A method for early abnormality screening of alzheimer's disease brain images

Abstract

This invention relates to the field of healthcare information technology, specifically to a method for early screening of abnormalities in brain images of Alzheimer's disease. The method acquires multi-temporal MRI images and synchronized blood pressure data to determine the edge pixels of the hippocampus and their offset relative to the initial time point; it divides the body and tail regions based on morphological analysis and calculates the atrophy rate by comparing the number of edge pixels in both regions; it identifies the time period affected by blood pressure based on blood pressure data and determines the intensity of blood pressure interference based on the atrophy rate and blood pressure trend within that time period; accordingly, it filters and removes abnormal interference moments significantly affected by hypertension, retaining the remaining sampling moments as valid analysis moments, thereby eliminating the influence of vascular confounding factors and improving the accuracy of identifying specific atrophy.

Description

Technical Field

[0001] This invention relates to the field of healthcare information technology, specifically to a method for early screening of abnormalities in brain images of Alzheimer's disease. Background Technology

[0002] Alzheimer's disease (AD) is a neurodegenerative disease characterized primarily by progressive cognitive decline, one of its early pathological features being significant hippocampal atrophy. Currently, clinical practice often uses brain imaging techniques (such as MRI) to longitudinally monitor changes in hippocampal volume or morphology as an important basis for early AD screening. Conventional methods typically calculate the rate of change in overall hippocampal volume over time or compare the volume differences between patients and healthy individuals to determine the presence of abnormal atrophy.

[0003] However, in actual clinical settings, especially among elderly patients, many also suffer from chronic vascular diseases such as hypertension and diabetes. Long-term hypertension can lead to cerebral small vessel disease, blood-brain barrier damage, and chronic cerebral perfusion abnormalities, subsequently causing non-specific atrophy of limbic structures such as the hippocampus. This type of atrophy manifests on imaging as a reduction in hippocampal volume, but its mechanism is fundamentally different from the neuronal degeneration in Alzheimer's disease (AD).

[0004] Current screening methods generally rely solely on the absolute degree or overall rate of hippocampal atrophy as criteria, which is easily affected by vascular confounding factors, leading to distortion in deformation measurements. They cannot effectively decouple the influence of neurodegenerative atrophy from vascular confounding factors, resulting in insufficient specificity and reliability in identifying abnormal atrophy. Summary of the Invention

[0005] To address the technical problem in related technologies where the influence of neurodegenerative atrophy and vascular confounding factors is not effectively decoupled, resulting in insufficient specificity and reliability in identifying abnormal atrophy, this invention provides an early screening method for abnormalities in Alzheimer's disease brain imaging. The specific technical solution adopted is as follows: This invention proposes a method for early screening of abnormalities in brain images of Alzheimer's disease, the method comprising: MRI images of the hippocampus at different sampling times and blood pressure data were acquired to determine the edge pixels of the hippocampus in the MRI images; Morphological analysis was performed on the edge pixels in the MRI images at each sampling time to determine the body and tail regions of the hippocampus; the specific deformation index at each sampling time was determined by comparing the number of edge pixels in the tail and body regions at different sampling times. The time period of blood pressure influence is determined based on blood pressure data; the intensity of blood pressure interference on atrophy is determined based on the specific deformation index and blood pressure data trend at the sampling time within the blood pressure influence time period. Based on the intensity of blood pressure interference, abnormal interference moments are screened out from the time period of blood pressure influence, and all other sampling moments that exclude abnormal interference moments are taken as valid analysis moments.

[0006] Further, determining the edge pixels of the hippocampus in the MRI image includes: Edge pixels are determined based on the Canny edge detection algorithm.

[0007] Furthermore, the morphological analysis of edge pixels in the MRI image at each sampling time to determine the body and tail regions of the hippocampus includes: Based on principal component analysis, the longest major axis of the region enclosed by the edge pixels is determined; the edge pixels are projected along the determined major axis curve, and the region is divided into three equal parts based on the cumulative distribution of the projection length. The middle part of the division is taken as the body region. Compare the number of pixels in the two sides and designate the area with fewer pixels as the tail area.

