Auxiliary film reading system suitable for patient with cryptococcus lung disease

Abstract

The invention relates to the technical field of lung image marking auxiliary film reading, in particular to an auxiliary film reading system suitable for a patient suffering from a cryptococcus lung disease. The method comprises the following steps: firstly, acquiring continuous CT slice images of a lung, and extracting a foreground connected domain of a lung parenchyma region as a connected domain to be analyzed; further obtaining the blood vessel possibility of the target connected domain according to the monotonicity of the width change of the target connected domain; further according to the monotonicity of the area change of the to-be-analyzed connected domain at the same position, obtaining the blood vessel penetrating possibility of the target connected domain; obtaining the quasi-circular degree of the target connected domain; and finally, marking the target connected domain according to the quasi-circular degree, the blood vessel fusion possibility and the blood vessel penetration possibility. According to the method, the possibility that the target connected domain belongs to the blood vessel is measured from the angles of width change and area change, the blood vessel fusion possibility and the blood vessel penetrating possibility are fused by means of the quasi-circular degree, the connected domain is accurately marked, and related personnel are assisted in reading the film.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lung image marking assisted reading, in particular to an auxiliary reading system suitable for patients with pulmonary cryptococcosis. BACKGROUND

[0002] Pulmonary infection caused by cryptococcus usually presents a nodular mass feature in the case of normal immune system of the patient, and such feature is easy to be identified, but is easy to be misjudged as tumor or tuberculosis; in the case of immune function suppression of the patient, it may present a variety of "satellite type" lesions, and the morphology of such lesions is variable, common diffuse, scattered infiltration, consolidation and other features, and due to the high density of lesion tissue and blood vessel tissue, a similar white feature is generated, so that the gray scale of lesion pixels is easy to overlap with that of lung blood vessel pixels, and interference is caused, so that it is difficult to distinguish the tiny lesion. SUMMARY

[0003] In order to solve the technical problem that the lesion tissue and blood vessel tissue of pulmonary cryptococcus are difficult to distinguish, the purpose of the present application is to provide an auxiliary reading system suitable for patients with pulmonary cryptococcosis, and the technical scheme is as follows: An image acquisition module: acquiring continuous CT slice images of the lung of the current patient; extracting a lung parenchyma region in each CT slice image through image segmentation; extracting a foreground connected domain of the lung parenchyma region as a connected domain to be analyzed; and matching the connected domains to be analyzed at the same position in all CT slice images; An image analysis module: selecting any connected domain to be analyzed as a target connected domain; acquiring a blood vessel possibility of the target connected domain according to the monotonicity of the width change of the target connected domain; acquiring a through blood vessel possibility of the target connected domain according to the monotonicity of the area change of the connected domains to be analyzed matched by the target connected domain; and acquiring a degree of circle-like shape of the target connected domain; An image labeling module: labeling the target connected domain according to the degree of circle-like shape, and fusing the blood vessel possibility and the through blood vessel possibility.

[0004] Further, the method for acquiring the blood vessel possibility comprises: Taking a line connecting two farthest pixel points in the target connected domain as a reference line, and moving from one end of the reference line to the other end with a single pixel point as a step; after each movement, acquiring a pixel width of the target connected domain in the vertical direction of the reference line, and constructing a width sequence of the target connected domain; According to the monotonicity of the width sequence, the blood vessel possibility of the target connected domain is acquired.

[0005] Further, the method for obtaining the blood vessel possibility of the target connected domain according to the monotonicity of the width sequence comprises: obtaining the blood vessel possibility of the target connected domain according to the correlation between the width sequence and a serial number sequence composed of serial numbers of elements in the width sequence; the correlation between the width sequence and the serial number sequence is positively correlated with the blood vessel possibility.

