Head anomaly detection method and device, equipment and medium

Abstract

The invention discloses a head anomaly detection method, device and equipment and a medium, and the method comprises the steps: carrying out the image processing of a head perfusion map, and obtaining a pixel ratio image; determining a target abnormal region in the head perfusion map based on the head perfusion map and the pixel ratio image; processing the target abnormal region to obtain a focus region; determining abnormal parameter data based on the target abnormal region and the focus region; and obtaining a head anomaly detection result based on the anomaly parameter data.

Description

Technical Field

[0001] The present invention relates to technical fields such as medical image processing, and in particular to a head abnormality detection method, device, equipment and medium. Background Art

[0002] Cerebrovascular disease is a hot topic in global health. The incidence of ischemic stroke is significantly higher than that of hemorrhagic stroke, accounting for 60% to 70% of all strokes. The progression of acute ischemic stroke can be divided into two phases: the pre-infarction phase and the acute infarction phase, based on changes in cerebral blood flow. The condition typically worsens over time. It should be noted that staging based solely on the duration of onset ignores individual differences in collateral circulation, cerebral circulatory reserve, and cerebral metabolic reserve, hindering a comprehensive and accurate assessment of the patient's condition. Clinically, it is generally accepted that CT perfusion imaging should be used to understand the actual condition of the patient's brain tissue, thereby formulating targeted, individualized treatment plans.

[0003] Related technologies assess perfusion abnormalities by analyzing changes in perfusion signals for each brain region. Brain regions are divided, and then abnormality assessment thresholds are set. Qualitative abnormality analysis is performed on each perfusion map for each brain region, and signal changes are annotated. This method provides qualitative results for each brain region, but does not identify the extent of the lesion based on perfusion signal changes. It also fails to visualize and quantitatively analyze the abnormal region within the parameter map. This results in inconsistent assessments by different doctors, influenced by experience and subjective judgment, and in low stability among the same doctor. Summary of the Invention

[0004] The embodiments of the present invention are intended to at least partially address one of the technical problems in the related art. To this end, one object of the present invention is to provide a head abnormality detection method, apparatus, device, and medium for assisting doctors in diagnosis, thereby improving the efficiency and accuracy of doctors' diagnosis.

[0005] An embodiment of the present invention provides a head abnormality detection method, which includes: performing image processing on a head perfusion map to obtain a pixel ratio image; determining a target abnormal area in the head perfusion map based on the head perfusion map and the pixel ratio image; processing the target abnormal area to obtain a lesion area; determining abnormal parameter data based on the target abnormal area and the lesion area; and obtaining a head abnormality detection result based on the abnormal parameter data.

[0006] Exemplarily, image processing is performed on the head perfusion map to obtain a pixel ratio image, including: performing geometric operations on the head perfusion map to obtain a symmetrical head perfusion map; after aligning the head perfusion map and the symmetrical head perfusion map, obtaining a pixel ratio image based on the pixel ratio of the head perfusion map and the symmetrical head perfusion map.

[0007] Exemplarily, determining a target abnormal area in the head perfusion image based on the head perfusion map and the pixel ratio image includes: determining an absolute abnormal area from the head perfusion map based on a pixel absolute threshold, and / or determining a relative abnormal area from the pixel ratio image based on a pixel relative threshold; and determining the target abnormal area in the head perfusion map based on the absolute abnormal area and / or the relative abnormal area.

[0008] Exemplarily, the head perfusion atlas includes N maps, each of the N maps includes a target abnormal area; processing the target abnormal area to obtain a lesion area includes: performing connectivity processing on the N target abnormal areas corresponding to the N map images to obtain the lesion area.

[0009] Exemplarily, the abnormal parameter data includes at least one of the lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio. Based on the target abnormal area and the lesion area, the abnormal parameter data is determined, including at least one of the following: determining the lesion volume based on the number of first target pixel points and the first pixel spacing in the lesion area; determining the pixel mean of the corresponding area in the head perfusion map based on the lesion area; determining the intersection area of ​​the lesion area and the target abnormal area, and determining the number of second pixel points and the second pixel spacing in the intersection area, and obtaining the mismatch volume and mismatch rate based on the lesion volume, the number of second pixel points, and the second pixel spacing; determining the intersection area of ​​the lesion area and the target abnormal area, and determining the pixel mean of the corresponding area from the pixel ratio image based on the intersection area to obtain the perfusion signal change ratio.

[0010] Exemplarily, the abnormal parameter data includes at least one of lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio. Based on the abnormal parameter data, a head abnormality detection result is obtained, including: performing weighted calculation on the lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio to obtain a weighted result; obtaining a perfusion analysis result based on the weighted result and a statistical threshold, wherein the perfusion analysis result includes at least one of insufficient perfusion, collateral circulation formation, blood reperfusion, and excessive perfusion; and obtaining a head abnormality detection result based on the perfusion analysis result.

[0011] Exemplarily, the head abnormality detection method further includes acquiring a head perfusion image sequence; separating cerebrospinal fluid regions based on the head perfusion image sequence to obtain a separated head perfusion image sequence; and performing deconvolution processing on the separated head perfusion image sequence to obtain a head perfusion map.

[0012] Illustratively, deconvolution processing is performed on the separated head perfusion image sequence to obtain a head perfusion atlas, including: obtaining a vascular bolus concentration curve and a tissue bolus concentration curve based on the separated head perfusion image sequence; deconvolution processing is performed on the vascular bolus concentration curve and the tissue bolus concentration curve to obtain a residual function; and obtaining the head perfusion atlas based on the residual function.