[0008] Furthermore, determining the specific deformation index at each sampling time based on a comparison of the number of edge pixels in the tail region and the body region at different sampling times includes: Calculate the ratio of the number of edge pixels in the tail region to the number of edge pixels in the body region at each sampling time, and use it as the volume ratio; A two-dimensional rectangular coordinate system for volume ratio is constructed with the sampling time as the x-axis and the volume ratio value as the y-axis, and the volume ratio coordinate point at each sampling time is determined. The volume ratio fitting curve is obtained by curve fitting the volume ratio coordinate points based on the least squares method. The instantaneous slope of the volume ratio fitting curve at different sampling times was determined and normalized to serve as a specific deformation index.

[0009] Furthermore, the determination of the time period of blood pressure influence based on blood pressure data includes: The sampling time when the blood pressure data is greater than the preset standard blood pressure is taken as the time of influence; The time period consisting of at least a preset number of consecutive influencing moments is defined as the blood pressure influence time period.

[0010] Furthermore, determining the intensity of blood pressure interference on atrophic blood pressure based on the specific deformation index and blood pressure data trend at the sampling time within the blood pressure influence period includes: A two-dimensional rectangular coordinate system for blood pressure is constructed with the sampling time as the x-axis and the blood pressure data value as the y-axis, and the blood pressure coordinate point at each sampling time is determined. The blood pressure fitting curve is obtained by performing curve fitting on the blood pressure coordinate points based on the least squares method. The nonspecific interference coefficient is determined based on the instantaneous slope and specific deformation index of the blood pressure fitting curve at different sampling times. Determine the blood pressure fluctuation stability coefficient based on the numerical fluctuations in blood pressure data; The blood pressure interference intensity is obtained by weighting and multiplying the non-specific interference coefficient with the blood pressure fluctuation stability coefficient and then normalizing the result.

[0011] Further, determining the non-specific interference coefficient based on the instantaneous slope and specific deformation index of the blood pressure fitting curve at different sampling times includes: The instantaneous slope of the blood pressure fitting curve at different sampling times was determined and normalized as a non-specific deformation index. The product of the non-specific deformation index and the specific deformation index at each sampling time is normalized and used as the non-specific interference coefficient.

[0012] Furthermore, determining the blood pressure fluctuation stability coefficient based on the numerical fluctuations of blood pressure data includes: Calculate the standard deviation of all blood pressure data, and normalize the negative of the standard deviation as the blood pressure fluctuation stability coefficient.

[0013] Furthermore, the step of filtering abnormal interference moments from the blood pressure influence time period based on the intensity of blood pressure interference includes: Determine the baseline value for the intensity of blood pressure interference; The sampling time when the blood pressure interference intensity is greater than the baseline value is regarded as the abnormal interference time.

[0014] Furthermore, the benchmark value for determining the intensity of blood pressure interference includes: Calculate the mean and standard deviation of blood pressure interference intensity at all sampling times within all blood pressure influence periods, and use the sum of the mean and standard deviation as the baseline value.

[0015] The present invention has the following beneficial effects: This invention combines temporal offset analysis of hippocampal edge pixels with calculation of body-tail region-specific deformation index, and introduces an interference identification mechanism based on blood pressure data. This effectively solves the problem of misjudgment caused by the failure to distinguish vascular confounding factors in existing screening methods. Specifically, by quantifying the offset of the hippocampal edge relative to its initial state at each sampling time and dividing the body and tail regions, it can capture the dynamic of region-selective atrophy unique to Alzheimer's disease. Based on blood pressure data, it determines the time period of blood pressure influence, and further combines the specific deformation index and blood pressure trend to determine the intensity of blood pressure interference. Based on this, abnormal interference times are screened out, so that the "effective analysis time" used for abnormal judgment excludes the significant interference of non-neurodegenerative factors such as hypertension. Thus, this method effectively decouples the influence of neurodegenerative atrophy and vascular confounding factors without relying on complex biomarkers, significantly improving the specificity and reliability of identifying abnormal hippocampal atrophy. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a method for early screening of abnormalities in brain images for Alzheimer's disease, provided as an embodiment of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an early abnormality screening method for Alzheimer's disease brain imaging proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] 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 this invention pertains.

[0020] The following describes in detail, with reference to the accompanying drawings, a specific scheme for an early abnormality screening method for Alzheimer's disease brain imaging provided by the present invention.