[0006] Further, the method for obtaining the through blood vessel possibility comprises: obtaining the area of the target connected domain matched with the connected domain to be analyzed, and sorting the area according to the acquisition order of the CT slice image to obtain an area sequence; obtaining the through blood vessel possibility of the target connected domain according to the monotonicity of the area sequence.

[0007] Further, the method for obtaining the through blood vessel possibility of the target connected domain according to the monotonicity of the area sequence comprises: obtaining the through blood vessel possibility of the target connected domain according to the number of extreme points other than the two ends in the area sequence; the number of extreme points other than the two ends is negatively correlated with the through blood vessel possibility.

[0008] Further, the method for marking the target connected domain comprises: fusing the circle-like degree and the blood vessel possibility to obtain a first normal possibility; the circle-like degree is negatively correlated with the first normal possibility; the blood vessel possibility is positively correlated with the first normal possibility; fusing the circle-like degree and the through blood vessel possibility to obtain a second normal possibility; the circle-like degree and the through blood vessel possibility are positively correlated with the second normal possibility; fusing the first normal possibility and the second normal possibility to obtain a normal tissue possibility of the target connected domain; the first normal possibility and the second normal possibility are positively correlated with the normal tissue possibility; when the normal tissue possibility is lower than a preset normal threshold, marking the target connected domain.

[0009] Further, the circle-like degree is the circularity of the contour of the outermost pixel point of the target connected domain.

[0010] Further, the method for matching the connected domain to be analyzed at the same position in all the CT slice images comprises: When the proportion of overlapping of the pixel point in the to-be-analyzed connected domain in one image and the pixel point in the to-be-analyzed connected domain in another image is greater than a preset threshold, it is determined that the two corresponding to-be-analyzed connected domains are matched.

[0011] Further, the lung parenchyma region of the CT slice image is extracted by a pre-trained neural network.

[0012] Further, the foreground connected domain acquisition method comprises: The region growing algorithm is selected, a seed point is randomly selected from the pixel point with the gray value greater than the gray mean value of the lung parenchyma region, and the foreground connected domain is obtained.

[0013] The present application has the following advantages: The present application firstly acquires the continuous CT slice image of the lung, obtains the analysis basis, further extracts the lung parenchyma region and the foreground connected domain therein as the to-be-analyzed connected domain, gradually narrows the analysis range and excludes irrelevant region interference, further acquires the blood vessel possibility and the through blood vessel possibility of the target connected domain according to the monotonicity of the width change of the target connected domain and the monotonicity of the area change of the to-be-analyzed connected domain at the same position, analyzes the shape change from the width change and the area change, measures the possibility that the target connected domain belongs to the blood vessel, provides the basis for subsequent marking image, further acquires the degree of circle of the target connected domain, adjusts the blood vessel possibility and the through blood vessel possibility, comprehensively analyzes the possibility that the target connected domain belongs to the blood vessel tissue, and thus prepares for marking. Finally, the target connected domain is marked according to the degree of circle, the blood vessel possibility and the through blood vessel possibility. The present application measures the possibility that the target connected domain belongs to the blood vessel from the width change and the area change, and fuses the blood vessel possibility and the through blood vessel possibility by means of the degree of circle, accurately marks the connected domain, and assists the relevant personnel in reading the film. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, a brief introduction will be given to the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only show some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0015] Figure 1 A system block diagram of an auxiliary film reading system suitable for a patient with pulmonary cryptococcosis provided by an embodiment of the present application; Figure 2 A CT slice image provided by an embodiment of the present application; Figure 3A lung parenchyma region image provided by an embodiment of the present application; Figure 4 A distribution diagram of the connected domain to be analyzed provided by an embodiment of the present application; Figure 5 A performance diagram of lung blood vessels in continuous slices provided by an embodiment of the present application; Figure 6 A performance diagram of lung blood vessels in a single slice provided by an embodiment of the present application; Figure 7 A diagram for calculating the pixel width provided by an embodiment of the present application; Figure 8 A diagram for the connected domain circumscribed circle reduction process provided by an embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to further illustrate the technical means and effects taken by the present application to achieve the predetermined purposes, the following describes in detail the specific implementation, structure, features and effects of an auxiliary reading system for patients with pulmonary cryptococcosis according to the present application, in combination with the preferred embodiments and the accompanying drawings. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0018] The following describes in detail the specific scheme of an auxiliary reading system for patients with pulmonary cryptococcosis provided by the present application, in combination with the accompanying drawings.