[0013] Another embodiment of the present invention provides a head abnormality detection device, which includes: a first acquisition module, used to perform image processing on the head perfusion map to obtain a pixel ratio image; a first determination module, used to determine the target abnormal area in the head perfusion map based on the head perfusion map and the pixel ratio image; a second acquisition module, used to process the target abnormal area to obtain a lesion area; a second determination module, used to determine abnormal parameter data based on the target abnormal area and the lesion area; and a third acquisition module, used to obtain a head abnormality detection result based on the abnormal parameter data.

[0014] An embodiment of the present invention provides an electronic device including a memory and a processor. The memory stores a computer program, and the processor implements the steps of the method of any one of the above embodiments when executing the computer program.

[0015] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method of any one of the above embodiments are implemented.

[0016] In the above-mentioned embodiment, the head abnormality detection method includes: performing image processing on the head perfusion map to obtain a pixel ratio image; determining a target abnormal region in the head perfusion map based on the head perfusion map and the pixel ratio image; processing the target abnormal region to obtain a lesion region; determining abnormal parameter data based on the target abnormal region and the lesion region; and obtaining a head abnormality detection result based on the abnormal parameter data. This method quantitatively analyzes the head perfusion map based on the abnormal parameter data, helping doctors comprehensively assess the patient's disease progression, thereby providing targeted, individualized treatment plans and improving the efficiency and accuracy of doctors' diagnoses.

[0017] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flow chart of a head abnormality detection method provided by an embodiment of the present invention;

[0019] Figure 2 Extraction of lesion area map based on different map target abnormal areas provided by the embodiment of the present invention;

[0020] Figure 3 A schematic diagram of an auxiliary analysis system provided in an embodiment of the present invention;

[0021] Figure 4 A diagram for setting the threshold value for abnormal analysis of perfusion images provided in an embodiment of the present invention;

[0022] Figure 5 A segmentation map of abnormal perfusion areas and lesion areas provided by an embodiment of the present invention;

[0023] Figure 6 The perfusion map based on abnormal areas of different maps provided by the embodiment of the present invention;

[0024] Figure 7 A diagram showing the quantitative analysis results of perfusion parameters in abnormal areas provided by an embodiment of the present invention;

[0025] Figure 8 A detailed flow chart of a head abnormality detection method based on an auxiliary analysis system provided in an embodiment of the present invention;

[0026] Figure 9 This is a block diagram of a head abnormality detection device provided in another embodiment of the present invention. DETAILED DESCRIPTION

[0027] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0028] Cerebrovascular disease is a hot topic in global health. The incidence of ischemic stroke is significantly higher than that of hemorrhagic stroke, accounting for 60% to 70% of all strokes. The progression of acute ischemic stroke can be divided into two phases: the pre-infarction phase and the acute infarction phase, based on changes in cerebral blood flow. The condition typically worsens over time. It should be noted that staging based solely on the duration of onset ignores individual differences in collateral circulation, cerebral circulatory reserve, and cerebral metabolic reserve, hindering a comprehensive and accurate assessment of the patient's condition. Clinically, it is generally accepted that CT perfusion imaging should be used to understand the actual condition of the patient's brain tissue, thereby formulating targeted, individualized treatment plans.

[0029] Currently, the perfusion maps (CBF (Cerebral Blood Flow), CBV (Cerebral Blood Volume), TTP (Time To Peak), MTT (Mean Transit Time)) that doctors can rely on in clinical diagnosis and treatment usually only directly display the various parameter maps, without in-depth exploration of the information contained in each parameter map. The non-quantitative direct display of perfusion maps may lead to poor consistency in the evaluation of different doctors and low stability of the evaluation of the same doctor due to the influence of experience, qualifications and subjective judgment during the diagnosis and treatment process. Therefore, it is possible to visualize the corresponding areas of the perfusion map with different degrees of abnormality under relative and absolute thresholds, and quantitatively analyze and calculate multiple parameters of each abnormal area, which will assist doctors in comprehensively evaluating the patient's disease progression and making efficient and accurate judgments on the patient's cerebral insufficiency, collateral circulation formation, reperfusion and overperfusion.

[0030] The perfusion abnormality analysis of related technologies only provides qualitative results for each brain region, indicating whether there is an abnormal increase or decrease. It does not identify the scope of the lesion based on the change in perfusion signal, nor can it provide a parameter map of the abnormal area itself. Visual display and quantitative results of the abnormal area, so it is impossible to provide perfusion abnormality analysis for specific lesions.

[0031] In view of this, an embodiment of the present application provides a method for detecting head abnormalities, which improves diagnostic efficiency and accuracy by processing and quantitatively analyzing head perfusion maps.

[0032] Figure 1 This is a flow chart of the head abnormality detection method provided in an embodiment of the present invention.

[0033] like Figure 1 As shown, the head abnormality detection method 100 includes steps S110 to S150.

[0034] Step S110 , performing image processing on the head perfusion map to obtain a pixel ratio image.

[0035] Exemplarily, the head perfusion map PerfMap includes, for example, a CT perfusion map, a magnetic resonance perfusion map, and the like. The embodiment of the present application is explained using the CT perfusion map as an example. The CT perfusion map includes multiple parameter maps, such as CBF (Cerebral Blood Flow), CBV (Cerebral Blood Volume), TTP (Time To Peak), MTT (Mean Transit Time), Tmax (Time to Maximum), and the like. Image processing of the head perfusion map includes operations such as translation, rotation, flipping, and registration. For example, a pixel ratio image calculates the ratio of each pixel point of the same parameter map before and after image processing, and uses multiple ratios as new pixel points to obtain a pixel ratio image.

[0036] Step S120 : determining a target abnormal area in the head perfusion map based on the head perfusion map and the pixel ratio image.