[0021] Please see Figure 1The diagram illustrates a flowchart of an early abnormality screening method for Alzheimer's disease brain images according to an embodiment of the present invention. The method includes: S101: Acquire MRI images of the hippocampus at different sampling times and blood pressure data to determine the edge pixels of the hippocampus in the MRI images.

[0022] In related technologies, relying solely on a single MRI image to assess hippocampal morphology makes it difficult to capture its subtle dynamic atrophy process. Directly comparing the hippocampal position in original images at different time points is susceptible to interference from factors such as blood pressure, leading to distortion in deformation measurements. To address this technical problem, this invention employs blood pressure interference analysis to filter out data with significant blood pressure interference; details are provided in subsequent embodiments.

[0023] In this embodiment of the invention, fixed scanning parameters such as tube voltage, tube current, and slice thickness of a spiral MRI device can be used, and the scanning range can cover the entire brain of the patient, with the first examination as the baseline.

[0024] It should be noted that the present invention aims to analyze the state characteristics of hippocampal atrophy, and this process takes a long time. Therefore, the embodiments of the present invention perform retrospective analysis based on the patient's previously stored periodic historical image sequences (such as fixed physical examination cycles) to obtain the relevant data stored in history. Data analysis is performed when the number of stored sampling times is greater than the preset number of sampling times (such as 10). If it is less than or equal to the preset number of sampling times, no further analysis is performed.

[0025] Within one hour before and after each MRI image acquisition, the patient's blood pressure was continuously collected by wearing an ambulatory blood pressure monitor, and the average blood pressure during the duration was used as the blood pressure data at the corresponding sampling time.

[0026] In this embodiment of the invention, the Canny edge detection algorithm can be used to extract edge pixels of the hippocampus, and morphological algorithms can be used to enhance the edges. Specifically, such as the local mean denoising algorithm, Gaussian noise and salt-and-pepper noise are suppressed while the gradient information of the edge pixels of the hippocampus is enhanced. The specific edge detection and enhancement are common knowledge in the field and will not be described in detail.

[0027] Hippocampal atrophy can cause the spatial position of the same edge pixel to shift at different sampling times, which manifests as the shift of edge pixels towards the central region in two-dimensional images.

[0028] S102: Perform morphological analysis on the edge pixels in the MRI image at each sampling time to determine the body and tail regions of the hippocampus; determine the specific deformation index at each sampling time by comparing the number of edge pixels in the tail and body regions at different sampling times.

[0029] In related technologies, shrinkage analysis is performed directly based on changes in the overall volume of the hippocampus. However, if only changes in the overall volume of the hippocampus are used as an indicator of shrinkage, it is difficult to capture the unique regional selective degradation patterns in the shrinkage process (such as preferential shrinkage of the body while the tail is relatively preserved), resulting in insufficient sensitivity to early abnormalities.

[0030] To address this issue, this step focuses on the refined analysis of the internal structure of the hippocampus. In the MRI images at each sampling time, morphological analysis is performed based on the spatial distribution of edge pixels to automatically identify and divide the anatomically significant body and tail regions. Furthermore, by comparing the relative changes in the number of edge pixels in these two regions at different sampling times, the dynamics of local atrophy are quantified.

[0031] This step enables an objective measurement of the specific shrinkage trend in the hippocampus region, making the calculation of the specific deformation index more reliable and significantly improving the detectability and discriminative power of early abnormal signals.

[0032] Furthermore, in some embodiments of the present invention, morphological analysis is performed on the edge pixels in the MRI image at each sampling time to determine the body region and tail region of the hippocampus, including: determining the longest major axis of the region enclosed by the edge pixels based on principal component analysis; determining the longest major axis of the region enclosed by the edge pixels based on principal component analysis; projecting the edge pixels along the determined major axis curve, dividing the region into three equal parts based on the cumulative distribution of the projection length, and taking the middle part of the division as the body region; comparing the number of pixels in the two sides, and taking the part with fewer pixels as the tail region.