[0019] Referring to Figure 1 , a system block diagram of an auxiliary reading system for patients with pulmonary cryptococcosis provided by an embodiment of the present application is shown, which includes an image acquisition module 101, an image analysis module 102 and an image labeling module 103.

[0020] The image acquisition module 101 acquires continuous CT slice images of the lung of the current patient, extracts the lung parenchyma region in each CT slice image through image segmentation, extracts the foreground connected domain of the lung parenchyma region as the connected domain to be analyzed, and matches the connected domains to be analyzed at the same position in all CT slice images.

[0021] In one embodiment of the present application, the chest is scanned by a high-resolution thin-layer CT machine, and the CT scanning layer thickness is limited to no more than 2 mm; the scanning time should be completed within 30 seconds as much as possible. At the same time, the CT of the scanned chest image is stored in real time, stored according to the scanning order, and the continuous CT slice image of the current patient's lung is obtained, which is the basis for analysis.

[0022] Please refer to Figure 2 which shows a CT slice image provided by one embodiment of the present application; from Figure 2 It can be seen that the obtained CT slice image includes the chest, spine and other irrelevant regions of the patient, so the lung parenchyma region in each CT slice image is extracted by image segmentation to reduce the analysis range.

[0023] In one embodiment of the present application, the lung parenchyma region of the CT slice image is extracted by a pre-trained neural network. The neural network architecture is a CNN network, the input is the scanned chest CT image, and the output is an image including only the lung parenchyma region; the training set is the manually labeled lung parenchyma region contour; and the loss function is a cross-entropy loss function. Please refer to Figure 3 which shows a lung parenchyma region image provided by one embodiment of the present application.

[0024] It should be noted that the neural network and its training method are well-known technical means to those skilled in the art, and in other embodiments of the present application, the implementer can also use existing graph cut algorithm and other segmentation methods, which will not be described here.

[0025] During the CT image shooting process, organs with low tissue density generally appear darker in color, so that the soft tissue appears gray-white features. Therefore, most of the lung parenchyma region contains air and appears darker in color, and the blood in the corresponding lung tissue has a larger density, and the blood vessels appear lighter in gray scale.

[0026] In addition, after pulmonary cryptococcosis infection, local tissues may produce nodules, exudation and ground glass shadows, resulting in local density increasing, so that the lung presents a white pixel region, which may be normal blood vessels on the one hand, and may also be diseased tissue on the other hand. Therefore, the foreground connected domain of the lung parenchyma region is further extracted as the connected domain to be analyzed to further reduce the analysis range.

[0027] Preferably, in one embodiment of the present application, the region growing algorithm is selected, and considering that the gray value of the blood vessel region and the diseased tissue region is high, a seed point is randomly selected from the pixel points with a gray value greater than the average gray value of the lung parenchyma region to obtain the foreground connected domain.

[0028] Specifically, the seed points are randomly selected uniformly, the number of the seed points is not less than 20, and the number is set to 25, the growth criterion is set to a relative threshold value of 30, and the neighborhood is set to an eight-neighborhood. When the absolute value of the difference between the gray value of the pixel point that is not grown in the eight-neighborhood of all the newly grown pixel points and the average gray value of the grown region is greater than 30, the growth is stopped.

[0029] Referring to Figure 4 which shows a distribution diagram of a to-be-analyzed connected domain provided by one embodiment of the present application, Figure 4 The connected domain with a gray value of 0 in the image is a background connected domain, and the connected domain with a gray value of 255 is a foreground connected domain, which is a to-be-analyzed connected domain.