[0037] For example, for different tissues and organs, there is usually a certain normal pixel ratio range, i.e., a threshold value. When the pixel ratio of a certain area exceeds this normal range, it may indicate an abnormality, and the target abnormal area can be obtained.

[0038] Step S130: Process the target abnormal area to obtain the lesion area.

[0039] Exemplarily, since the head perfusion map PerfMap includes multiple parameter maps, multiple target abnormal areas can be obtained based on different parameter maps and thresholds. The multiple target abnormal areas are processed, such as by taking a union, connected domain processing, etc., to obtain a merged target abnormal area, that is, a lesion area.

[0040] Step S140: determining abnormal parameter data based on the target abnormal area and the lesion area.

[0041] For example, abnormal parameter data can be calculated for each parameter map based on the target abnormal area and the lesion area. Abnormal parameter data corresponding to each parameter map can be obtained. Abnormal parameter data can include custom parameters such as lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio.

[0042] Step S150: Obtaining a head abnormality detection result based on the abnormal parameter data.

[0043] For example, the abnormal head detection results include insufficient perfusion, collateral circulation formation, blood reperfusion, excessive perfusion, etc. The abnormal parameter data corresponding to each parameter map is processed, and the changes in each parameter map and the perfusion type to which it belongs can be judged based on the empirical threshold to obtain the abnormal head detection results.

[0044] According to the embodiments of the present application, quantitative analysis of the head perfusion map based on abnormal parameter data helps doctors to comprehensively evaluate the patient's disease progression, thereby providing targeted individualized treatment plans and improving the efficiency and accuracy of doctors' diagnosis.

[0045] Perform image processing on the head perfusion map PerfMap to obtain a pixel ratio image, including: performing geometric operations on the head perfusion map PerfMap to obtain a symmetrical head perfusion map PerfMap f ; For head perfusion map PerfMap and symmetrical head perfusion map PerfMap f After registration, based on the head perfusion map PerfMap and the symmetrical head perfusion map PerfMap f The pixel ratio of is obtained to obtain a pixel ratio image.

[0046] For example, the geometric operations include translation, rotation, flipping and the like, and the registration includes non-rigid registration, so that the head perfusion map PerfMap and the symmetrical head perfusion map PerfMap f The shape and structure of the head are accurately aligned, based on the accurately aligned head perfusion map PerfMap and the symmetrical head perfusion map PerfMap f The pixel ratio of the corresponding pixel points is obtained to obtain the pixel ratio image Radio map .

[0047] For example, based on the head perfusion map PerfMap, the geometric center of the head perfusion map PerfMap is calculated, the geometric center of the head perfusion map PerfMap is translated to the center of the image (for example, the image is a square, and the center of the image is the center of the square), the positions of the superior and inferior sagittal sinus corner points of the head perfusion map PerfMap are identified, and the upper and lower corner points are connected to form a symmetry axis L s Based on the symmetry axis L s , calculate the angle theta of the head perfusion map PerfMap to be adjusted (i.e. the angle between the symmetry axis and the vertical direction of the image), and use the image center as the torque center to rotate the perfusion map by the angle theta; based on the rotated perfusion map, flip it symmetrically along the center line to obtain a symmetrical perfusion map PerfMap fBased on the two sets of atlases after registration (one set includes multiple registered perfusion parameter maps, and the other set includes multiple registered symmetrical perfusion parameter maps), considering the structural symmetry of the two hemispheres of the brain, the ratio map of the pixel values ​​corresponding to each point in each parameter map is obtained. map (i.e., pixel ratio image) to represent the relative abnormality of the current perfusion parameter map at each pixel.

[0048] Based on the head perfusion map and the pixel ratio image, a target abnormal area in the head perfusion map is determined, including: determining the absolute abnormal area from the head perfusion map based on the pixel absolute threshold, and / or determining the relative abnormal area from the pixel ratio image based on the pixel relative threshold; determining the target abnormal area in the head perfusion map based on the absolute abnormal area and / or the relative abnormal area.

[0049] For example, the pixel absolute threshold is obtained by statistical experience, and the pixel relative threshold is a critical value for identifying abnormal areas that are significantly different from surrounding areas or normal states through relative relationships.

[0050] For example, based on the absolute threshold, the absolute abnormal area is calculated and displayed using the threshold method. In one example, the absolute abnormal area can be used as the target abnormal area. Set the relative threshold of the abnormal area of ​​interest and perform the pixel ratio image Radio map The relative abnormal area Mask is calculated using threshold segmentation and morphological analysis map In another example, the relative abnormal area Mask map As the target abnormal area. In another example, the relative abnormal area and the absolute abnormal area can be taken as the union, so that the connected domains with intersecting positions are merged into the same connected domain to obtain the Mask all , Mask all That is, the target abnormal area.

[0051] In the above embodiment, the calculation of abnormal areas based on the set absolute thresholds and relative thresholds facilitates the assessment of brain tissue perfusion status, collateral circulation establishment, reperfusion and overperfusion at the lesion site, and allows for rapid formulation of individualized and precise diagnosis and treatment plans.

[0052] The head perfusion map PerfMap includes N maps, where N is a positive integer. Each map in the N maps includes the target abnormal area Mask. all ; Process the target abnormal area to obtain the lesion area Lesion n , including: Connectivity processing of N target abnormal regions corresponding to N maps to obtain the lesion region Lesion n .