[0033] It should be noted that the hippocampus extends in an arc shape along its long axis: Head: the anterior end is enlarged and close to the amygdala; Body: the middle part is relatively slender and has a more uniform cross-section; Tail: the posterior end gradually tapers and terminates in the trigone of the lateral ventricle. Therefore, the position along the main axis is key to distinguishing the three segments.

[0034] In this embodiment of the invention, all edge pixels can be combined into an edge pixel set, its covariance matrix can be calculated, and the first principal component (the eigenvector corresponding to the largest eigenvalue) can be solved. The direction of this principal component is the best fit long axis direction of the hippocampus, reflecting its overall orientation. Then, the edge pixels are projected along the determined long axis, and the projection length is divided into three equal parts based on the cumulative distribution. The middle part of the division is taken as the body region. The number of pixels in the two sides is compared. Since the tail of the hippocampus is usually smaller than the head and has fewer pixels, the part with fewer pixels is taken as the tail region.

[0035] This scheme determines the hippocampal long axis through principal component analysis → trisects the projection interval → automatically identifies the tail based on the number of pixels, realizing a fully automatic and individualized three-part hippocampal division without prior directional information, providing a reliable anatomical basis for subsequent "body vs. tail-specific deformation index comparison".

[0036] Based on the comparison of the number of edge pixels in the tail region and body region at different sampling times, the specific deformation index at each sampling time is determined. This includes: calculating the ratio of the number of edge pixels in the tail region and body region at each sampling time as the volume ratio; constructing a two-dimensional rectangular coordinate system for the volume ratio with the sampling time as the abscissa and the volume ratio value as the ordinate, and determining the volume ratio coordinate point at each sampling time; performing curve fitting (nonlinear fitting, such as quadratic or cubic polynomial fitting) on ​​the volume ratio coordinate points based on the least squares method to obtain the volume ratio fitting curve; and determining the instantaneous slope of the volume ratio fitting curve at different sampling times and normalizing it as the specific deformation index.

[0037] Due to interference from non-neurodegenerative factors such as hypertension, the body of the hippocampus, which is rich in vulnerable neurons, atrophies faster than the tail, resulting in a continuous increase in the tail / body ratio over time. In non-AD atrophy (such as normal aging or vascular damage), this ratio changes more gradually or irregularly.

[0038] Calculate the ratio of the number of edge pixels in the tail region to the number of edge pixels in the body region, and use this ratio as the volume ratio.

[0039] It should be noted that the volume ratio does not directly represent absolute volume, but rather reflects the relative morphological proportions of the internal structures of the hippocampus. Its core physiological significance lies in the difference in the specific deformation index between the body and tail regions during the temporal progression. The tail region (Tail) is located at the posterior end of the hippocampus and has a relatively dense structure, resulting in less involvement in the early stages of Alzheimer's disease (AD). The body region (Body) is located in the middle of the hippocampus and is rich in CA1 neurons that are sensitive to AD pathology. It undergoes significant atrophy in the early stages of AD. As AD progresses, the atrophy rate of the body is significantly faster than that of the tail, leading to a faster decrease in the number of edge pixels in the body region, while the tail region remains relatively stable.

[0040] To analyze this relationship between speed and rate, the embodiments of the present invention calculate the volume ratio. The faster the volume ratio increases, the more severe the body shrinkage. If the volume ratio tends to stabilize or decreases, it indicates that it is caused by normal aging, blood pressure interference, etc.

[0041] In this embodiment of the invention, a specific deformation index is used to characterize this fast-slow relationship. That is, a two-dimensional rectangular coordinate system of volume ratio is constructed with the sampling time as the abscissa and the volume ratio value as the ordinate, and the volume ratio coordinate point at each sampling time is determined. The volume ratio coordinate point is then fitted with a curve based on the least squares method to obtain the volume ratio fitting curve. The instantaneous slope of the volume ratio fitting curve at different sampling times is determined and normalized to serve as the specific deformation index.

[0042] In one embodiment of the present invention, the normalization process can be specifically, for example, maximum and minimum value normalization. Furthermore, the normalization in subsequent steps can all adopt maximum and minimum value normalization. In other embodiments of the present invention, other normalization methods can be selected according to the specific range of the numerical values, which will not be elaborated further.