[0030] It should be noted that in other embodiments of the present application, the foreground pixel points can also be extracted by using an Otsu threshold segmentation algorithm. The Otsu threshold segmentation algorithm and the region growing algorithm are both known technical means to those skilled in the art, and will not be described here.

[0031] It is considered that the same blood vessel tissue or lesion tissue has different performances in different depth CT slice images, so the to-be-analyzed connected domains at the same position in all CT slice images are matched, so as to facilitate the analysis of the change characteristics of the connected domains of the same tissue at different depths.

[0032] Preferably, in one embodiment of the present application, when the overlapping ratio of the pixel points in the to-be-analyzed connected domain in one frame image and the pixel points in the to-be-analyzed connected domain in another frame image is greater than a preset threshold value, it is determined that the corresponding two to-be-analyzed connected domains are matched.

[0033] As an example, the overlapping area of the to-be-analyzed connected domains in the two frames of images is taken as the numerator, the minimum value of the areas of the two to-be-analyzed connected domains is taken as the denominator, the ratio of the numerator to the denominator is taken as the overlapping ratio, and the preset threshold value is 0.5. When the overlapping ratio is greater than 0.5, it is determined that the corresponding two to-be-analyzed connected domains are matched.

[0034] In another embodiment of the present application, a matching restriction condition of a similar centroid can also be added, such as calculating the Euclidean distance of the centroid coordinates of the to-be-analyzed connected domains in adjacent CT slices, setting a matching threshold value such as 5, and determining that the corresponding two to-be-analyzed connected domains are matched when the overlapping ratio is higher than 0.5 and the Euclidean distance is less than 5.

[0035] It should be noted that obtaining the centroid of the connected domain, establishing the coordinate system and calculating the Euclidean distance are all known technical means to those skilled in the art, and will not be described here.

[0036] The image analysis module 102: selects any to-be-analyzed connected domain as a target connected domain; obtains the blood vessel possibility of the target connected domain according to the monotonicity of the width variation of the target connected domain; obtains the through blood vessel possibility of the target connected domain according to the monotonicity of the area variation of the to-be-analyzed connected domain matched by the target connected domain; and obtains the degree of circularity of the target connected domain.

[0037] Considering that the gray scales of the blood vessel tissue and the lesion tissue are similar and it is difficult to distinguish them from the gray scale angle, the possibility that the to-be-analyzed connected domain belongs to the blood vessel tissue is determined by analyzing the morphological variation, so as to finally mark the lesion region which is most likely to be the lung cryptococcus and assist the relevant personnel in reading the film; first, any to-be-analyzed connected domain is selected as a target connected domain, so as to facilitate the analysis one by one; the analysis method for each to-be-analyzed connected domain is consistent, and here, only the target connected domain is taken as an example for description, and the repeated description is not given.

[0038] Referring to Figure 5 , a schematic diagram showing the performance of the lung blood vessel in continuous slices is shown; Figure 5 Three continuous CT slices are shown in the figure, and the spatial relationship between the lung blood vessel and the slices is shown.

[0039] Referring to Figure 6 , a schematic diagram showing the performance of the lung blood vessel in a single slice is shown; Figure 6 The slice transversely passes through the lung blood vessel and presents a tubular condition; and the slice longitudinally passes through the lung blood vessel and presents a circular condition.

[0040] Combined with Figure 5 and Figure 6 It can be seen that the normal blood vessel tissue is diffused from the main lung blood vessel to the periphery in the three-dimensional space, and the pipe diameter of the blood vessel from the main blood vessel to the end presents a variation characteristic from large to small, so that the width of the normal blood vessel presents a continuous monotonic variation; Meanwhile, part of the blood vessel may present a hole on a single CT image, indicating that the blood vessel penetrates the current CT slice, so that the size of the through hole at the same position in the continuous slices also has strong monotonicity, so the blood vessel possibility of the target connected domain is obtained according to the monotonicity of the width variation of the target connected domain; the through blood vessel possibility of the target connected domain is obtained according to the monotonicity of the area variation of the to-be-analyzed connected domain matched by the target connected domain; and the possibility that the target connected domain belongs to the blood vessel is measured from the width variation and the area variation respectively, thereby providing a basis for subsequent marking of the image.