[0053] For example, the map is the parameter map mentioned above. The head perfusion map PerfMap includes N parameter maps, such as CBF (Cerebral Blood Flow), CBV (Cerebral Blood Volume), TTP (Time To Peak), MTT (Mean Transit Time), Tmax (Time to Maximum), etc. This application uses 5 parameter maps as an example for explanation, that is, N=5, and the head perfusion map PerfMap (map=CBF, CBV, MTT, TTP, Tmax) is not specifically limited. The first map is related to the lesion area Lesion n The corresponding area is represented as Lesion 1-n , Lesion area 1-n Here, 1 represents the first map, and n represents the nth lesion area in the first map.

[0054] Common algorithms for connectivity processing include region growing algorithms: first, select a suitable seed point (usually a pixel within the target area), and then, based on pre-set growth criteria (such as similarity of pixel grayscale values, texture features, etc.), gradually merge the pixels that are connected to the seed point and meet the growth criteria into the current area to form a connected target area. Graph theory-based methods: The medical image is regarded as a graph, with pixels or regions as nodes of the graph, and the connection relationship and similarity between nodes as the weight of the edge. Use graph theory algorithms (such as minimum spanning tree, graph cut, etc.) to find the optimal segmentation scheme and determine the connected area. Deep learning algorithms: Use network structures such as convolutional neural networks (CNN), fully convolutional networks (FCN), and U-Net to directly learn image features and connectivity information from medical image data to achieve automatic segmentation of medical images and identification of connected areas.

[0055] Figure 2 The embodiment of the present invention provides a lesion area map based on different map target abnormal areas.

[0056] For example, Figure 2 As shown in Figure 1, the head perfusion map PerfMap includes 5 parameter maps, namely map = CBF, CBV, MTT, TTP, Tmax, Figure 2 30%, 35%, 40%, 6s, and 40% are the abnormality analysis thresholds set for each parameter graph (relative threshold and / or absolute threshold, where 6s is the absolute threshold and 30%, 35%, 40%, and 40% are relative thresholds). Figure 2Two lesion areas Lesion1 and Lesion2 are shown in the figure, and the target abnormal area Mask of the 5 parameter maps is all Merge the connected domain to get the Label n , mark it as the lesion area Lesion n Each parameter map can include multiple target abnormal area masks all ,for example Figure 2 It shows that each parameter map includes two target abnormal regions, the first target abnormal region is located in the green dotted box, and the second target abnormal region is located in the red dotted box.

[0057] The abnormal parameter data includes at least one of lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio. The abnormal parameter data is determined based on the target abnormal area and the lesion area, including at least one of the following:

[0058] determining a lesion volume based on the number of first target pixels and the first pixel spacing in the lesion area;

[0059] Based on the lesion area, the pixel mean of the corresponding area in the head perfusion map is determined;

[0060] Determining an intersection area between the lesion area and the target abnormal area, and determining the number of second pixels and the second pixel spacing in the intersection area, and obtaining a mismatch volume and a mismatch rate based on the lesion volume, the number of second pixels, and the second pixel spacing;

[0061] The intersection area of ​​the lesion area and the target abnormal area is determined, and the pixel mean of the corresponding area is determined from the pixel ratio image based on the intersection area to obtain the perfusion signal change ratio.

[0062] Specifically, each parameter map of the head perfusion map (map: CBF, CBV, MTT, TTP, Tmax) is analyzed for each lesion. n The quantitative values ​​of the perfusion parameters were analyzed and the implementation process was as follows:

[0063] a: Calculate the lesion volume V of each lesion area Lesionn Ln , the specific nth lesion area Lesion n Lesion volume V Ln As shown in formula (1):

[0064] V Ln =Sum(Lesion n )×spacing x ×spacing y ×spacing z / 1000(1)

[0065] Among them, Sum is the nth lesion area Lesion n The number of pixels with a pixel value of 1 (the number of the first target pixel), spacing x 、spacing y 、spacing z is the pixel spacing of the image in the x, y, and z directions (the first pixel spacing), where 1000 represents the unit conversion factor, which is used to convert spacing x 、spacing y 、spacing z The unit of the multiplication result, such as cubic millimeters, is converted to cubic centimeters.

[0066] b: For each perfusion parameter map, calculate the perfusion parameter map in the nth lesion area Lesion one by one n The mean value of the map signal within the range is denoted as M map-n .

[0067] c: For each perfusion parameter map, calculate the mismatch volume MV (mismatch volume) and mismatch ratio MR (mismatch ratio) for each lesion area Lesion, specifically:

[0068] Get the target abnormal area Mask in the current perfusion parameter map map Lesion n The intersection of map-n , the mismatch volume MV of the nth lesion area Lesion in the current perfusion parameter map map-n As shown in formula (2):

[0069] MV map-n =V Ln -Sum(Lesion map-n )×spacing x ×spacing y ×spacing z / 1000(2)

[0070] 1000 represents the unit conversion factor, which is used to convert spacing x 、spacing y 、spacing z The unit of the multiplication result, such as cubic millimeters, is converted to cubic centimeters.

[0071] The mismatch rate MR of the nth Lesion in the current map map-n As shown in formula (3):

[0072] MV map-n =Sum(Lesion map-n )×spacing x ×spacing y ×spacing z / 1000 / V Ln (3)

[0073] Among them, Sum is the nth lesion area Lesion in the current map n The number of pixels with a pixel value of 1 (the number of the second target pixel points), spacing x 、spacing y 、spacing z is the pixel spacing of the image in the x, y, and z directions (second pixel spacing), where 1000 represents the unit conversion factor, which is used to convert spacing x 、spacing y 、spacing z The unit of the multiplication result, such as cubic millimeters, is converted to cubic centimeters.