[0043] The maximum and minimum value normalization in this embodiment of the invention is intended to perform standardization processing. The maximum and minimum values ​​of the normalization processing can be set according to specific scenarios and data numerical characteristics. Adjusting, calibrating or optimizing the maximum and minimum values ​​does not constitute a limitation of this invention.

[0044] Furthermore, unless otherwise specified, the normalization functions mentioned in this invention all employ maximum-minimum value normalization to normalize the normalization results to the [0, 1] interval or other continuous intervals. The maximum and minimum values ​​used in maximum-minimum value normalization can be obtained according to the actual situation. For example, when multiple values ​​can be obtained during implementation and it is necessary to compare the magnitudes of different values, multiple values ​​can be statistically analyzed to obtain the maximum and minimum values. However, when only a single value can be obtained during implementation or it is not necessary to compare the magnitudes of different values, the maximum and minimum values ​​can be statistically obtained based on a large amount of historical experimental data or prior data. When performing maximum-minimum value analysis based on a large amount of historical experimental data, the maximum and minimum values ​​are used as cutoff values ​​for the analysis. When the data is greater than the historical maximum or less than the historical minimum, the value is truncated and determined as the corresponding cutoff value for maximum-minimum value normalization calculation.

[0045] Therefore, the higher the value of the specific deformation index, the more severe the body shrinkage. If the specific deformation index tends to stabilize or decreases, it indicates that it is caused by normal aging.

[0046] S103: Determine the time period of blood pressure influence based on blood pressure data; determine the intensity of blood pressure interference on atrophy based on the specific deformation index and blood pressure data trend at the sampling time within the blood pressure influence time period.

[0047] Determining the time period of blood pressure influence based on blood pressure data includes: taking the sampling time when the blood pressure data is greater than the preset standard blood pressure as the time of influence; and taking the time period consisting of at least a preset number of consecutive time periods of influence as the time period of blood pressure influence.

[0048] To identify the time intervals in which non-neurodegenerative factors such as hypertension significantly interfere with the hippocampal atrophy process, it is necessary to extract analytically significant events such as persistent hypertension from continuous blood pressure monitoring data, rather than isolated transient increases in blood pressure.

[0049] The preset standard blood pressure is a standard value for blood pressure comparison analysis (e.g., systolic blood pressure of 140 mmHg or 24-hour mean arterial pressure of 105 mmHg; this threshold can be adjusted according to clinical guidelines or individual patient baseline). Only when blood pressure is consistently higher than the preset standard blood pressure can it have a substantial abnormal effect on the hippocampal structure through mechanisms such as blood-brain barrier damage and cerebral small vessel disease, and this time is taken as the time of impact.

[0050] However, a single, occasional increase in blood pressure (such as that caused by emotional stress) is usually insufficient to analyze abnormal situations. Therefore, in this embodiment of the invention, a time period consisting of at least a predetermined number of consecutive influencing moments is also used as the blood pressure influence time period.

[0051] The preset number can be, for example, three, meaning a time period consisting of three or more consecutive influencing moments is considered the blood pressure influence period. Influencing moments that are isolated moments, where neither the preceding nor following sampling moment is an influencing moment, are not analyzed.

[0052] Through the two-step screening process described above—threshold determination and continuity verification—the interference of random fluctuations or transient increases was effectively eliminated, accurately pinpointing the active period of non-neurodegenerative factors such as hypertension (i.e., the blood pressure influence period) that may truly have a confounding effect on hippocampal atrophy. This blood pressure influence period will serve as input for subsequent analysis to assess the interference strength of blood pressure on the specific deformation index, thus providing a crucial time window for distinguishing between specific atrophy caused by Alzheimer's disease and vascular non-specific atrophy.

[0053] Furthermore, in some embodiments of the present invention, in order to quantify the degree of interference of non-neurodegenerative factors such as hypertension on the hippocampal atrophy process, a multi-dimensional "blood pressure interference intensity" index is constructed by comprehensively considering the coupling relationship between blood pressure change trends and atrophy dynamic response within a determined blood pressure influence period.