[0041] Preferably, in one embodiment of the present application: considering that the blood vessels usually appear as an elongated structure on the CT slice, taking the line connecting the two farthest pixels in the target connected domain as the reference line can make the reference line roughly aligned with the main axis direction of the blood vessels, thus presenting the main morphology of the target connected domain and making the width measurement as close as possible to the cross-sectional width of the real blood vessels; Moving from one end of the reference line to the other end by single pixel point as a step; after each movement, acquiring the pixel width of the target connected domain in the vertical direction of the reference line to construct the width sequence of the target connected domain; wherein, in order to avoid the inaccuracy of the width change analysis caused by measuring the width of multiple blood vessel branches, only the pixel width of the target connected domain in the vertical direction of the reference line is acquired.

[0042] Please refer to Figure 7 which shows a schematic diagram for calculating the pixel width provided by the present application; Figure 7 The target connected domain is contained in the figure, the dashed straight line in the figure is the reference line, the straight line perpendicular to the reference line is used to count the pixel width, and only the pixel width of the target connected domain in the vertical direction of the reference line is shown; the pixel width is the number of pixel points.

[0043] The width sequence represents the width change characteristics of the target connected domain, so the blood vessel possibility of the target connected domain is further acquired according to the monotonicity of the width sequence.

[0044] As an example: considering that the width change presents strong monotonicity, the width sequence presents monotonic change with the position sequence number, and the blood vessel possibility of the target connected domain is acquired according to the correlation between the width sequence and the sequence number sequence composed of the element sequence number in the width sequence; since the sequence number sequence presents monotonicity, the stronger the correlation between the width sequence and the sequence number sequence at this time, the more in line with the width change characteristics of the blood vessel structure, so the correlation between the width sequence and the sequence number sequence is positively correlated with the blood vessel possibility.

[0045] Specifically, considering that the Spearman rank correlation coefficient is a statistical method for measuring the monotonic relationship between two variables, the greater the absolute value of the Spearman rank correlation coefficient, the stronger the correlation, and the stronger the monotonicity of the width sequence, the absolute value of the Spearman rank correlation coefficient between the width sequence and the sequence number sequence is taken as the blood vessel possibility of the target connected domain.

[0046] In another embodiment of the present application, the skeleton of the target connected domain can also be extracted, the topological structure and key morphological characteristics thereof are retained, the skeleton line where the two farthest pixels are located is selected as the reference line, the pixel width of the original connected domain corresponding to the reference line is acquired, and the width sequence is obtained.

[0047] In other embodiments of the present application, the Pearson correlation coefficient can also be used to analyze the correlation between the width sequence and the sequence number, and the absolute value of the Pearson correlation coefficient can be used as the blood vessel possibility. The skeleton extraction method and the Pearson correlation coefficient and the Spearman rank correlation coefficient are all prior art and will not be described here.

[0048] Preferably, in an embodiment of the present application, considering that part of the blood vessels passes through the continuous CT slices, the area change of the connected domain of such blood vessels in the continuous slices reflects the change of the diameter of the blood vessels, so the area of the connected domain to be analyzed matched with the target connected domain is obtained, and the area sequence is obtained according to the acquisition sequence of the CT slice images. According to the monotonicity of the area sequence, the possibility of the target connected domain being penetrated by the blood vessels is obtained.