[0074] d: For each perfusion parameter map, calculate the perfusion signal change ratio P for each lesion region Lesion. Specifically, the perfusion signal change ratio P of the nth lesion region Lesion in the pixel ratio image corresponding to the current perfusion parameter map map-n As shown in formula (4):

[0075] P map-n =mean[Ratio map (Lesion map-n )]×100%(4)

[0076] Among them, Mean[] is the pixel mean calculation function; Ratio map (Lesion map-n ) is the pixel ratio image corresponding to the current perfusion parameter map Ratio map Mask of abnormal target area map Lesion in the lesion area map-n The pixels within the target abnormal area are the pixels within the intersection area of ​​the lesion area and the target abnormal area.

[0077] The abnormal parameter data includes at least one of the lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio. Based on the abnormal parameter data, a head abnormality detection result is obtained, including: performing weighted calculation on the lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio to obtain a weighted result; based on the weighted result and the statistical threshold, a perfusion analysis result is obtained, wherein the perfusion analysis result includes at least one of insufficient perfusion, collateral circulation formation, blood reperfusion, and excessive perfusion; based on the perfusion analysis result, a head abnormality detection result is obtained.

[0078] For example, based on the combined weighting of custom parameters such as lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio, the abnormality of the perfusion map is analyzed according to the empirical threshold or statistical threshold. Referring to Table 1, the severity of the perfusion abnormality of the lesion can be classified.

[0079] Table 1 Perfusion type judgment basis

[0080] Perfusion type TTP MTT CBF CBV Insufficient perfusion Significantly prolonged Significantly prolonged decline decline Collateral circulation formation extend extend Normal or slightly increased Normal or slightly increased Blood reperfusion Normal or shortened Normal or shortened Normal or slightly increased Increase Overperfusion Normal or shortened Normal or shortened Significant increase Significant increase

[0081] When all parameter maps of the head perfusion map PerfMap meet the above conditions at the same time, the severity of the perfusion abnormality of the lesion can be classified to obtain the head abnormality detection result.

[0082] In the above embodiment, by counting the number of lesions and analyzing the abnormal increase or decrease of TTP, CBF, CBV, and MTT signals for each lesion, the lesion volume, the percentage of increase or decrease, the mismatch volume MV (mismatch volume) and the mismatch ratio MR (mismatch ratio), the perfusion change trend, the abnormality classification, etc. are output. The operation of using a single lesion as the calculation object can solve the problem that the abnormality analysis based on the whole brain or brain region is not detailed and accurate enough, and can truly reflect the severity of any single lesion when multiple lesions are present.

[0083] The head abnormality detection method also includes obtaining a head perfusion image sequence; separating cerebrospinal fluid areas based on the head perfusion image sequence to obtain a separated head perfusion image sequence; and performing deconvolution processing on the separated head perfusion image sequence to obtain a head perfusion map PerfMap.

[0084] For example, a contrast agent is rapidly injected intravenously (usually 30-50 ml, the injection speed depends on the patient's condition and equipment requirements), and a CT scan is started at the same time. Continuous dynamic scanning is performed within a short period of time after the contrast agent injection (usually 40-60 seconds) to obtain a series of brain images at different time points, namely the head perfusion image sequence.

[0085] Based on the head perfusion image sequence, the cerebrospinal fluid mask M is extracted using the threshold method. f , where the pixels occupied by tissue within the cerebrospinal fluid are marked as 1, and the rest are marked as 0 to avoid the influence of cerebrospinal fluid on abnormal areas.

[0086] The cerebrospinal fluid mask is essentially a binary image or a data matrix in which the area corresponding to cerebrospinal fluid is marked with a specific value (usually 1 or other non-zero value), while other non-cerebrospinal fluid areas are marked with 0 or other specific background values. The cerebrospinal fluid mask is like a "template" or "mask" that can be overlaid on the original medical image (such as MRI and CT images) to separate the cerebrospinal fluid area from the entire image for subsequent analysis and processing.

[0087] The separated head perfusion image sequence is subjected to deconvolution processing to obtain a head perfusion atlas, including: obtaining a vascular bolus concentration curve and a tissue bolus concentration curve based on the separated head perfusion image sequence; performing deconvolution processing on the vascular bolus concentration curve and the tissue bolus concentration curve to obtain a residual function; and obtaining the head perfusion atlas based on the residual function.

[0088] For example, the separated head perfusion image sequence is segmented using a threshold method to calculate the AIF curve (Arterial Input Function), which can obtain the vascular bolus concentration curve (Ca); based on the multi-phase images of the perfusion data, a set of brain tissue TDC curves (Time-Density Curve, i.e., the tissue bolus concentration curve (Ct) is obtained.

[0089] AIF curve: The arterial input function describes the time-dependent changes in contrast agent concentration in arterial blood. Accurately acquiring the AIF curve is crucial in perfusion imaging (such as CT perfusion imaging and magnetic resonance perfusion imaging), as it forms the basis for calculating other perfusion parameters, such as cerebral blood flow and cerebral blood volume. The AIF curve is constructed by measuring the contrast agent concentration over time in a region of interest (ROI) within an artery (usually a large vessel such as the carotid artery).

[0090] TDC curves: In CT perfusion imaging, these are often called time-density curves. They depict the temporal evolution of contrast agent concentration within a specific region of interest (e.g., a tissue region or tumor area within the brain). These curves are plotted with density (CT value) as the vertical axis and time as the horizontal axis. Analysis of TDC curves can provide information on tissue perfusion status, aiding in disease diagnosis and assessment.

