A method for constructing a three-substance model based on double-layer spectral CT and a quantitative analysis system for bone density and bone marrow fat content

Abstract

The present invention relates to the field of medical data processing technology, and discloses a method for constructing a three-substance model based on double-layer spectral CT and a quantitative analysis system for bone density and bone marrow fat content, including: according to the double-layer spectral CT data of the target group located at the lumbar vertebral body, using the three-substance decomposition method, the bone matrix volume fraction of the target group and the yellow bone marrow volume fraction representing the bone marrow fat content are obtained; according to the bone matrix volume fraction of the target group, the bone density of the target group is obtained; using the double-layer spectral CT data of the target group located at the lumbar vertebral body as sample data, and using the bone density and bone marrow fat content of the target group as label data, the three-substance model based on the double-layer spectral CT is trained. The present invention can accurately quantify bone density and bone marrow fat content based on the HU value of the double-layer spectral CT image at the same time, reduce the burden of medical resources, help to better understand the role of bone marrow fat in bone metabolic diseases, and provide a reliable reference basis for clinical diagnosis and treatment.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a method for constructing a three-substance model based on double-layer spectral CT and a quantitative analysis system for bone density and bone marrow fat content. Background Art

[0002] Osteoporosis is a common chronic disease in the elderly, leading to complications such as brittle fractures and pain, which seriously affect the quality of life. In recent years, the number of osteoporosis patients has increased year by year, with a prevalence of 32.0% in people over 65 years old, 51.6% in women and 10.9% in men. Among people over 50 years old worldwide, about 33% to 50% of women and 20% of men will experience at least one fracture. In addition, among people over 40 years old, the vertebral fracture rates of male and female patients are as high as 10.5% and 9.7%, respectively. Patients with osteoporotic fractures are often accompanied by serious complications such as pain, deep vein thrombosis, limited activity, acute cardiovascular and cerebrovascular events, and recurrent fractures, increasing the risk of death and disability. In addition, studies have shown that decreased bone density may be associated with increased bone marrow fat content in patients with osteoporosis. Therefore, the assessment of bone marrow fat content may become an important clinical tool for accurately assessing bone density.

[0003] Bone mineral density plays a central role in diagnosing osteoporosis, predicting fracture risk, and monitoring the efficacy of osteoporosis treatment. Globally, dual-energy X-ray absorptiometry (DXA) is the most widely used method for assessing bone mineral density (BMD). However, the area BMD (aBMD, mg / cm2) measured by DXA is not very accurate. 2 ) is a two-dimensional imaging method that is susceptible to artifacts caused by vertebral degeneration. Quantitative computed tomography (QCT) is an alternative method for measuring BMD. It evaluates the trabecular bone volume (vBMD, mg / cm 3 ) overcomes the limitations of DXA and shows superior sensitivity for identifying patients with osteoporosis. However, QCT measures vBMD at a single effective photon energy, and the results are subject to measurement errors caused by changes in bone marrow fat content. Magnetic Resonance Imaging (MRI) technology uses magnetic fields and harmless radio waves to create detailed images with good contrast for soft tissue and bone marrow fat. In osteoporosis research, MRI can be used to assess the distribution and content of bone marrow fat, but it cannot directly measure bone density.

[0004] In summary, although these techniques are helpful in assessing BMD and marrow fat content, none of them can measure both vertebral BMD and marrow fat content simultaneously. Therefore, it is necessary to develop a method that can accurately quantify BMD and marrow fat content simultaneously to correct for the potential effect of marrow fat on BMD. Summary of the invention

[0005] The present invention provides a method for constructing a three-substance model based on double-layer spectral CT and a quantitative analysis system for bone density and bone marrow fat content, so as to solve the defect that the prior art cannot accurately and quantitatively analyze bone density and bone marrow fat content.

[0006] The present invention provides a method for constructing a three-substance model based on double-layer spectral CT, comprising:

[0007] Obtaining double-layer spectral CT data of the target group located at the lumbar vertebral body, wherein the target group includes an osteoporosis patient group and a healthy group;

[0008] According to the double-layer spectral CT data of the target group located at the lumbar vertebral body, the bone matrix volume fraction, the yellow bone marrow volume fraction, and the red bone marrow volume fraction of the target group are obtained by using the three-substance decomposition method, wherein the yellow bone marrow volume fraction represents the bone marrow fat content;

[0009] According to the bone matrix volume fraction of the target group, the bone density of the target group is obtained;

[0010] The double-layer spectral CT data of the target group at the lumbar vertebrae were used as sample data, and the bone density and bone marrow fat content of the target group were used as label data, and a three-substance model based on double-layer spectral CT was trained.

[0011] In one embodiment, the double-slice spectral CT data of the target group located at the lumbar vertebral body includes: performing double-slice spectral CT on the lumbar vertebral body of the target group at energy E 1 and energy E 2 The HU (E 1 ) value and HU(E 2 )value.

[0012] In one embodiment, the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group are obtained by three-substance decomposition method based on the double-layer spectral CT data of the target group located at the lumbar vertebral body, wherein the yellow bone marrow volume fraction represents the bone marrow fat content, including:

[0013] Get the target group in energy E 1 and energy E 2The water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient under the target group are set as variables, and the sum of the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group is set to 1;

[0014] According to the target group in energy E 1 and energy E 2 The water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient under the energy E are combined with water density, bone matrix density, yellow bone marrow density, and red bone marrow density to obtain the target group 1 and energy E 2 The linear attenuation coefficient of water, the linear attenuation coefficient of bone matrix, the linear attenuation coefficient of yellow bone marrow, and the linear attenuation coefficient of red bone marrow;

[0015] According to the target group in energy E 1 and energy E 2 The HU (E 1 ) value and HU(E 2 ) value, combined with the target group in energy E 1 and energy E 2 The water linear attenuation coefficient under the energy E is obtained. 1 and energy E 2 The linear attenuation coefficient μ(E 1 ) and μ(E 2 );

[0016] According to the target group in energy E 1 and energy E 2 The linear attenuation coefficient μ(E 1 ) and μ(E 2 ), combined with the target group in energy E 1 and energy E 2 The bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient under the conditions are obtained to obtain the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group.

[0017] In one embodiment, the energy E 1 and energy E 2 The water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient under the energy E are combined with water density, bone matrix density, yellow bone marrow density, and red bone marrow density to obtain the target group 1 and energy E 2The water linear attenuation coefficient, bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient under the condition include:

[0018] According to the target group in energy E 1 and energy E 2 The water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient under the energy E are obtained by combining the water density, bone matrix density, yellow bone marrow density, and red bone marrow density using the first expression. 1 and energy E 2 The water linear attenuation coefficient, bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient under the condition are as follows:

[0019] μ water (E 1 )=μ m1(water) × water density,

[0020] μ water (E 2 )=μ m2(water) × water density,

[0021] μ BM (E 1 )=μ m1(BM) × the density of BM,

[0022] μ BM (E 2 )=μ m2(BM) × the density of BM,

[0023] μ YM (E 1 )=μ m1(YM) ×YM density,

[0024] μ YM (E 2 )=μ m2(YM) ×YM density,

[0025] μ RM (E 1 )=μ m1(RM) × the density of RM,

[0026] μ RM (E 2 )=μ m2(RM) × the density of RM,

[0027] In the first expression, μ water (E 1 ) and μ water (E 2) indicates that the target group is in energy E 1 and energy E 2 Water linear attenuation coefficient under m1(water) and μ m2(water) Indicates that the target group is in energy E 1 and energy E 2 The water mass attenuation coefficient under water, the density of water represents the density of water, μ bM (E 1 ) and μ BM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of bone matrix under m1(BM) and μ m2(BM) Indicates that the target group is in energy E 1 and energy E 2 The bone matrix mass attenuation coefficient under BM is , and the density of BM represents the density of bone matrix, μ YM (E 1 ) and μ YM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of yellow bone marrow under m1(YM) and μ m2(YM) Indicates that the target group is in energy E 1 and energy E 2 The yellow bone marrow mass attenuation coefficient under the condition of YM represents the density of yellow bone marrow, μ RM (E 1 ) and μ RM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of red bone marrow under m1(RM) and μ m2(RM) Indicates that the target group is in energy E 1 and energy E 2 The red bone marrow mass attenuation coefficient under RM, the density of RM represents the density of red bone marrow.