[0049] As an example: considering that the stronger the monotonicity of the data in the area sequence is, the fewer the extreme points are, and the extreme points at both ends of the sequence cannot reflect the monotonicity of the data change, the possibility of the target connected domain being penetrated by the blood vessels is obtained according to the number of extreme points other than the extreme points at both ends in the area sequence. The number of extreme points other than the extreme points at both ends is negatively correlated with the possibility of being penetrated by the blood vessels.

[0050] Specifically, the sum of the number of extreme points other than the extreme points at both ends in the area sequence and the preset positive parameter 1 is taken as the reciprocal and linearly normalized, and the normalized result is used as the possibility of being penetrated by the blood vessels.

[0051] It should be noted that the area of the connected domain is the number of pixel points of the connected domain, and the method of obtaining the extreme points is a known technical means in the art and will not be described here.

[0052] In the CT slice images, the cross-sectional shape of the blood vessels may change due to the slice direction and the shape of the blood vessels themselves, resulting in different reference values of the blood vessel possibility and the possibility of being penetrated by the blood vessels in different cases, so the degree of circularity of the target connected domain is also obtained as an additional judgment factor, which can improve the accuracy.

[0053] Preferably, in an embodiment of the present application, the degree of circularity is the circularity of the contour of the outermost pixel points of the target connected domain.

[0054] It should be noted that the calculation method of the circularity is a prior art and will not be described here.

[0055] In another embodiment of the present application, considering that when the connecting circle of the circular connected domain gradually shrinks inward, the smooth pit in the continuous space appears as a stepped edge in the discrete pixels, resulting in a jump increase in the number of difference pixels when the circumscribed circle first contacts, and the number of pixels that changes is large when it is closest to the boundary of the circle, the degree of circularity is obtained by using this characteristic.

[0056] Referring to Figure 8 Fig. 2 shows a schematic diagram of a circumscribed circle reduction process according to an embodiment of the present application; Figure 8 The figure contains a to-be-analyzed connected domain, the largest circle is the minimum circumscribed circle of the to-be-analyzed connected domain, the single pixel point is gradually reduced to the center of the circle, the arrow represents the reduction direction, and the smallest circle is the circle after the first reduction. After each reduction, the number of pixel points of the to-be-analyzed connected domain passing through the circle is obtained, and the absolute value of the difference between the number of pixel points corresponding to adjacent two reductions and the average value of the radius are recorded. The absolute value of the difference between the number of pixel points corresponding to adjacent two reductions is taken as the numerator, and the average value of the radius of the two reductions is taken as the denominator. After linear normalization, the fraction ratio is taken as the circular performance degree corresponding to each reduction operation. The maximum value of the circular performance degrees corresponding to all reduction operations is taken as the degree of circularity.

[0057] The image labeling module 103 labels the target connected domain according to the degree of circularity, fuses the blood vessel possibility and the through blood vessel possibility, and labels the target connected domain.

[0058] The blood vessel possibility and the through blood vessel possibility measure the possibility of the target connected domain belonging to the blood vessel from the width change and the area change angles, respectively. The degree of circularity is also obtained to adjust the reference weight of the two possibilities. Therefore, the blood vessel possibility and the through blood vessel possibility are fused according to the degree of circularity to improve the recognition accuracy of blood vessels with different morphological manifestations, label the target connected domain, ensure the labeling accuracy, and improve the auxiliary film reading effect.

[0059] Preferably, in an embodiment of the present application, the first normal possibility is obtained by fusing the degree of circularity and the blood vessel possibility. It is considered that the smaller the degree of circularity is, the more likely the target connected domain is a cross section of a blood vessel, the greater the reference value of the blood vessel possibility is, the greater the blood vessel possibility is, the stronger the monotonicity of the width change of the target connected domain is, the more consistent with the width change characteristics of the blood vessel, and the more likely the target connected domain is normal tissue. Therefore, the degree of circularity is negatively correlated with the first normal possibility, and the blood vessel possibility is positively correlated with the first normal possibility.