[0091] Based on the AIF and TDC curves, the deconvolution method is used to obtain the residual function r(t), as shown in formula (5):

[0092]

[0093] Based on the residual function r(t), the head perfusion map PerfMap is calculated, which includes: the residual function is used to measure the difference between the model prediction value and the actual measurement value. In perfusion analysis, common models are based on the dynamics of contrast agents, such as the two-compartment model and the single-compartment model. In order to obtain the perfusion map, it is necessary to estimate the parameters in the model, such as cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT), etc. These parameters are related to the dynamic process of the contrast agent in the tissue. Use optimization algorithms (such as the least squares method, maximum likelihood estimation method, etc.) to adjust the model parameters to minimize the value of the residual function. In the process of adjusting the parameters, the contrast agent concentration predicted by the model is continuously updated until the residual function reaches a minimum. At this time, the parameter value obtained is the optimal estimate.

[0094] For each region of interest, parameter estimation is performed to obtain the perfusion parameter values ​​(such as CBF and CBV) for that region. The perfusion parameter values ​​for all regions of interest in the entire image are mapped to obtain a distribution map of the perfusion parameters, known as a perfusion atlas. For example, the CBF value of each pixel is mapped to a color or grayscale range to generate a cerebral blood flow perfusion atlas. Differences in color or grayscale reflect the level of cerebral blood flow in different regions.

[0095] In order to avoid the influence of noise on the results, the perfusion map was filtered by mean filtering to obtain the denoised perfusion map PerfMap (CBF, CBV, MTT, TTP, Tmax). The perfusion parameter values ​​are calculated as shown in formula (6), formula (7), formula (8), formula (9), and formula (10):

[0096]

[0097]

[0098] Among them, k AV represents the arteriovenous correction constant, H SV Correction constant for different hematocrit levels in large vessels, H SV A correction constant representing different capillary hematocrit levels, ρ: represents brain tissue density. For example, based on previous experience, set HSV = 0.25, HLV = 0.45, and ρ = 1.04 g / ml.

[0099] Based on the above calculation and analysis process, a fully automatic auxiliary analysis system is established.

[0100] Figure 3 A schematic diagram of an auxiliary analysis system provided by an embodiment of the present invention, such as Figure 3 As shown, the auxiliary analysis system includes an image acquisition module, a threshold adjustment module, an abnormal area calculation and display module, and a quantitative analysis module.

[0101] The image acquisition module is used to establish an image acquisition module to obtain head perfusion images; the threshold adjustment module: users can manually set the relative threshold and absolute threshold through the threshold adjustment module, such as Figure 4 As shown in the figure, the abnormal analysis threshold of each perfusion parameter map is set (the threshold can be adjusted according to the needs). Abnormal area calculation and display module: The perfusion image and the preset threshold are input into the abnormal area calculation and display module, and the segmentation results of the abnormal perfusion area and lesion area of ​​each parameter map under the preset threshold are automatically calculated, as shown in the figure below. Figure 5 As shown: Quantitative analysis calculation module: According to the above perfusion map abnormal area segmentation results and lesion area, the perfusion parameters are quantitatively analyzed, such as Figure 6 、 Figure 7 As shown in the figure, for each extracted lesion area, the quantitative analysis results include:

[0102] ① The brain region where the lesion is located;

[0103] ② Lesion volume;

[0104] ③ Mismatch Volume / Ratio: Mismatch volume and Mismatch Ratio of abnormal CBF, CBV, MTT, TTP, Tmax and lesion volume in each lesion;

[0105] ④. Analysis of perfusion change trends: The trend and percentage of abnormal increase or decrease of CBF, CBV, MTT, and TTP in each lesion, and the classification of perfusion abnormalities are given.

[0106] In the above embodiment, by establishing a system for auxiliary analysis of perfusion abnormality, parameter adjustment and automatic display of perfusion abnormality quantitative analysis results can be achieved through human-computer interaction.

[0107] Figure 8 A detailed flow chart of a head abnormality detection method based on an auxiliary analysis system provided in an embodiment of the present invention.

[0108] like Figure 8 As shown, the detailed head abnormality detection method 800 based on the auxiliary analysis system includes steps S810 to S860.

[0109] Step S810: Acquire a head perfusion image sequence.

[0110] Step S820: Calculate the perfusion map based on the original image sequence.

[0111] Exemplarily, the original image sequence is the head perfusion image sequence acquired in step S810 .

[0112] Step S830: Calculate and display abnormal areas according to threshold settings. The threshold setting methods include absolute thresholds and relative thresholds.

[0113] Step S840: Calculate the lesion area based on the abnormal area.

[0114] Step S850: Analyze and calculate the quantitative value of the perfusion parameter according to each lesion.

[0115] Step S860: Establish an auxiliary analysis system.

[0116] This embodiment provides a detailed head abnormality detection method based on the auxiliary analysis system, which includes: obtaining a head perfusion imaging sequence; calculating the cerebrospinal fluid mask M of the head image; f and the head symmetry axis L s Calculate the perfusion map PerfMap (CBF, CBV, MTT, TTP, Tmax) of the head image; Based on the perfusion map PerfMap, set the threshold to calculate the abnormal area, the threshold setting method includes absolute threshold and relative threshold; Based on the absolute threshold, use the threshold method to calculate the absolute abnormal area; Based on the relative threshold, use the brain symmetry and threshold method to calculate the relative abnormal area; Based on the above five perfusion maps, use the abnormal area range Mask map (map = CBF, CBV, MTT, TTP, Tmax), calculate the lesion area Lesion using the connected domain union method n (n=1, 2, 3...); for each lesion n , analyze and calculate the quantitative values ​​of perfusion parameters, including: volume, mean, mismatch information, and percentage of signal change; based on the abnormal area display and analysis implementation method, a fully automatic system can be established to assist doctors in evaluating patients' cerebral perfusion conditions.

[0117] Figure 9 This is a block diagram of a head abnormality detection device provided in another embodiment of the present invention.