[0028] In one embodiment, the energy E 1 and energy E 2 The HU (E 1 ) value and HU(E 2 ) value, combined with the target group in energy E 1 and energy E 2 The water linear attenuation coefficient under the energy E is obtained. 1 and energy E 2 The linear attenuation coefficient μ(E1 ) and μ(E 2 ),include:

[0029] According to the target group in energy E 1 and energy E 2 The HU (E 1 ) value and HU(E 2 ) value, combined with the target group in energy E 1 and energy E 2 The water linear attenuation coefficient under the condition of E is obtained by using the second expression: 1 and energy E 2 The linear attenuation coefficient μ(E 1 ) and μ(E 2 ), where the second expression is:

[0030]

[0031] In the second expression, μ(E 1 ) and μ(E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient under 1 ) value and HU(E 2 ) value indicates the target group in energy E 1 and energy E 2 HU value of the single-energy image measured at water (E 1 ) and μ water (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The water linear attenuation coefficient under 1000 [HU] represents the constant of the Hornsfield unit, which is used to convert HU (Hohnsfield unit) to the linear attenuation coefficient (μ). Specifically, 1000 [HU] in the formula is used to standardize the HU value and convert it into the corresponding linear attenuation coefficient, so that the calculation is more accurate and meaningful.

[0032] In one embodiment, the energy E 1 and energy E 2 The linear attenuation coefficient μ(E 1 ) and μ(E 2 ), combined with the target group in energy E 1 and energy E 2 The bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient under the above conditions are used to obtain the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group, including:

[0033] According to the target group in energy E 1 and energy E 2 The linear attenuation coefficient μ(E 1 ) and μ(E 2 ), combined with the target group in energy E 1 and energy E 2 The linear attenuation coefficient of bone matrix, the linear attenuation coefficient of yellow bone marrow, and the linear attenuation coefficient of red bone marrow under the above conditions are used to obtain a plurality of quantitative determination coefficients by using the third expression;

[0034] According to the plurality of quantitative determination coefficients, the bone matrix volume fraction, the yellow bone marrow volume fraction, and the red bone marrow volume fraction of the target population are obtained using the fourth expression.

[0035] In one embodiment, the third expression is:

[0036] D 1 =μ YM (E 1 )*μ RM (E 2 )-μ RM (E 1 )*μ(E 2 )-μ(E 1 )*μ YM (E 2 )+μ(E 1 )*μ RM (E 2 )+

[0037] μ(E 2 )*μ YM (E 1 )+μ RM (E 1 )*μ YM (E 2 ),

[0038] D 2 =μ BM (E 1 )*μ RM (E 2 )-μ RM (E 1 )*μ(E 2 )-μ(E 1 )*μ BM (E 2 )+μ(E 1 )*μ RM (E 2 )+

[0039] μ(E 2 )*μBM (And 1 )+μ RM (And 1 )*μ BM (And 2 ),

[0040] D 3 =μ BM (And 1 )*μ YM (And 2 )-μ YM (And 1 )*μ(E 2 )-μ(E 1 )*μ BM (And 2 )+μ(E 1 )*μ YM (And 2 )+

[0041] μ(E 2 )*μ BM (And 1 )+μ YM (And 1 )*μ BM (And 2 ),

[0042] D 4 =μ BM (And 1 )*μ YM (And 2 )+μ YM (And 1 )*μ BM (And 2 )+μ RM (And 1 )*μ BM (And 2 )-μ RM (And 1 )*μ YM (And 2 )

[0043] -μ YM (And 1 )*μ BM (And 2 )-μ BM (And 1 )*μ RM (And 2 ),

[0044] In the third expression, D 1 、D 2 、D 3 、D4 Represents multiple quantitative determination coefficients, μ(E 1 ) and μ(E 2 ) indicates that the target group is in energy E 1 and energy E 2 Linear attenuation coefficient under YM (E 1 ) and μ YM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of bone matrix under BM (E 1 ) and μ BM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of bone matrix under RM (E 1 ) and μ RM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of red bone marrow under .

[0045] In one embodiment, the fourth expression is:

[0046]

[0047] f 1 +f 2 +f 3 =1,

[0048] In the fourth expression, f 1 represents the bone matrix volume fraction of the target population, f 2 represents the yellow bone marrow volume fraction of the target population, f 3 Represents the red bone marrow volume fraction of the target population.

[0049] In one embodiment, obtaining the bone density of the target population according to the bone matrix volume fraction of the target population comprises:

[0050] According to the bone matrix volume fraction of the target group, the bone density of the target group is obtained using the bone density expression, where the bone density expression is:

[0051] vBMD=f 1 ×f HA × density of HA × 1000,

[0052] In the bone density expression, vBMD represents the bone density of the target group, f 1represents the bone matrix volume fraction of the target population, f HA represents the known volume fraction of hydroxyapatite, the density of HA represents the known density of hydroxyapatite, and 1000 is used for unit conversion.

[0053] Among them, bone matrix components: 58% hydroxyapatite (HA) [Ca 10 (PO 4 ) 6 (OH) 2 ]、32% collagen[C 57 H 91 N 19 O 16 ], (19) and 10% water.

[0054] Bone matrix mass attenuation coefficient = 58% × mass attenuation coefficient of hydroxyapatite (HA) (known) + 32% × mass attenuation coefficient of collagen (known) + 10% × mass attenuation coefficient of water (known).

[0055] In one embodiment, the double-layer spectral CT data of the target group at the lumbar vertebral body is used as sample data, and the bone density and bone marrow fat content of the target group are used as label data to train a three-substance model based on the double-layer spectral CT, including:

[0056] The double-layer spectral CT data of the target group at the lumbar vertebrae were used as sample data, and the bone density and bone marrow fat content of the target group were used as label data to train a three-substance model based on double-layer spectral CT. During the training process, the density of YM and the density of RM were regarded as free parameters, and the free parameters were adjusted so that the average volume of YM accounted for half of the total bone marrow volume. The three-substance model based on double-layer spectral CT was trained, in which the total bone marrow volume was the sum of the yellow bone marrow volume and the red bone marrow volume.

[0057] The present invention also provides a system for constructing a three-substance model based on double-layer spectral CT, comprising:

[0058] A data acquisition module is used to: acquire double-layer spectral CT data of a target group located at a lumbar vertebral body, wherein the target group includes an osteoporosis patient group and a healthy group;

[0059] A decomposition module is used to: obtain the bone matrix volume fraction, the yellow bone marrow volume fraction, and the red bone marrow volume fraction of the target group according to the double-layer spectral CT data of the target group located at the lumbar vertebral body by using a three-substance decomposition method, wherein the yellow bone marrow volume fraction represents the bone marrow fat content;

[0060] A data processing module is used to obtain the bone density of the target group according to the bone matrix volume fraction of the target group;

[0061] The training module is used to: use the double-layer spectral CT data of the target group located at the lumbar vertebrae as sample data, and use the bone density and bone marrow fat content of the target group as label data to train a three-substance model based on the double-layer spectral CT.