[0060] The second normal possibility is further obtained by fusing the degree of circularity and the through blood vessel possibility. It is considered that the greater the degree of circularity is, the more likely the target connected domain is a through surface of a blood vessel, the greater the reference value of the through blood vessel possibility is, the greater the through blood vessel possibility is, the stronger the monotonicity of the area change of the target connected domain in the continuous slices is, the more consistent with the width change characteristics of the blood vessel, and the more likely the target connected domain is normal tissue. Therefore, the degree of circularity and the through blood vessel possibility are positively correlated with the second normal possibility.

[0061] Further fuse the first normal possibility and the second normal possibility to obtain a normal tissue possibility of the target connected domain; It is considered that the greater the first normal possibility and the second normal possibility, the more likely it is normal tissue, so the first normal possibility and the second normal possibility are positively correlated with the normal tissue possibility.

[0062] Finally, when the normal tissue possibility is lower than a preset normal threshold, the target connected domain is marked.

[0063] As an example, for the target connected domain, a product of a constant 1 and a difference value of the circle-like degree and the blood vessel possibility is taken as the first normal possibility, a product of the circle-like degree and the through blood vessel possibility is taken as the second normal possibility, and a sum value of the first normal possibility and the second normal possibility is taken as the normal tissue possibility; the preset normal threshold is 0.6, and when the normal tissue possibility is lower than 0.6, the target connected domain is marked as a pulmonary cryptococcosis infection area.

[0064] In the calculation of the difference value of the constant 1 and the circle-like degree, the circle-like degree is the minuend, and the circle-like degree is negatively correlated by subtracting the circle-like degree from 1, so that the calculation logic relationship is adjusted.

[0065] It should be noted that the marking of the target connected domain is only used to assist the relevant personnel in reading the film and is not used for disease diagnosis.

[0066] In another embodiment of the present application, the medical record of the patient is also obtained, and related keywords of the pulmonary cryptococcosis infection symptoms such as hemoptysis, dyspnea, and shortness of breath are set, which are extracted by comparison with the patient's relevant medical record, and if the keywords coincide, a warning prompt of the pulmonary cryptococcosis infection is provided to the diagnosing doctor.

[0067] In summary, in view of the technical problem that the lesion tissue and the blood vessel tissue of the pulmonary cryptococcosis are difficult to distinguish, the present application provides an auxiliary reading system suitable for patients with pulmonary cryptococcosis. The present application first obtains the continuous CT slice image of the lung, the lung parenchyma region, and extracts the foreground connected domain of the lung parenchyma region as the connected domain to be analyzed; further selects any connected domain to be analyzed as a target connected domain; according to the monotonicity of the width change of the target connected domain, the blood vessel possibility of the target connected domain is obtained; further according to the monotonicity of the area change of the connected domain to be analyzed at the same position of the target connected domain, the through blood vessel possibility of the target connected domain is obtained; the circle-like degree of the target connected domain is obtained; finally, according to the circle-like degree, the blood vessel possibility and the through blood vessel possibility are fused, and the target connected domain is marked.

[0068] It is to be noted that the sequential order of the above-described embodiments of the present application only for the purpose of description, but not the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0069] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments.

Claims

1. An auxiliary reading system for a patient with pulmonary cryptococcosis, characterized by, The system comprises: An image acquisition module: acquiring continuous CT slice images of a current patient's lung; extracting lung parenchyma regions in each CT slice image through image segmentation; extracting foreground connected domains of the lung parenchyma regions as connected domains to be analyzed; and matching the connected domains to be analyzed at the same position in all CT slice images; An image analysis module: selecting any connected domain to be analyzed as a target connected domain; acquiring a blood vessel possibility of the target connected domain according to the monotonicity of the width variation of the target connected domain; acquiring a through blood vessel possibility of the target connected domain according to the monotonicity of the area variation of the connected domains to be analyzed matched by the target connected domain; and acquiring a degree of circularity of the target connected domain; An image labeling module: labeling the target connected domain according to the degree of circularity, the blood vessel possibility and the through blood vessel possibility.