[0118] The embodiment of the present invention provides a head abnormality detection device 900, please refer to Figure 9 The head abnormality detection device 900 includes: a first obtaining module 910, a first determining module 920, a second obtaining module 930, a second determining module 940, and a third obtaining module 950.

[0119] Illustratively, the first obtaining module 910 is configured to perform image processing on the head perfusion map to obtain a pixel ratio image.

[0120] Illustratively, the first determining module 920 is configured to determine a target abnormal region in the head perfusion map based on the head perfusion map and the pixel ratio image.

[0121] Illustratively, the second obtaining module 930 is used to process the target abnormal region to obtain a lesion region.

[0122] Exemplarily, the second determination module 940 is configured to determine abnormal parameter data based on the target abnormal area and the lesion area.

[0123] Exemplarily, the third obtaining module 950 is used to obtain a head abnormality detection result based on the abnormal parameter data.

[0124] Exemplarily, the first acquisition module 910 is further configured to perform image processing on the head perfusion map to obtain a pixel ratio image, including: performing geometric operations on the head perfusion map to obtain a symmetrical head perfusion map; and aligning the head perfusion map with the symmetrical head perfusion map, and then obtaining a pixel ratio image based on the pixel ratio of the head perfusion map and the symmetrical head perfusion map.

[0125] Exemplarily, the first determination module 920 is further configured to determine a target abnormal area in the head perfusion map based on the head perfusion map and the pixel ratio image, including: determining an absolute abnormal area from the head perfusion map based on a pixel absolute threshold, and / or determining a relative abnormal area from the pixel ratio image based on a pixel relative threshold; and determining a target abnormal area in the head perfusion map based on the absolute abnormal area and / or the relative abnormal area.

[0126] Exemplarily, the second obtaining module 930 is further configured to perform connectivity processing on the N target abnormal regions corresponding to the N maps to obtain lesion regions.

[0127] Exemplarily, the second determination module 940 is also used to determine the lesion volume based on the number of first target pixel points and the first pixel spacing in the lesion area; determine the pixel mean of the corresponding area in the head perfusion map based on the lesion area; determine the intersection area of ​​the lesion area and the target abnormal area, and determine the number of second pixel points and the second pixel spacing in the intersection area, and obtain the mismatch volume and mismatch rate based on the lesion volume, the number of second pixel points, and the second pixel spacing; determine the intersection area of ​​the lesion area and the target abnormal area, determine the pixel mean of the corresponding area from the pixel ratio image based on the intersection area, and obtain the perfusion signal change ratio.

[0128] Exemplarily, the third acquisition module 950 is also used to perform weighted calculations on the lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio to obtain weighted results; based on the weighted results and statistical thresholds, perfusion analysis results are obtained, wherein the perfusion analysis results include at least one of insufficient perfusion, collateral circulation formation, blood reperfusion, and excessive perfusion; based on the perfusion analysis results, head abnormality detection results are obtained.

[0129] Exemplarily, the head abnormality detection method further includes acquiring a head perfusion image sequence; separating cerebrospinal fluid regions based on the head perfusion image sequence to obtain a separated head perfusion image sequence; and performing deconvolution processing on the separated head perfusion image sequence to obtain a head perfusion map.

[0130] Illustratively, deconvolution processing is performed on the separated head perfusion image sequence to obtain a head perfusion atlas, including: obtaining a vascular bolus concentration curve and a tissue bolus concentration curve based on the separated head perfusion image sequence; deconvolution processing is performed on the vascular bolus concentration curve and the tissue bolus concentration curve to obtain a residual function; and obtaining the head perfusion atlas based on the residual function.

[0131] It can be understood that for the specific description of the head abnormality detection device 900, please refer to the description of the head abnormality detection method above, which will not be repeated here.

[0132] An embodiment of the present invention provides an electronic device including a memory and a processor. The memory stores a computer program, and the processor implements the steps of the method of any one of the above embodiments when executing the computer program.

[0133] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method of any one of the above embodiments are implemented.

[0134] This application utilizes abnormal regions analyzed from head perfusion maps and performs visualization and quantitative analysis, enabling more comprehensive utilization and detailed analysis of head perfusion maps. For example, multi-level abnormal region visualization based on relative thresholds of head perfusion maps can calculate patient-specific thresholds to segment and display the location and size of abnormal regions in patients with unilateral lesions, as well as mismatches between abnormal regions of different levels within the same map and between different maps. Multi-level abnormal region visualization based on absolute thresholds of head perfusion maps can accurately display abnormal regions in patients with bilateral lesions. The complementary nature of relative and absolute thresholds makes the display of abnormal regions more comprehensive and accurate, expanding the applicable population and reducing the probability of misdiagnosis in clinical applications. For another example, abnormal region analysis and calculation primarily focuses on individual lesions, calculating the average increase or decrease in perfusion parameters within a single lesion and the mismatch between different perfusion parameters within a single lesion. This calculation based on individual lesions addresses the lack of detailed and accurate abnormality analysis based on the entire brain or brain region, and can accurately reflect the severity of any single lesion in the presence of multiple lesions. The implementation of the above functions can speed up doctors' diagnosis and treatment while ensuring the precision and accuracy of diagnosis, and is more conducive to providing targeted individualized treatment plans for each patient.