[0054] Specifically, based on the specific deformation index and blood pressure data trend at each sampling time within the blood pressure influence period, the intensity of blood pressure interference on atrophic blood pressure is determined. This includes: constructing a two-dimensional rectangular coordinate system for blood pressure with the sampling time as the abscissa and the blood pressure data value as the ordinate, and determining the blood pressure coordinate point at each sampling time; obtaining a blood pressure fitting curve by curve fitting the blood pressure coordinate points using the least squares method; determining the non-specific interference coefficient based on the instantaneous slope and specific deformation index of the blood pressure fitting curve at different sampling times; determining the blood pressure fluctuation stability coefficient based on the numerical fluctuation of the blood pressure data; and obtaining the blood pressure interference intensity by weighted multiplication of the non-specific interference coefficient and the blood pressure fluctuation stability coefficient and normalization.

[0055] Specifically, a two-dimensional rectangular coordinate system for blood pressure is constructed, and curve fitting is performed to obtain the blood pressure fitting curve. The fluctuation characteristics of the blood pressure fitting curve are then used to analyze the overall impact of blood pressure abnormalities.

[0056] Based on the instantaneous slope and specific deformation index of the blood pressure fitting curve at different sampling times, the non-specific interference coefficient is determined, including: determining the instantaneous slope of the blood pressure fitting curve at different sampling times and normalizing it as the non-specific deformation index; and normalizing the product of the non-specific deformation index and the specific deformation index at each sampling time as the non-specific interference coefficient.

[0057] In this embodiment of the invention, the first derivative (i.e., instantaneous slope) at different sampling times on the blood pressure fitting curve can be calculated. After maximum and minimum value normalization, the non-specific deformation index is obtained, which reflects the rate of rise or fall of blood pressure. The maximum and minimum value normalization of blood pressure data is a standardization process. Its maximum and minimum values ​​can be obtained based on historical experience. If the non-specific deformation index at certain sampling times is less than the previously obtained minimum value or greater than the maximum value, it is directly used as the corresponding minimum or maximum value for threshold truncation to avoid calculation errors.

[0058] It should be noted that, within the time period of blood pressure influence, the larger the values ​​of the specific deformation index and the non-specific deformation index at the sampling time, the greater the abnormal influence caused by non-neurodegenerative factors such as hypertension. Furthermore, both the specific and non-specific deformation indices are normalized values ​​and are dimensionless data. Therefore, in this embodiment of the invention, the product of the non-specific and specific deformation indices is directly calculated and normalized to obtain the non-specific interference coefficient. The normalization process in the calculation of the non-specific interference coefficient is also a maximum-minimum value normalization, which will not be elaborated further.

[0059] The nonspecific interference coefficient characterizes the interference value obtained by coupling the atrophy effect in the combined MRI image with the effect of blood pressure temporal changes. The larger the value, the more obvious the blood pressure interference effect at the sampling time. However, since blood pressure is a fluctuating feature, it is also necessary to analyze whether this interference is a stable interference feature. Therefore, it is also necessary to determine the blood pressure fluctuation stability coefficient.

[0060] Based on the numerical fluctuations of blood pressure data, the blood pressure fluctuation stability coefficient is determined, including: calculating the standard deviation of all blood pressure data, and normalizing the negative of the standard deviation as the blood pressure fluctuation stability coefficient.

[0061] The blood pressure fluctuation stability coefficient characterizes the reliability of the aforementioned non-specific interference coefficients. The larger the value of the blood pressure fluctuation stability coefficient, the stronger the reliability of the non-specific interference coefficient. Conversely, the smaller the value of the blood pressure fluctuation stability coefficient, the more likely the blood pressure fluctuation is caused by physiological activities or other influences. Such blood pressure fluctuations cannot reliably reflect the overall impact, and the non-specific interference coefficient is less reliable.

[0062] Since the nonspecific interference coefficient is a characteristic of the sampling time, while the blood pressure fluctuation stability coefficient is the reliability of the blood pressure influence time period, and both data are dimensionless parameters after normalization, in this embodiment of the invention, the blood pressure fluctuation stability coefficient is directly used as the weight to calculate the product of the blood pressure fluctuation stability coefficient and the nonspecific interference coefficient, and then normalized to obtain the blood pressure interference intensity.

[0063] The intensity of blood pressure interference not only reflects the magnitude of the driving effect of blood pressure on atrophy, but also takes into account the stability of its effect, providing a quantitative basis for subsequent accurate identification and elimination of abnormal moments interfered with by vascular factors, thereby improving the specificity and reliability of early screening.