[0062] The present invention also provides a quantitative analysis system for bone density and bone marrow fat content based on double-layer spectral CT, comprising:

[0063] A data receiving module, used for: receiving double-layer spectral CT data of a subject at a lumbar vertebral body uploaded by at least one client, wherein the type of the double-layer spectral CT data of the subject at a lumbar vertebral body is consistent with the type of the double-layer spectral CT data of the target group at a lumbar vertebral body in any of the above-mentioned methods for constructing a three-substance model based on double-layer spectral CT, and the subject is an individual outside a preset filtered-out group, and the preset filtered-out group includes a group with scoliosis, multiple spinal fractures, multiple bone tumors, and multiple metal implants;

[0064] A quantitative analysis module, for predicting the bone density and bone marrow fat content of the subject according to the double-layer spectral CT data of the subject at the lumbar vertebral body, using a three-substance model based on double-layer spectral CT obtained by the construction method of a three-substance model based on double-layer spectral CT according to any one of the above-mentioned items;

[0065] The data output module is used to send the bone density and bone marrow fat content of the subject to be tested back to at least one client.

[0066] The present invention also provides an electronic device, comprising a processor and a memory storing a computer program, wherein the processor implements any of the above-mentioned methods for constructing a three-substance model based on double-layer spectral CT when executing the computer program.

[0067] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for constructing a three-substance model based on double-layer spectral CT.

[0068] The present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute any of the above-mentioned methods for constructing a three-substance model based on double-layer spectral CT.

[0069] The present invention provides a method for constructing a three-substance model based on double-layer spectral CT and a quantitative analysis system for bone density and bone marrow fat content. Based on double-layer spectral CT and combined with a three-substance decomposition method, a three-substance model based on double-layer spectral CT is trained, which can accurately and quantitatively analyze bone density and bone marrow fat content based on the HU value of the double-layer spectral CT image. Compared with traditional methods, the present invention can not only simultaneously measure lumbar vertebrae volume bone density and bone marrow fat content, but also significantly shorten the patient's examination time, reduce the burden of medical resources, and reduce the patient's economic cost. In addition, the present invention not only solves the problem that traditional imaging methods are difficult to accurately measure bone marrow fat content, but also helps to better understand the role of bone marrow fat in bone metabolic diseases, and provide a reliable reference for clinical diagnosis and treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0071] Figure 1 A schematic flow chart of a method for constructing a three-substance model based on double-layer spectral CT provided by the present invention.

[0072] Figure 2 (a) shows the VOI delineated on the DLCT virtual single-energy image using 3D slicer software; Figure 2 (b) shows the VOI delineated on the MR Dixon FF image using 3D slicer software; Figure 2 (c) shows the VOI delineated on the DLCT unenhanced axial image using QCT software.

[0073] Figure 3 A shows the volume fractions of bone matrix (BM), yellow bone marrow (YM), and red bone marrow (RM) of the 100 subjects in the training set. The mass densities of YM and RM are 0.93 g / cm 3 and 1.065g / cm 3 , so that the mean adjusted YM volume fraction (=YM / [YM+RM]) was 0.50, ±1.96, and the standard deviation limits were 0.25 and 0.75, respectively. Figure 3 B shows the quality assurance diagram of 50 and 150 keV single energy HU measurements. The diamond ABCD shows that the BM, YM and RM are 1.92, 0.93 and 1.065 g / cm 3Predictive limits of physically feasible solutions of the three-material model for density and vBMD values ​​between 50 and 200 mg / cm 3 between.

[0074] Figure 4 A shows the volume fractions of bone matrix (BM), yellow marrow (YM), and red marrow (RM) of 100 subjects in the training set, where the mass densities of BM, YM, and RM were set to 1.92, 0.92, and 1.02 g / cm, respectively. 3 . Figure 4 B shows the quality assurance diagram for the 50 and 150 keV single energy HU measurements. Diamonds ABCD show the predicted limits of physically feasible solutions to the three-material model for vBMD values ​​between 50 and 200 mg / cm 3 between.

[0075] Figure 5 A shows the quality assurance diagram for the 50 and 150 keV single energy HU measurements. Diamonds ABCD show the predicted limits of physically feasible solutions to the three-material model for vBMD values ​​between 50 and 200 mg / cm 3 This is the data of 125 participants in the validation set (with Figure 3 The training set is similar to that in Figure B for comparison). Data points for three vertebrae of the European Spine Model (ESP) have been added to the figure. Figure 5 B shows diamond ABCD redrawn from the same data, assuming the single energies are 55 and 150 keV, respectively, and the densities of YM and RM are 0.925 and 1.02 g / cm, respectively. 3 .

[0076] Figure 6 This is a schematic diagram of the structure of a bone density and bone marrow fat content quantitative analysis system based on double-layer spectral CT provided by the present invention.

[0077] Figure 7 This is a schematic structural diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments, and they should not be understood as limitations on the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0079] Dual-layer spectral CT (DLCT) is a new imaging technology that can analyze the composition of various substances with the spectral information generated by each scan. Two superimposed detectors are used to separate low-energy and high-energy photons, and provide specific energy information of X-ray attenuation, and then obtain the density information of specific substances. This makes it feasible for this technology to quantify bone density and bone marrow fat content at the same time. The present invention has developed for the first time a three-substance model based on DLCT, which can accurately measure bone density and bone marrow fat content at the same time, eliminate the interference of bone marrow fat content on bone density, and more clearly understand the effect of bone marrow fat (MAT) on bone density. The successful application of this model will provide a new imaging tool and method for the diagnosis and management of osteoporosis. The following is combined with Figure 1-Figure 7 The present invention describes a method for constructing a three-substance model based on double-layer spectral CT and a quantitative analysis system for bone density and bone marrow fat content. It should be noted that the execution subject of the method for constructing a three-substance model based on double-layer spectral CT provided by the present invention can be any network-side device / terminal-side device that meets the technical requirements, such as a bone density and bone marrow fat content quantitative analysis device based on double-layer spectral CT.

[0080] Figure 1 It is a schematic flow chart of the method for constructing a three-substance model based on double-layer spectral CT provided by the present invention.

[0081] Reference Figure 1 The present invention provides a method for constructing a three-substance model based on double-layer spectral CT, which may include:

[0082] Step S110, obtaining double-layer spectral CT data of the target group located at the lumbar vertebral body, wherein the target group includes an osteoporosis patient group and a healthy group, and the double-layer spectral CT data of the target group located at the lumbar vertebral body includes performing double-layer spectral CT on the lumbar vertebral body of the target group at energy E 1 and energy E 2 The HU (E 1 ) value and HU(E 2 )value.

[0083] Step S120: According to the double-layer spectral CT data of the target group at the lumbar vertebral body, the bone matrix volume fraction, the yellow bone marrow volume fraction, and the red bone marrow volume fraction of the target group are obtained by using a three-substance decomposition method, wherein the yellow bone marrow volume fraction represents the bone marrow fat content.

[0084] In one embodiment, step S120 may include:

[0085] Get the target group in energy E 1 and energy E 2The water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient under energy E are calculated. The water density, bone matrix density, yellow bone marrow density, and red bone marrow density are set as variables, and the sum of the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group is set to 1. Specifically, the target group can be obtained from the open source X-ray mass attenuation coefficient database at energy E. 1 and energy E 2 Water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient under the condition;

[0086] According to the target group in energy E 1 and energy E 2 The water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient under the energy E are obtained by combining the water density, bone matrix density, yellow bone marrow density, and red bone marrow density using the first expression. 1 and energy E 2 The water linear attenuation coefficient, bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient under the condition are as follows:

[0087] μ water (E 1 )=μ m1(water) × water density,

[0088] μ water (E 2 )=μ m2(water) × water density,

[0089] μ BM (E 1 )=μ m1(BM) × the density of BM,

[0090] μ BM (E 2 )=μ m2(BM) × the density of BM,

[0091] μ YM (E 1 )=μ m1(YM) ×YM density,

[0092] μ YM (E 2 )=μ m2(YM) ×YM density,

[0093] μ RM (E 1 )=μ m1(RM) × the density of RM,

[0094] μ RM (E 2 )=μ m2(RM) × the density of RM,

[0095] In the first expression, μ water (E 1 ) and μ water (E 2 ) indicates that the target group is in energy E 1 and energy E 2 Water linear attenuation coefficient under m1(water) and μ m2(water) Indicates that the target group is in energy E 1 and energy E 2 The water mass attenuation coefficient under water, the density of water represents the density of water, μ BM (E 1 ) and μ BM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of bone matrix under m1(BM) and μ m2(BM) Indicates that the target group is in energy E 1 and energy E 2 The bone matrix mass attenuation coefficient under BM is , and the density of BM represents the density of bone matrix, μ YM (E 1 ) and μ YM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of yellow bone marrow under m1(YM) and μ m2(YM) Indicates that the target group is in energy E 1 and energy E 2 The yellow bone marrow mass attenuation coefficient under the condition of YM represents the density of yellow bone marrow, μ RM (E 1 ) and μ RM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of red bone marrow under m1(RM) and μ m2(RM) Indicates that the target group is in energy E 1 and energy E 2 The red bone marrow mass attenuation coefficient under RM, the density of RM represents the density of red bone marrow;

[0096] According to the target group in energy E 1 and energy E 2The HU (E 1 ) value and HU(E 2 ) value, combined with the target group in energy E 1 and energy E 2 The water linear attenuation coefficient under the condition of E is obtained by using the second expression: 1 and energy E 2 The linear attenuation coefficient μ(E 1 ) and μ(E 2 ), where the second expression is:

[0097]

[0098] In the second expression, μ(E 1 ) and μ(E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient under 1 ) value and HU(E 2 ) value indicates the target group in energy E 1 and energy E 2 HU value of the single-energy image measured at water (E 1 ) and μ water (E 2 ) indicates that the target group is in energy E 1 and energy E 2 Water linear attenuation coefficient under ;

[0099] According to the target group in energy E 1 and energy E 2 The linear attenuation coefficient μ(E 1 ) and μ(E 2 ), combined with the target group in energy E 1 and energy E 2 The linear attenuation coefficient of bone matrix, the linear attenuation coefficient of yellow bone marrow, and the linear attenuation coefficient of red bone marrow under the condition are obtained by using the third expression, wherein the third expression is:

[0100] D 1 =μ YM (E 1 )*μ RM (E 2 )-μ RM (E 1 )*μ(E 2 )-μ(E 1 )*μ YM (E 2 )+μ(E 1 )*μRM (And 2 )+

[0101] μ(E 2 )*μ YM (And 1 )+μ RM (And 1 )*μ YM (And 2 ),

[0102] D 2 =μ BM (And 1 )*μ RM (And 2 )-μ RM (And 1 )*μ(E 2 )-μ(E 1 )*μ BM (And 2 )+μ(E 1 )*μ RM (And 2 )+

[0103] μ(E 2 )*μ BM (And 1 )+μ RM (And 1 )*μ BM (And 2 ),

[0104] D 3 =μ BM (And 1 )*μ YM (And 2 )-μ YM (And 1 )*μ(E 2 )-μ(E 1 )*μ BM (And 2 )+μ(E 1 )*μ YM (And 2 )+

[0105] μ(E 2 )*μ BM (And 1 )+μ YM (And 1 )*μ BM (And 2 ),

[0106] D 4 =μ BM (And1 )*μ YM (E 2 )+μ YM (E 1 )*μ BM (E 2 )+μ RM (E 1 )*μ BM (E 2 )-μ RM (E 1 )*μ YM (E 2 )

[0107] -μ YM (E 1 )*μ BM (E 2 )-μ BM (E 1 )*μ RM (E 2 ),

[0108] In the third expression, D 1 , D 2 , D 3 , D 4 Represents multiple quantitative determination coefficients, μ(E 1 ) and μ(E 2 ) indicates that the target group is in energy E 1 and energy E 2 Linear attenuation coefficient under YM (E 1 ) and μ YM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of bone matrix under BM (E 1 ) and μ BM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 The linear attenuation coefficient of bone matrix under RM (E 1 ) and μ RM (E 2 ) indicates that the target group is in energy E 1 and energy E 2 Linear attenuation coefficient of red bone marrow under ;

[0109] According to the multiple quantitative determination coefficients, the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group are obtained using the fourth expression, wherein the fourth expression is:

[0110]

[0111] f 1 +f 2 +f 3 =1,

[0112] In the fourth expression, f 1 represents the bone matrix volume fraction of the target population, f 2 represents the yellow bone marrow volume fraction of the target population, f 3 Represents the red bone marrow volume fraction of the target population.

[0113] Step S130: Obtain the bone density of the target group according to the bone matrix volume fraction of the target group. Specifically, step S130 can obtain the bone density of the target group according to the bone matrix volume fraction of the target group using a bone density expression, wherein the bone density expression is:

[0114] vBMD=f 1 ×f HA × density of HA × 1000,

[0115] In the bone density expression, vBMD represents the bone density of the target group, f 1 represents the bone matrix volume fraction of the target population, f HA represents the known volume fraction of hydroxyapatite, and the density of HA represents the known density of hydroxyapatite, wherein the volume fraction and density of hydroxyapatite can be obtained from an open source database.

[0116] Step S140: Using the double-layer spectral CT data of the target group at the lumbar vertebral body as sample data, and using the bone density and bone marrow fat content of the target group as label data, a three-substance model based on double-layer spectral CT is trained.

[0117] Specifically, step S140 can use the double-layer spectral CT data of the target group located at the lumbar vertebral body as sample data, and the bone density and bone marrow fat content of the target group as label data to train the three-substance model based on the double-layer spectral CT. During the training process, the density of YM and the density of RM are regarded as free parameters, and by adjusting the free parameters, the average volume of YM accounts for half of the total bone marrow volume, so as to train the three-substance model based on the double-layer spectral CT, wherein the total bone marrow volume is the sum of the yellow bone marrow volume and the red bone marrow volume.

[0118] After obtaining the three-substance model based on double-layer spectral CT, the double-layer spectral CT data of the subject at the lumbar vertebrae that are consistent with the type of double-layer spectral CT data of the target group at the lumbar vertebrae can be input into the three-substance model based on double-layer spectral CT to predict the bone density and bone marrow fat content of the subject, that is, the bone density and bone marrow fat content of the subject can be obtained at the same time.

[0119] Specifically, this example retrospectively included the dual-layer spectral CT data and bone density (bone matrix volume fraction), and bone marrow fat content (yellow bone marrow volume fraction) of 100 patients in the spine outpatient clinic of Beijing Jishuitan Hospital as a training set. All patients underwent lumbar spine plain scan examination on dual-layer spectral CT. Patient inclusion criteria are as follows: (1) age ≥ 18 years old; (2) spine outpatients; (3) all underwent lumbar spine CT plain scan on dual-layer spectral CT. Exclusion criteria are as follows: (1) the presence of primary or metastatic bone tumors in the spine; (2) severe scoliosis; (3) fractures involving more than two vertebrae; (4) lumbar metal implants after spinal surgery; (5) missing spectral data.

[0120] The training set was used to train the three-substance model based on dual-layer spectral CT, and then a separate validation set was used for validation. In this example, 125 subjects from the community near Beijing Jishuitan Hospital were prospectively recruited as the validation set. All subjects underwent non-contrast enhanced lumbar DLCT and spine MR examinations. The inclusion criteria for the subjects were as follows: (1) age ≥ 18 years old; (2) self-awareness and communication barrier; (3) no contraindications to MR and CT examinations; (4) all subjects underwent lumbar spine plain scan examinations on dual-layer spectral CT and 3.0T MR. The exclusion criteria were as follows: (1) patients taking glucocorticoids; (2) patients with a history of bone tumors or other diseases that affect bone metabolism; (3) patients with severe degenerative changes in the lumbar spine or more than two vertebral fractures; (4) patients after spinal surgery or with lumbar implants; (5) patients with claustrophobia. The validation set was used to verify the accuracy of bone density and marrow adipose tissue (MAT) predicted by the three-substance model based on dual-layer spectral CT.

