to obtain a quantization value corresponding to the liver segmentation result.

[0178] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0179] According to the orientation information and the shape information of each anatomical structure in the liver segmentation result and the corresponding first quantification mode, a first quantification value corresponding to the liver segmentation result is calculated; according to the orientation information of each anatomical structure in the liver segmentation result, a relative position relationship between the anatomical structures is obtained, and according to the relative position relationship between the anatomical structures and the corresponding second quantification mode, a second quantification value corresponding to the liver segmentation result is calculated; according to the shape information and the orientation information of each anatomical structure in the liver segmentation result, image information of the whole liver in the liver segmentation result is obtained, and according to the image information of the whole liver and the corresponding third quantification mode, a third quantification value corresponding to the liver segmentation result is calculated; at least two of the first quantification value, the second quantification value and the third quantification value are taken as initial quantification values.

[0180] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0181] According to the orientation information of each anatomical structure and the corresponding first orientation quantification mode, a first orientation quantification value corresponding to each anatomical structure is calculated; the first orientation quantification mode is used to quantize whether the corresponding anatomical structure exists in the liver segmentation result through the actual orientation information of the anatomical structure; according to the shape information of each anatomical structure and the corresponding first shape quantification mode, a first shape quantification value corresponding to each anatomical structure is calculated; the first shape quantification mode is used to quantize the shape of the anatomical structure in the liver segmentation result through the actual shape information of the anatomical structure; the first orientation quantification value and the first shape quantification value are comprehensively calculated to obtain the first quantification value.

[0182] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0183] According to the first relative position relationship and the corresponding second quantification mode, the first relative position relationship is quantitatively calculated to obtain a second quantification value corresponding to the first relative position relationship; according to the second relative position relationship and the corresponding second quantification mode, the second relative position relationship is quantitatively calculated to obtain a second quantification value corresponding to the second relative position relationship; according to the second quantification value corresponding to the first relative position relationship and the second quantification value corresponding to the second relative position relationship, the second quantification value corresponding to the liver segmentation result is determined.

[0184] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0185] According to the data quality of the original data and the corresponding third quantization manner, the data quality of the original data is quantized to obtain a third quantization value corresponding to the data quality of the original data; according to the contour fitting degree of the anatomical structure and the original data in the liver segmentation result and the corresponding third quantization manner, the contour fitting degree is quantized to obtain a third quantization value corresponding to the contour fitting degree; and according to the third quantization value corresponding to the data quality of the original data and the third quantization value corresponding to the contour fitting degree, a third quantization value corresponding to the liver segmentation result is determined.

[0186] In one embodiment, the anatomical structure includes at least one of a liver, a liver segment, and a blood vessel.

[0187] It should be noted that the data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or fully authorized by all parties.

[0188] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0189] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0190] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for assessing the quality of liver segmentation, characterized in that, The method includes: Obtain the liver segmentation result to be evaluated; the liver segmentation result includes the orientation and morphological information of at least one anatomical structure; the orientation information includes the relative positional relationship between the anatomical structures, the relative positional relationship between the anatomical structures includes the first relative positional relationship between the segmentation plane of the liver segment and the blood vessels, and the second relative positional relationship between the interior of the liver segment and the portal vein in the blood vessels; The liver segmentation result is input into a preset quantization model for quantization processing to determine multiple initial quantization values ​​corresponding to the liver segmentation result. Based on these initial quantization values, the corresponding quantization value for the liver segmentation result is then determined. The quantization model is determined based on the liver segmentation result and multiple quantization methods for the liver segmentation result. Each quantization method is related to the actual orientation and morphological information of the anatomical structures in the liver segmentation result. Each quantization method corresponds to a segmentation result in a different dimension of the liver segmentation result. The multiple initial quantization values ​​correspond to the multiple quantization methods. The quantization model is trained using the gold standard quantization results of the data in each dimension of the liver segmentation result under each quantization method. The different dimensions are related to the orientation information and the morphological information. Based on the quantization value corresponding to the liver segmentation result and the preset quantization value range, the quality category corresponding to the liver segmentation result is determined; wherein, each quantization value range corresponds to a quality category.