[0064] S104: Based on the intensity of blood pressure interference, screen out abnormal interference moments from the time period of blood pressure influence, and take all other sampling moments that exclude abnormal interference moments as valid analysis moments.

[0065] In early screening, directly including all sampling times in the abnormality assessment can lead to significant confounding interference from non-neurodegenerative factors such as hypertension affecting hippocampal atrophy during certain time periods, resulting in misdiagnosis of vascular nonspecific atrophy as an AD pathological progression. To address this issue, this step focuses on the precise removal of interfering times. Based on the calculated intensity of blood pressure interference, sampling times with particularly prominent interference effects are identified within the blood pressure-affected time periods and marked as abnormal interference times for exclusion. All remaining sampling times that are not excluded are uniformly defined as valid analysis times.

[0066] Among them, the process of screening abnormal interference moments from the blood pressure influence time period based on the intensity of blood pressure interference includes: determining the baseline value of blood pressure interference intensity; and taking the sampling moments with blood pressure interference intensity greater than the baseline value as abnormal interference moments.

[0067] Regarding the baseline value for blood pressure interference intensity, in this embodiment of the invention, the mean and standard deviation of blood pressure interference intensity at all sampling times within all blood pressure influence time periods are calculated, and the sum of the mean and standard deviation is used as the baseline value.

[0068] Of course, in other embodiments of the present invention, the reference value can also be determined directly based on prior experience, for example, the reference value is 0.75.

[0069] The baseline value represents the normal level. That is, when the blood pressure interference intensity is greater than the baseline value, it means that the influence of blood pressure interference is more obvious, and the corresponding sampling time is regarded as the abnormal interference time.

[0070] Subsequently, for all sampling times (including periods affected by blood pressure and periods not affected by blood pressure), the remaining sampling times after filtering out abnormal interference times were selected as valid analysis times. These valid analysis times are those with minimal blood pressure interference. Thus, after excluding significant interference from vascular factors such as hypertension, the remaining sampling time points that can reliably assess the specific hippocampal atrophy process in Alzheimer's disease (AD) have a certain degree of pathological representativeness, thereby improving the effectiveness of subsequent analyses.

[0071] This invention combines temporal offset analysis of hippocampal edge pixels with the calculation of a body-tail region-specific deformation index, and introduces an interference identification mechanism based on blood pressure data. This effectively solves the problem of misjudgment caused by the failure to distinguish vascular confounding factors in existing screening methods. Specifically, by quantifying the offset of the hippocampal edge relative to its initial state at each sampling time and dividing the body and tail regions, it can capture the dynamic of region-selective atrophy unique to Alzheimer's disease. Based on blood pressure data, it determines the time period of blood pressure influence, and further combines the specific deformation index and blood pressure trend to determine the intensity of blood pressure interference. Based on this, abnormal interference times are screened out, so that the "effective analysis time" used for abnormal judgment excludes the significant interference of non-neurodegenerative factors such as hypertension. Thus, this method significantly improves the specificity and reliability of identifying abnormal hippocampal atrophy without relying on complex biomarkers.

[0072] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0073] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for early screening of abnormalities in brain imaging in Alzheimer's disease, characterized in that, The method includes: MRI images of the hippocampus at different sampling times and blood pressure data were acquired to determine the edge pixels of the hippocampus in the MRI images; Morphological analysis was performed on the edge pixels in the MRI images at each sampling time to determine the body and tail regions of the hippocampus; the specific deformation index at each sampling time was determined by comparing the number of edge pixels in the tail and body regions at different sampling times. The time period of blood pressure influence is determined based on blood pressure data; the intensity of blood pressure interference on atrophy is determined based on the specific deformation index and blood pressure data trend at the sampling time within the blood pressure influence time period. Based on the intensity of blood pressure interference, abnormal interference moments are screened out from the time period of blood pressure influence, and all other sampling moments that exclude abnormal interference moments are taken as valid analysis moments.

2. The method for early screening of abnormalities in brain images for Alzheimer's disease as described in claim 1, characterized in that, The determination of edge pixels of the hippocampus in the MRI image includes: Edge pixels are determined based on the Canny edge detection algorithm.