[0121] To further verify the accuracy of the model in predicting bone density and bone marrow fat content, this embodiment also uses volumetric bone density measured by QCT and proton density fat fraction (PDFF) measured by MR as reference standards for verifying the model. In this embodiment, the patient group used to obtain the training set data is the target group, or the patient group used to obtain the training set data and the subject group used to obtain the validation set data are jointly used as the target group.

[0122] Among them, the patient's lumbar spine can be scanned by a DLCT scanner (IQon Spectral CT, Philips Healthcare). The patient's standard tube voltage is 120 kV, the tube current is 125 mA, the slice thickness is 1 mm, the matrix is ​​512 × 512, and the pitch is 1. The spectral base image (SBI) of the axial soft tissue with a slice thickness of 1 mm was reconstructed into a virtual mono-energetic image (VMI) of 50 keV and 150 keV using IntelliSpace Portal software (version 11.1.1, Philips Healthcare). In addition, the 1 mm soft tissue original image obtained by the DLCT scan was imported into the QCT Pro software (Mindways software, Inc, TX, USA) for vBMD value measurement. Model 4 and European Spinal Body Membrane (ESP) were used for asynchronous calibration (Mindways Software, Inc., TX, USA) and quality assurance (QA). The CT value of air was always around -1000, while the CT value of water was approximately 0. In addition, prospectively recruited volunteers underwent conventional lumbar spine MRI scans on a 3.0T MRI scanner (Vida, SIEMENS, Magnetom, Germany) with an axial 3D Vibe-Dixon scanning sequence.

[0123] Then, the different X-ray attenuation coefficients of different substances are used to decompose the material components. Different substances produce different attenuation coefficients because the photoelectric effect produced by the interaction between X-rays and substances is highly dependent on the effective atomic number and photon energy of the substance, while Compton scattering is weakly dependent on both. 1 and E 2 When expressed in HU (Hounsfield units), at energy E 1 and E 2 The HU values ​​of the single-energy images measured at 1 ) and HU(E 2 ). Its corresponding linear attenuation coefficient μ(E 1 ) and μ(E 2 ) is as follows:

[0124]

[0125] μ wateR(E) represents the linear attenuation coefficient of water at photon energy E (obtained by multiplying the mass attenuation coefficient of water in NIST by the density of water), which comes from the X-ray mass attenuation coefficient database of the United States National Institute of Technology and Standards (NIST). In addition to considering the linear attenuation coefficients measured in the two monoenergetic images, it is also necessary to consider that the sum of the volume fractions f of various substances must be equal to 1, as shown in the following formula. Therefore, three linear equations can be solved using the volume fractions of the three selected substances, namely:

[0126] μ(E 1 )=f 1 μ 1 (E 1 )+f 2 μ 2 (E 1 )+f 3 μ 3 (E 1 ),

[0127] μ(E 2 )=f 1 μ 1 (E 2 )+f 2 μ 2 (E 2 )+f 3 μ 3 (E 2 ),

[0128] f 1 +f 2 +f 3 =1,

[0129] In order to measure bone density, the three substances in the research model of this embodiment are bone matrix (bonematrix, BM), yellow marrow (yellow marrow, YM), and red marrow (red marrow, RM). The linear attenuation coefficients corresponding to the three substances are expressed in μ 1,2,3 (E) indicates. 1 、f 2 、f 3Represent the volume fractions of BM, YM and RM respectively. The chemical composition of YM and RM is taken from the International Commission on Radiation Units and Measurements Report 46. When searching for the chemical composition of the corresponding substances in this embodiment, the basic principle is to assume that the bone matrix of all subjects included in this embodiment is normally mineralized (i.e., the subjects do not have osteomalacia). The selected photon energies E1 and E2 are 50 and 150 keV, respectively. The mass attenuation coefficient derived from the NIST table is converted into a linear attenuation coefficient by multiplying the physical density of the corresponding substance.

[0130] To measure the QCT vBMD (mg / cm 3 ), all CT images were transferred to the Mindways QCT Pro workstation, which facilitated the generation and semi-automatic analysis of VOIs (see Figure 2 (c)). The height of the VOI was 9 mm and was drawn as a cylindrical cross section, as large as possible without intersecting the vertebral cortex or including dense bone islands, venous plexuses, or local lesions. If a vertebral fracture occurred, it was replaced by an adjacent vertebra. The 50 and 150 keV single-energy images and MR DixonFF images generated by DLCT reconstruction were imported into the 3D Slicer software to outline the VOIs of the three vertebrae (see Figure 2 (a) and (b)), calculate HU(E 1 )、HU(E 2 ) and MR PDFF values. All VOI measurements were performed independently by two observers. DLCT L1-3 vBMD, QCT L1-3vBMD, and MR L1-3PDFF were measured simultaneously to ensure the consistency of VOI position and VOI size for each individual in all measurements. According to step S120, the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of each individual in the target population were obtained.

[0131] According to step S130, the double-layer spectral CT data of the target group at the lumbar vertebral body is used as sample data, and the bone density and bone marrow fat content of the target group are used as label data to train a three-substance model based on double-layer spectral CT. During the model training process, the densities of YM and RM are regarded as free parameters, and these parameters are adjusted so that the average volume of YM accounts for half of the total bone marrow volume, i.e., 0.50, and 95% of the measured values ​​(i.e., ±1.96SD) are between 0.25 and 0.75 (see Figure 3 A). The densities of YM and RM obtained were 0.924 g / cm 3 and 1.057 g / cm 3To further ensure the quality of the model, Figure 3 B shows a scatter plot of HU values ​​measured at 50 keV vs. 150 keV, with a diamond ABCD superimposed on it, which outlines the three-substance model at 50 and 200 mg / cm 3 Only the data points between line segments AD and BC provide reasonable solutions for the volume fractions of YM and RM (i.e., not negative and not greater than 1).

[0132] Specifically, Figure 3 A shows the volume fractions of bone matrix (BM), yellow bone marrow (YM), and red bone marrow (RM) of the 100 patients in the training set. The composition and density of BM can be obtained from the open source information repository. The mass density of YM and RM is 0.93 g / cm 3 and 1.065g / cm 3 , so that the mean adjusted YM volume fraction (=YM / [YM+RM]) was 0.50, ±1.96, and the standard deviation limits were 0.25 and 0.75, respectively. Figure 3 B shows the quality assurance graph of 50 and 150 keV single energy HU measurements. Diamond graph ABCD shows that the BM, YM and RM are 1.92, 0.93 and 1.065 g / cm 3 Predictive limits of physically feasible solutions of the three-material model for density and vBMD values ​​between 50 and 200 mg / cm 3 A physically feasible solution to the three-material model should have all three volume fractions (fBM, fYM, and fRM) in the range of 0 to 1, their sum should be 1, and all vBMDs between 50 and 200 mg / cm 3 The points between should lie within the diamond ABCD. A similar plot for the validation set is shown in Figure 5 A.

[0133] To ensure that the volume fractions of the three substances (BM, YM and RM) are reliable and in the range of 0 to 1, this example regards the mass densities of YM and RM as free parameters and fits the model to 100 subjects obtained from the training set.

[0134] Step 1: This example uses the YM and RM densities used in the open literature, which are 0.92 and 1.02 g / cm 2 Applying this to the three-substance model, we find that these densities lead to negative YM volume fractions for some subjects in the training set (see Figure 4 A), and the points on the 50 vs. 150 keV HU diagram are outside the predicted diamond ABCD (see Figure 4 B).

[0135] Specifically, Figure 4 A shows the volume fractions of bone matrix (BM), yellow marrow (YM), and red marrow (RM) of the 100 participants in the training set, where the mass densities of BM, YM, and RM were set to 1.92, 0.92, and 1.02 g / cm, respectively. 3 The mean adjusted YM volume fraction (=YM / [YM+RM]) was 0.13, ±1.96, and the standard deviation limits were -0.22 and 0.48, respectively. Figure 4 B shows the quality assurance diagram for the 50 and 150 keV single energy HU measurements. Diamond ABCD shows the predicted limits of physically feasible solutions to the three-material model for vBMD values ​​between 50 and 200 mg / cm 3 Between. Figure 4 The values ​​for the systematically low YM volume fraction shown in A are consistent with that of many individual vertebrae, with data points located outside the diamonds.