2. The method according to claim 1, characterized in that, The quantification model is a machine learning model, or a model that characterizes the correlation between liver segmentation results and their quantification methods.

3. The method according to claim 1, characterized in that, The step of inputting the liver segmentation result into a preset quantization model for quantization processing to determine the quantization value corresponding to the liver segmentation result includes: Based on the orientation and morphological information of each anatomical structure in the liver segmentation result and the corresponding quantization method, calculate multiple initial quantization values ​​corresponding to the liver segmentation result. The multiple initial quantization values ​​corresponding to the liver segmentation result are combined to obtain the quantization value corresponding to the liver segmentation result.

4. The method according to claim 3, characterized in that, The step involves calculating multiple initial quantization values ​​corresponding to the liver segmentation result based on the orientation and morphological information of each anatomical structure in the liver segmentation result and the corresponding quantization method, including: Based on the orientation and morphological information of each anatomical structure in the liver segmentation result and the corresponding first quantization method, calculate the first quantization value corresponding to the liver segmentation result; Based on the orientation information of each anatomical structure in the liver segmentation result, the relative positional relationship between each anatomical structure is obtained, and based on the relative positional relationship between each anatomical structure and the corresponding second quantization method, the second quantization value corresponding to the liver segmentation result is calculated. Based on the morphological and orientational information of each anatomical structure in the liver segmentation result, the image information of the whole liver in the liver segmentation result is obtained, and based on the image information of the whole liver and the corresponding third quantization method, the third quantization value corresponding to the liver segmentation result is calculated. At least two of the first quantization value, the second quantization value, and the third quantization value are used as the initial quantization value.

5. The method according to claim 4, characterized in that, The step of calculating the first quantization value corresponding to the liver segmentation result based on the orientation and morphological information of each anatomical structure in the liver segmentation result and the corresponding first quantization method includes: Based on the orientation information of each anatomical structure and the corresponding first orientation quantification method, the first orientation quantification value corresponding to each anatomical structure is calculated; the first orientation quantification method is used to quantify whether there is a corresponding anatomical structure in the liver segmentation result through the actual orientation information of the anatomical structure. Based on the morphological information of each anatomical structure and the corresponding first morphological quantification method, the first morphological quantification value corresponding to each anatomical structure is calculated; the first morphological quantification method is used to quantify the morphology of the anatomical structures in the liver segmentation result through the actual morphological information of the anatomical structures. The first quantization value is obtained by comprehensively calculating the first orientation quantization value and the first morphology quantization value.

6. The method according to claim 4, characterized in that, The step of calculating the second quantization value corresponding to the liver segmentation result based on the relative positional relationship between the anatomical structures and the corresponding second quantization method includes: Based on the first relative positional relationship and its corresponding second quantization method, the first relative positional relationship is quantized to obtain the second quantization value corresponding to the first relative positional relationship; Based on the second relative position relationship and its corresponding second quantization method, the second relative position relationship is quantized to obtain the second quantization value corresponding to the second relative position relationship; The second quantization value corresponding to the liver segmentation result is determined based on the second quantization value corresponding to the first relative positional relationship and the second quantization value corresponding to the second relative positional relationship.

7. The method according to claim 4, characterized in that, The overall image information of the liver includes the data quality of the original data corresponding to the liver segmentation result and the contour fit between the anatomical structure in the liver segmentation result and the original data. The step of calculating the third quantization value corresponding to the liver segmentation result based on the overall image information of the liver and the corresponding third quantization method includes: Based on the data quality of the original data and its corresponding third quantization method, the data quality of the original data is quantized to obtain the third quantization value corresponding to the data quality of the original data. Based on the contour fit between the anatomical structure and the original data in the liver segmentation results and its corresponding third quantization method, the contour fit is quantized to obtain the third quantization value corresponding to the contour fit. The third quantization value corresponding to the liver segmentation result is determined based on the third quantization value corresponding to the data quality of the original data and the third quantization value corresponding to the contour fit.

8. The method according to any one of claims 1-7, characterized in that, The anatomical structures include at least one of the following: liver, liver segments, and blood vessels.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images