[0135] It should be noted that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device), or in conjunction with such instruction execution system, apparatus, or device. For purposes of the present invention, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by an instruction execution system, apparatus, or device, or in conjunction with such instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wires (electronic device), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0136] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0137] In the description of the present invention, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In the present invention, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0138] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like to indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0139] In addition, the terms "first" and "second" used in the embodiments of the present invention are only used for descriptive purposes and should not be understood as indicating or implying relative importance, or implicitly indicating the number of technical features indicated in this embodiment. Therefore, the features defined by the terms "first" and "second" in the embodiments of the present invention can explicitly or implicitly indicate that the embodiment includes at least one of such features. In the description of the present invention, the word "plurality" means at least two or two or more, such as two, three, four, etc., unless otherwise clearly and specifically defined in the embodiments.

[0140] In the present invention, unless otherwise clearly specified or limited in the embodiments, the terms "installed," "connected," "connect," and "fixed" appearing in the embodiments should be understood in a broad sense. For example, the connection may be a fixed connection, a detachable connection, or an integral connection. It can also be a mechanical connection, an electrical connection, etc.; of course, it can also be a direct connection, or an indirect connection through an intermediate medium, or it can be internal communication between two elements, or an interaction between two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood based on the specific implementation.

[0141] In the present invention, unless otherwise expressly specified or limited, when a first feature is "above" or "below" a second feature, it may mean that the first and second features are in direct contact, or that the first and second features are in indirect contact through an intermediary. Furthermore, when a first feature is "above," "above," or "above" a second feature, it may mean that the first feature is directly above or diagonally above the second feature, or simply means that the first feature is at a higher level than the second feature. When a first feature is "below," "below," or "below" a second feature, it may mean that the first feature is directly below or diagonally below the second feature, or simply means that the first feature is at a lower level than the second feature.

[0142] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A head abnormality detection method, characterized in that: The method comprises: Perform image processing on the head perfusion map to obtain a pixel ratio image; determining a target abnormal area in the head perfusion map based on the head perfusion map and the pixel ratio image; Processing the target abnormal area to obtain a lesion area; Determining abnormal parameter data based on the target abnormal area and the lesion area; Based on the abnormal parameter data, a head abnormality detection result is obtained.

2. The method according to claim 1, characterized in that The step of performing image processing on the head perfusion atlas to obtain a pixel ratio image includes: performing geometric operations on the head perfusion map to obtain a symmetrical head perfusion map; After registering the head perfusion map and the symmetrical head perfusion map, the pixel ratio image is obtained based on the pixel ratio of the head perfusion map and the symmetrical head perfusion map.

3. The method according to claim 1, characterized in that The step of determining a target abnormal area in the head perfusion map based on the head perfusion map and the pixel ratio image includes: determining an absolute abnormal region from the head perfusion map based on a pixel absolute threshold, and / or determining a relative abnormal region from the pixel ratio image based on a pixel relative threshold; Based on the absolute abnormal area and / or the relative abnormal area, a target abnormal area in the head perfusion map is determined.

4. The method according to claim 1, wherein The head perfusion atlas includes N map images, and each map image in the N map images includes the target abnormal area; The processing of the target abnormal area to obtain the lesion area includes: Connectivity processing is performed on the N target abnormal regions corresponding to the N maps to obtain the lesion region.

5. The method according to any one of claims 1 to 4, characterized in that The abnormal parameter data includes at least one of a lesion volume, a pixel mean, a mismatch volume, a mismatch rate, and a perfusion signal change ratio. The abnormal parameter data is determined based on the target abnormal area and the lesion area, including at least one of the following: determining a lesion volume based on the number of first target pixels and a first pixel spacing in the lesion area; Based on the lesion area, determining the pixel mean of the corresponding area in the head perfusion map; Determining an intersection area between the lesion area and the target abnormal area, and determining the number of second pixels and the second pixel spacing in the intersection area, and obtaining a mismatch volume and a mismatch rate based on the lesion volume, the number of second pixels, and the second pixel spacing; An intersection area between the lesion area and the target abnormal area is determined, and pixel means of corresponding areas are determined from the pixel ratio image based on the intersection area to obtain a perfusion signal change ratio.

6. The method according to claim 1, characterized in that The abnormal parameter data includes at least one of lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio. The head abnormality detection result is obtained based on the abnormal parameter data, including: Performing weighted calculation on the lesion volume, pixel mean, mismatch volume, mismatch rate, and perfusion signal change ratio to obtain a weighted result; Obtaining a perfusion analysis result based on the weighted result and a statistical threshold, wherein the perfusion analysis result includes at least one of insufficient perfusion, collateral circulation formation, blood reperfusion, and excessive perfusion; The head abnormality detection result is obtained based on the perfusion analysis result.

7. The method according to any one of claims 1 to 4 and 6, characterized in that: The method further includes Acquire head perfusion imaging sequences; separating the cerebrospinal fluid region based on the head perfusion image sequence to obtain a separated head perfusion image sequence; The separated head perfusion image sequence is subjected to deconvolution processing to obtain the head perfusion atlas.

8. The method according to claim 7, characterized in that The deconvolution processing is performed on the separated head perfusion image sequence to obtain the head perfusion atlas, including: Based on the separated head perfusion image sequence, the vascular bolus concentration curve and tissue bolus concentration curve are obtained; performing deconvolution processing on the vascular bolus concentration curve and the tissue bolus concentration curve to obtain a residual function; The head perfusion map is obtained based on the residual function.

9. A head abnormality detection device, characterized in that: The device comprises: The first acquisition module is used to perform image processing on the head perfusion map to obtain a pixel ratio image; a first determining module, configured to determine a target abnormal area in the head perfusion map based on the head perfusion map and the pixel ratio image; A second obtaining module is used to process the target abnormal area to obtain a lesion area; A second determining module is configured to determine abnormal parameter data based on the target abnormal area and the lesion area; The third obtaining module is used to obtain a head abnormality detection result based on the abnormal parameter data.

10. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

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

Images