[0136] Step 2: As mentioned above, this example treats the densities of YM and RM as free parameters and corrects them to obtain an Adj-YM (YM volume fraction expressed as a fraction of total marrow volume) volume fraction with a mean of 0.50, a standard deviation of ±1.96, and upper and lower quartiles of 0.25 and 0.75 ( Figure 3 ), the densities of YM and RM are 0.93 and 1.065 g / cm 2 The density of RM was found to be about 4% higher than expected.

[0137] In order to investigate this difference, the RM density was set to 1.02 g / cm 3 , and adjust some other parameters to make them consistent with the mean, upper and lower limits of the Adj-YM volume fraction. This example studies three alternative ways to change the model parameters: (1) changing the single energy settings of 50 and 150 keV; (2) changing the chemical composition of the RM by adjusting its effective atomic number (Zeff) and electron density; (3) by changing the mass density of hydroxyapatite and collagen, the original values ​​of which are 3.06 and 1.38 g / cm 3 This example tests these solutions one by one, and believes that the RM obtained is the most ideal value, so as to ensure that the RM value substituted into the three-substance model is reliable and the results obtained by the model are ideal.

[0138] [1] Single energy keV setting. By changing the low energy keV value, this example found that the single energy of 55keV (the densities of YM and RM are 0.925 and 1.02 g / cm respectively) 3 ) and 50keV single energy (densities of YM and RM are 0.93 and 1.065g / cm3 ) has similar results to the Adj-YM distribution (see Figure 5 ). Increasing the low photon energy from 50 to 55 keV results in the sides AD and BC of the rhombus ABCD having a more gradual slope, closer to the slope of the points in the HU measurement diagram. However, it is not possible for the low single energy keV error to be so large. This example scans the European Spine Phantom (ESP) 10 times and plots the HU values ​​of the three vertebrae on the rhombus diagram (see Figure 5 ). Since ESP is composed of only two materials, hydroxyapatite and soft tissue equivalent resin, the three points of ESP should be located on a straight line parallel to the diamond sides AD and BC. Figure 5 It can be seen that the possibility of selecting 55keV low single energy is excluded. This embodiment also modifies 150keV high single energy, but it is found that Figure 3 The placement of the QA diamonds in the scatter plot is very insensitive to this change. This is because at higher photon energies, Compton scattering is the dominant attenuation process, and the attenuation coefficients of materials such as YM, RM, and collagen vary very little with photon energy. Therefore, their HU measurements change very little when the attenuation coefficient and photon energy vary.

[0139] Specifically, Figure 5 A shows the quality assurance diagram for the 50 and 150 keV single energy HU measurements. Diamonds ABCD show the predicted limits of physically feasible solutions to the three-material model for vBMD values ​​between 50 and 200 mg / cm 3 This is the data of 125 participants in the validation set (with Figure 3 The training set is similar to that in Figure B for comparison). Data points for three vertebrae of the European Spine Model (ESP) have been added to the figure. Figure 5 B shows diamond ABCD redrawn from the same data, assuming the single energies are 55 and 150 keV, respectively, and the densities of YM and RM are 0.925 and 1.02 g / cm, respectively. 3 . Please note that Figure 4 In contrast to B, all the subjects with vBMD values ​​between 50 and 250 mg / cm 3 However, this model was excluded because the soft tissue equivalent component of the ESP vertebrae changed with changes in vBMD.

[0140] [2] Elemental composition of RM. Another possible explanation for the higher-than-expected RM density is that the elemental composition of the RM is different from that specified in ICRU Publication 46. To examine this, the mass attenuation coefficient of the RM is expressed as a function of the effective atomic number (which determines the contribution from the photoelectric effect) and the electron density (which determines the contribution from Compton scattering). From the NIST mass attenuation coefficient, the Zeff values ​​of YM and RM are calculated to be 6.29 and 7.07, and the electron densities are 3.35 and 3.32 x 1023 electrons / g, respectively. In the scatter plot, the position of the QA diamond is very insensitive to the Zeff value of the RM, and a decrease in the Zeff value to 2.5 is required to increase the mean of the Adj-YM volume fraction from 0.13 to 0.50. Changing the electron density is a more effective approach, and increasing the RM from 3.32 to 3.50 x 1023 electrons / g (a 5% increase) produces a volume fraction distribution of Adj-YM that is almost identical to an increase in mass density from 1.02 to 1.065 g / cm 3 (4% increase) The ICRU publication 46 reports an average electron density of 3.33 for 71 different types of soft tissues, with minimum and maximum values ​​of 3.30 and 3.36 x 1023 electrons / g, respectively. Therefore, the three-matter model solution of 3.50 x 1023 electrons / g is clearly an extreme outlier.

[0141] [3] Mass density of hydroxyapatite and collagen. Finally, this example adjusts the mass density of hydroxyapatite and collagen to ensure that the density of BM is consistent with its composition by weight of 58% HA, 32% collagen, and 10% water. However, changing these densities has little effect on the Adj-YM volume fraction and vBMD results. This is because the volume fraction of BM is relatively small, approximately 10%. For example, using a higher density hydroxyapatite results in a slightly smaller volume fraction, and vBMD (vBMD = density x volume fraction) and Adj-YM remain almost unchanged.

[0142] The present invention provides a method for constructing a three-substance model based on double-layer spectral CT and a quantitative analysis system for bone density and bone marrow fat content. Based on double-layer spectral CT and combined with a three-substance decomposition method, a three-substance model based on double-layer spectral CT is trained, which can accurately and quantitatively analyze bone density and bone marrow fat content based on the HU value of the double-layer spectral CT image. Compared with traditional methods, the present invention can not only simultaneously measure lumbar vertebrae volume bone density and bone marrow fat content, but also significantly shorten the patient's examination time, reduce the burden of medical resources, and reduce the patient's economic cost. In addition, the present invention not only solves the problem that traditional imaging methods are difficult to accurately measure bone marrow fat content, but also helps to better understand the role of bone marrow fat in bone metabolic diseases, and provide a reliable reference for clinical diagnosis and treatment.

[0143] The bone density and bone marrow fat content quantitative analysis system based on double-layer spectral CT provided by the present invention is described below.

[0144] Reference Figure 6 The present invention provides a quantitative analysis system for bone density and bone marrow fat content based on double-layer spectral CT, which may include:

[0145] A data receiving module, used for: receiving double-layer spectral CT data of a subject at a lumbar vertebral body uploaded by at least one client, wherein the type of the double-layer spectral CT data of the subject at a lumbar vertebral body is consistent with the type of the double-layer spectral CT data of the target group at a lumbar vertebral body in any of the above-mentioned methods for constructing a three-substance model based on double-layer spectral CT, and the subject can be a group of people other than those with severe scoliosis, multiple spinal fractures, multiple bone tumors, and multiple metal implants;

[0146] A quantitative analysis module, for predicting the bone density and bone marrow fat content of the subject according to the double-layer spectral CT data of the subject at the lumbar vertebral body, using a three-substance model based on double-layer spectral CT obtained by the construction method of a three-substance model based on double-layer spectral CT according to any one of the above-mentioned items;

[0147] The data output module is used to send the bone density and bone marrow fat content of the subject to be tested back to at least one client.

[0148] Specifically, before using the bone density and bone marrow fat content quantitative analysis system based on double-layer spectral CT, the double-layer spectral CT data obtained by the subject undergoing DLCT can be obtained first, and then the double-layer spectral CT data of the subject at the lumbar vertebral body can be uploaded to the bone density and bone marrow fat content quantitative analysis system based on double-layer spectral CT through the client, received by the data receiving module, and then predicted by the quantitative analysis module using the three-substance model based on double-layer spectral CT to obtain the subject's bone density and bone marrow fat content, and then sent back to at least one client through the data output module. This helps to quantitatively analyze the subject's bone density and bone marrow fat content with high precision, and provide a reliable reference for the subject's clinical diagnosis and treatment.