2. The system for assisting in reading a radiograph of a patient suspected of having pulmonary cryptococcosis according to claim 1, wherein, The method for acquiring the blood vessel possibility comprises: Taking a line connecting two most distant pixel points in the target connected domain as a reference line, and moving from one end of the reference line to the other end with a single pixel point as a step; after each movement, acquiring the pixel width of the target connected domain in the vertical direction of the reference line, and constructing a width sequence of the target connected domain; Acquiring the blood vessel possibility of the target connected domain according to the monotonicity of the width sequence.

3. The system for assisting in reading radiographs of patients with pulmonary cryptococcosis according to claim 2, wherein The method for acquiring the blood vessel possibility of the target connected domain according to the monotonicity of the width sequence comprises: Acquiring the blood vessel possibility of the target connected domain according to the correlation between the width sequence and a serial number sequence composed of serial numbers of elements in the width sequence; and the correlation between the width sequence and the serial number sequence is positively correlated with the blood vessel possibility.

4. The system for assisting in reading radiographs of patients with pulmonary cryptococcosis according to claim 1, wherein The method for acquiring the through blood vessel possibility comprises: Acquiring the areas of the connected domains to be analyzed matched by the target connected domain, and sorting the areas according to the acquisition sequence of the CT slice images to obtain an area sequence; Acquiring the through blood vessel possibility of the target connected domain according to the monotonicity of the area sequence.

5. The system for assisting in reading radiographs of patients with pulmonary cryptococcosis according to claim 4, wherein The method for acquiring the through blood vessel possibility of the target connected domain according to the monotonicity of the area sequence comprises: Acquiring the through blood vessel possibility of the target connected domain according to the number of extreme points other than the two ends in the area sequence; and the number of extreme points other than the two ends is negatively correlated with the through blood vessel possibility.

6. The system for assisting in reading radiographs of patients with pulmonary cryptococcosis according to claim 1, wherein The method for labeling the target connected domain comprises: Fusing the degree of circularity and the blood vessel possibility to obtain a first normal possibility; the degree of circularity is negatively correlated with the first normal possibility; and the blood vessel possibility is positively correlated with the first normal possibility; Fusing the degree of circularity and the through blood vessel possibility to obtain a second normal possibility; the degree of circularity and the through blood vessel possibility are positively correlated with the second normal possibility; Fusing the first normal possibility and the second normal possibility to obtain a normal tissue possibility of the target connected domain; the first normal possibility and the second normal possibility are positively correlated with the normal tissue possibility. When the normal tissue possibility is lower than a preset normal threshold, the target connected domain is marked.

7. The system for assisting in reading radiographs of patients with pulmonary cryptococcosis according to claim 1, wherein The circle-like degree is a circularity of an outermost pixel contour of the target connected domain.

8. The system for assisting in reading radiographs of patients with pulmonary cryptococcosis according to claim 1, wherein The method for matching the to-be-analyzed connected domains at the same position in all the CT slice images comprises: When the overlapping ratio of a pixel in the to-be-analyzed connected domain in one image and a pixel in the to-be-analyzed connected domain in another image is greater than a preset threshold, it is determined that the two corresponding to-be-analyzed connected domains are matched.

9. The system for assisting in reading radiographs of patients with pulmonary cryptococcosis according to claim 1, wherein, A lung parenchyma region of the CT slice image is extracted through a pre-trained neural network.

10. The system for assisting in reading a radiograph of a patient suspected of having pulmonary cryptococcosis according to claim 1, wherein, The method for obtaining the foreground connected domain comprises: A region growing algorithm is selected, a seed point is randomly selected from a pixel with a gray value greater than a gray mean value of the lung parenchyma region, and the foreground connected domain is obtained.

Images