3. The method for early screening of abnormalities in brain images for Alzheimer's disease as described in claim 1, characterized in that, The morphological analysis of edge pixels in the MRI image at each sampling time to determine the body and tail regions of the hippocampus includes: Based on principal component analysis, the longest major axis of the region enclosed by the edge pixels is determined; the edge pixels are projected along the determined major axis curve, and the region is divided into three equal parts based on the cumulative distribution of the projection length. The middle part of the division is taken as the body region. Compare the number of pixels in the two sides and designate the area with fewer pixels as the tail area.

4. The method for early screening of abnormalities in Alzheimer's disease brain imaging as described in claim 1, characterized in that, The determination of the specific deformation index at each sampling time based on a comparison of the number of edge pixels in the tail region and the body region at different sampling times includes: Calculate the ratio of the number of edge pixels in the tail region to the number of edge pixels in the body region at each sampling time, and use it as the volume ratio; A two-dimensional rectangular coordinate system for volume ratio is constructed with the sampling time as the x-axis and the volume ratio value as the y-axis, and the volume ratio coordinate point at each sampling time is determined. The volume ratio fitting curve is obtained by curve fitting the volume ratio coordinate points based on the least squares method. The instantaneous slope of the volume ratio fitting curve at different sampling times was determined and normalized to serve as a specific deformation index.

5. The method for early screening of abnormalities in brain images for Alzheimer's disease as described in claim 1, characterized in that, The determination of the time period of blood pressure influence based on blood pressure data includes: The sampling time when the blood pressure data is greater than the preset standard blood pressure is taken as the time of influence; The time period consisting of at least a preset number of consecutive influencing moments is defined as the blood pressure influence time period.

6. The method for early screening of abnormalities in brain images for Alzheimer's disease as described in claim 1, characterized in that, The determination of the intensity of blood pressure interference on atrophic blood pressure based on the specific deformation index and blood pressure data trend at the sampling time within the blood pressure influence period includes: A two-dimensional rectangular coordinate system for blood pressure is constructed with the sampling time as the x-axis and the blood pressure data value as the y-axis, and the blood pressure coordinate point at each sampling time is determined. The blood pressure fitting curve is obtained by performing curve fitting on the blood pressure coordinate points based on the least squares method. The nonspecific interference coefficient is determined based on the instantaneous slope and specific deformation index of the blood pressure fitting curve at different sampling times. Determine the blood pressure fluctuation stability coefficient based on the numerical fluctuations in blood pressure data; The blood pressure interference intensity is obtained by weighting and multiplying the non-specific interference coefficient with the blood pressure fluctuation stability coefficient and then normalizing the result.

7. The method for early screening of abnormalities in brain images for Alzheimer's disease as described in claim 6, characterized in that, The determination of the nonspecific interference coefficient based on the instantaneous slope and specific deformation index of the blood pressure fitting curve at different sampling times includes: The instantaneous slope of the blood pressure fitting curve at different sampling times was determined and normalized as a non-specific deformation index. Calculate the product of the non-specific deformation index and the specific deformation index at each sampling time, and normalize it to obtain the non-specific interference coefficient.

8. The method for early screening of abnormalities in brain images for Alzheimer's disease as described in claim 6, characterized in that, The determination of the blood pressure fluctuation stability coefficient based on the numerical fluctuation of blood pressure data includes: Calculate the standard deviation of all blood pressure data, and normalize the negative of the standard deviation as the blood pressure fluctuation stability coefficient.

9. The method for early screening of abnormalities in brain images for Alzheimer's disease as described in claim 1, characterized in that, The method of filtering abnormal interference moments from the time period of blood pressure influence based on the intensity of blood pressure interference includes: Determine the baseline value for the intensity of blood pressure interference; The sampling time when the blood pressure interference intensity is greater than the baseline value is regarded as the abnormal interference time.

10. The method for early screening of abnormalities in brain images for Alzheimer's disease as described in claim 9, characterized in that, The benchmark values ​​for determining the intensity of blood pressure interference include: Calculate the mean and standard deviation of blood pressure interference intensity at all sampling times within all blood pressure influence periods, and use the sum of the mean and standard deviation as the baseline value.

Images