[0149] Figure 7 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 7As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to perform the following steps:

[0150] Receiving double-layer spectral CT data of a subject at a lumbar vertebral body uploaded by at least one client, wherein the type of the double-layer spectral CT data of the subject at a lumbar vertebral body is consistent with the type of double-layer spectral CT data of a target group at a lumbar vertebral body in any of the above-mentioned methods for constructing a three-substance model based on double-layer spectral CT, and the subject is an individual outside a preset filtered-out group, and the preset filtered-out group includes a group with scoliosis, multiple spinal fractures, multiple bone tumors, and multiple metal implants;

[0151] According to the double-layer spectral CT data of the subject at the lumbar vertebral body, the bone density and bone marrow fat content of the subject are predicted by using the three-substance model based on double-layer spectral CT obtained by the construction method of the three-substance model based on double-layer spectral CT according to any of the above items;

[0152] The bone density and bone marrow fat content of the subject to be tested are sent back to at least one client.

[0153] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0154] On the other hand, the present invention further provides a computer program product, the computer program product comprising a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and when the computer program is executed by a processor, the computer can perform the following steps:

[0155] Receiving double-layer spectral CT data of a subject at a lumbar vertebral body uploaded by at least one client, wherein the type of the double-layer spectral CT data of the subject at a lumbar vertebral body is consistent with the type of double-layer spectral CT data of a target group at a lumbar vertebral body in any of the above-mentioned methods for constructing a three-substance model based on double-layer spectral CT, and the subject is an individual outside a preset filtered-out group, and the preset filtered-out group includes a group with scoliosis, multiple spinal fractures, multiple bone tumors, and multiple metal implants;

[0156] According to the double-layer spectral CT data of the subject at the lumbar vertebral body, the bone density and bone marrow fat content of the subject are predicted by using the three-substance model based on double-layer spectral CT obtained by the construction method of the three-substance model based on double-layer spectral CT according to any of the above items;

[0157] The bone density and bone marrow fat content of the subject to be tested are sent back to at least one client.

[0158] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented to perform the following steps when executed by a processor:

[0159] Receiving double-layer spectral CT data of a subject at a lumbar vertebral body uploaded by at least one client, wherein the type of the double-layer spectral CT data of the subject at a lumbar vertebral body is consistent with the type of double-layer spectral CT data of a target group at a lumbar vertebral body in any of the above-mentioned methods for constructing a three-substance model based on double-layer spectral CT, and the subject is an individual outside a preset filtered-out group, and the preset filtered-out group includes a group with scoliosis, multiple spinal fractures, multiple bone tumors, and multiple metal implants;

[0160] According to the double-layer spectral CT data of the subject at the lumbar vertebral body, the bone density and bone marrow fat content of the subject are predicted by using the three-substance model based on double-layer spectral CT obtained by the construction method of the three-substance model based on double-layer spectral CT according to any of the above items;

[0161] The bone density and bone marrow fat content of the subject to be tested are sent back to at least one client.

[0162] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0163] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a three-substance model based on double-layer spectral CT, characterized in that: include: Obtaining double-layer spectral CT data of the target group located at the lumbar vertebral body, wherein the target group includes an osteoporosis patient group and a healthy group; According to the double-layer spectral CT data of the target group located at the lumbar vertebral body, the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group were obtained by using the three-substance decomposition method, wherein the yellow bone marrow volume fraction represents the bone marrow fat content; According to the bone matrix volume fraction of the target group, the bone density of the target group is obtained; The double-layer spectral CT data of the target group at the lumbar vertebrae were used as sample data, and the bone density and bone marrow fat content of the target group were used as label data to train a three-substance model based on double-layer spectral CT. The double-layer spectral CT data of the target group located at the lumbar vertebral body includes the HU(E1) value and the HU(E2) value of the monoenergetic image measured at energy E1 and energy E2 of the lumbar vertebral body of the target group by double-layer spectral CT. The bone matrix volume fraction, the yellow bone marrow volume fraction, and the red bone marrow volume fraction of the target group are obtained by three-substance decomposition method based on the double-layer spectral CT data of the target group located at the lumbar vertebral body, wherein the yellow bone marrow volume fraction represents the bone marrow fat content, including: Obtain the water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient of the target group under energy E1 and energy E2, set water density, bone matrix density, yellow bone marrow density, and red bone marrow density as variables, and set the sum of the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group to 1; According to the water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient of the target group at energy E1 and energy E2, combined with water density, bone matrix density, yellow bone marrow density, and red bone marrow density, the water linear attenuation coefficient, bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient of the target group at energy E1 and energy E2 are obtained; According to the HU(E1) value and HU(E2) value of the single energy image measured at energy E1 and energy E2 of the target group, combined with the water linear attenuation coefficient of the target group at energy E1 and energy E2, the linear attenuation coefficients μ(E1) and μ(E2) of the target group at energy E1 and energy E2 are obtained; According to the linear attenuation coefficients μ(E1) and μ(E2) of the target group at energy E1 and energy E2, combined with the bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient of the target group at energy E1 and energy E2, the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group are obtained.

2. The method for constructing a three-substance model based on double-layer spectral CT according to claim 1, characterized in that: The water linear attenuation coefficient, bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient of the target group under energy E1 and energy E2 are obtained by combining water density, bone matrix density, yellow bone marrow density, and red bone marrow density, including: According to the water mass attenuation coefficient, bone matrix mass attenuation coefficient, yellow bone marrow mass attenuation coefficient, and red bone marrow mass attenuation coefficient of the target group at energy E1 and energy E2, combined with water density, bone matrix density, yellow bone marrow density, and red bone marrow density, the first expression is used to obtain the water linear attenuation coefficient, bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient of the target group at energy E1 and energy E2, wherein the first expression is: μ water (E1) = μ m1(water) × water density, μ water (E2) = μ m2(water) × water density, μ BM (E1) = μ m1(BM) × the density of BM, μ BM (E2) = μ m2(BM) × the density of BM, μ YM (E1) = μ m1(YM) ×YM density, μ YM (E2) = μ m2(YM) ×YM density, μ RM (E1) = μ m1(RM) × the density of RM, μ RM (E2) = μ m2(RM) × the density of RM, In the first expression, μ water (E1) and μ water (E2) represents the water linear attenuation coefficient of the target group at energy E1 and energy E2, μ m1(water) and μ m2(water) represents the water mass attenuation coefficient of the target group under energy E1 and energy E2, the density of water represents the density of water, μ BM (E1) and μ BM (E2) represents the linear attenuation coefficient of bone matrix of the target group at energy E1 and energy E2, μ m1(BM) and μ m2(BM) represents the bone matrix mass attenuation coefficient of the target group under energy E1 and energy E2, the density of BM represents the density of bone matrix, μ YM (E1) and μ YM (E2) represents the linear attenuation coefficient of the yellow bone marrow of the target group at energy E1 and energy E2, μ m1(YM) and μ m2(YM) represents the yellow bone marrow mass attenuation coefficient of the target group under energy E1 and energy E2, the density of YM represents the density of yellow bone marrow, μ RM (E1) and μ RM (E2) represents the linear attenuation coefficient of the red bone marrow of the target group at energy E1 and energy E2, μ m1(RM) and μ m2(RM) It represents the red bone marrow mass attenuation coefficient of the target group under energy E1 and energy E2, and the density of RM represents the density of red bone marrow.

3. The method for constructing a three-substance model based on double-layer spectral CT according to claim 2, characterized in that: The method of obtaining the linear attenuation coefficients μ(E1) and μ(E2) of the target group at energy E1 and energy E2 by combining the HU(E1) value and HU(E2) value of the monoenergetic image measured at energy E1 and energy E2 of the target group with the water linear attenuation coefficient of the target group at energy E1 and energy E2 comprises: According to the HU(E1) value and HU(E2) value of the single energy image measured at energy E1 and energy E2 of the target group, combined with the water linear attenuation coefficient of the target group at energy E1 and energy E2, the linear attenuation coefficients μ(E1) and μ(E2) of the target group at energy E1 and energy E2 are obtained using the second expression, where the second expression is: In the second expression, μ(E1) and μ(E2) represent the linear attenuation coefficients of the target group at energy E1 and energy E2, HU(E1) value and HU(E2) value represent the HU values ​​of the single-energy image measured at energy E1 and energy E2 for the target group, and μ water (E1) and μ water (E2) represents the water linear attenuation coefficient of the target group at energy E1 and energy E2.

4. The method for constructing a three-substance model based on double-layer spectral CT according to claim 3, characterized in that: The method of obtaining the bone matrix volume fraction, yellow bone marrow volume fraction, and red bone marrow volume fraction of the target group according to the linear attenuation coefficients μ(E1) and μ(E2) of the target group at energy E1 and energy E2, combined with the bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient of the target group at energy E1 and energy E2, includes: According to the linear attenuation coefficients μ(E1) and μ(E2) of the target group at energy E1 and energy E2, combined with the bone matrix linear attenuation coefficient, yellow bone marrow linear attenuation coefficient, and red bone marrow linear attenuation coefficient of the target group at energy E1 and energy E2, a third expression is used to obtain multiple quantitative determination coefficients; According to the plurality of quantitative determination coefficients, using the fourth expression, the bone matrix volume fraction, the yellow bone marrow volume fraction, and the red bone marrow volume fraction of the target group are obtained; Among them, the third expression is: D1=μ YM (E1)*m RM (E2)-m RM (E1)*μ(E2)-μ(E1)*μ YM (E2)+μ(E1)*μ RM (E2)+ μ(E2)*μ YM (E1)+μ RM (E1)*m YM (E2), D2=μ BM (E1)*m RM (E2)-m RM (E1)*μ(E2)-μ(E1)*μ BM (E2)+μ(E1)*μ RM (E2)+ μ(E2)*μ BM (E1)+μ RM (E1)*m BM (E2), D3=μ BM (E1)*m YM (E2)-m YM (E1)*μ(E2)-μ(E1)*μ BM (E2)+μ(E1)*μ YM (E2)+ μ(E2)*μ BM (E1)+μ YM (E1)*m BM (E2), D4=μ BM (E1)*m YM (E2)+μ YM (E1)*m BM (E2)+μ RM (E1)*m BM (E2)-m RM (E1)*m YM (E2)-m YM (E1)*m BM (E2)-m BM (E1)*m RM (E2), In the third expression, D1, D2, D3, and D4 represent multiple quantitative determination coefficients, μ(E1) and μ(E2) represent the linear attenuation coefficients of the target group at energy E1 and energy E2, and μ YM (E1) and μ YM (E2) represents the linear attenuation coefficient of bone matrix of the target group at energy E1 and energy E2, μ BM (E1) and μ BM (E2) represents the linear attenuation coefficient of bone matrix of the target group at energy E1 and energy E2, μ RM (E1) and μ RM (E2) represents the linear attenuation coefficient of red bone marrow of the target group at energy E1 and energy E2; The fourth expression is: f1+f2+f3=1, In the fourth expression, f1 represents the bone matrix volume fraction of the target population, f2 represents the yellow bone marrow volume fraction of the target population, and f3 represents the red bone marrow volume fraction of the target population.

5. The method for constructing a three-substance model based on double-layer spectral CT according to claim 4, characterized in that: The step of obtaining the bone density of the target group according to the bone matrix volume fraction of the target group includes: According to the bone matrix volume fraction of the target group, the bone density of the target group is obtained using the bone density expression, where the bone density expression is: vBMD=f1×f HA × density of HA × 1000, In the bone density expression, vBMD represents the bone density of the target group, f1 represents the bone matrix volume fraction of the target group, and f HA represents the known volume fraction of hydroxyapatite, and the density of HA represents the known density of hydroxyapatite.

6. The method for constructing a three-substance model based on double-layer spectral CT according to claim 5, characterized in that: The double-layer spectral CT data of the target group at the lumbar vertebral body is used as sample data, and the bone density and bone marrow fat content of the target group are used as label data to train a three-substance model based on the double-layer spectral CT, including: The double-layer spectral CT data of the target group at the lumbar vertebrae were used as sample data, and the bone density and bone marrow fat content of the target group were used as label data to train a three-substance model based on double-layer spectral CT. During the training process, the density of YM and the density of RM were regarded as free parameters, and the free parameters were adjusted so that the average volume of YM accounted for half of the total bone marrow volume. The three-substance model based on double-layer spectral CT was trained, in which the total bone marrow volume was the sum of the yellow bone marrow volume and the red bone marrow volume.

7. A quantitative analysis system of bone density and bone marrow fat content based on double-layer spectral CT, characterized in that: include: A data receiving module, used for: receiving double-layer spectral CT data of a subject at a lumbar vertebral body uploaded by at least one client, wherein the type of the double-layer spectral CT data of the subject at a lumbar vertebral body is consistent with the type of double-layer spectral CT data of a target group at a lumbar vertebral body in the method for constructing a three-substance model based on double-layer spectral CT according to any one of claims 1 to 6, and the subject is an individual outside a preset filtered-out group, and the preset filtered-out group includes a group with scoliosis, multiple spinal fractures, multiple bone tumors, and multiple metal implants; A quantitative analysis module, for predicting the bone density and bone marrow fat content of the subject according to the double-layer spectral CT data of the subject at the lumbar vertebral body, using a three-substance model based on double-layer spectral CT obtained by the method for constructing a three-substance model based on double-layer spectral CT according to any one of claims 1 to 6; The data output module is used to send the bone density and bone marrow fat content of the subject to be tested back to at least one client.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for constructing a three-substance model based on double-layer spectral CT according to any one of claims 1 to 6 is implemented, or when the processor executes the program, the following steps are implemented: receiving double-layer spectral CT data of a subject at a lumbar vertebral body uploaded by at least one client, wherein the type of the double-layer spectral CT data of the subject at a lumbar vertebral body is consistent with the type of the double-layer spectral CT data of the target group at a lumbar vertebral body in the method for constructing a three-substance model based on double-layer spectral CT according to any one of claims 1 to 6, and the subject is an individual outside a preset filtered-out group, and the preset filtered-out group includes a group with scoliosis, multiple spinal fractures, multiple bone tumors, and multiple metal implants; According to the double-layer spectral CT data of the subject at the lumbar vertebral body, the bone density and bone marrow fat content of the subject are predicted by using the three-substance model based on double-layer spectral CT obtained by the method for constructing a three-substance model based on double-layer spectral CT according to any one of claims 1 to 6; The bone density and bone marrow fat content of the subject to be tested are sent back to at least one client.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for constructing a three-substance model based on double-layer spectral CT according to any one of claims 1 to 6 is implemented, or when the computer program is executed by a processor, the following steps are implemented: receiving double-layer spectral CT data of a subject at a lumbar vertebral body uploaded by at least one client, wherein the type of the double-layer spectral CT data of the subject at a lumbar vertebral body is consistent with the type of the double-layer spectral CT data of the target group at a lumbar vertebral body in the method for constructing a three-substance model based on double-layer spectral CT according to any one of claims 1 to 6, and the subject is an individual outside a preset filtered-out group, and the preset filtered-out group includes a group with scoliosis, multiple spinal fractures, multiple bone tumors, and multiple metal implants; According to the double-layer spectral CT data of the subject at the lumbar vertebral body, the bone density and bone marrow fat content of the subject are predicted by using the three-substance model based on double-layer spectral CT obtained by the method for constructing a three-substance model based on double-layer spectral CT according to any one of claims 1 to 6; The bone density and bone marrow fat content of the subject to be tested are sent back to at least one client.

Images