Carotid plaque and vascular inflammation assessment method and system

Abstract

The invention belongs to the technical field of medical care informatics, and particularly relates to a carotid plaque and vascularization assessment method and system based on FF and R2 * atlas, T1W-TSE and FF / R2 * atlas are generated through single MRI scanning, the vascularization and plaque components represented by PVAT can be quantitatively assessed at the same time, such as LRNC (through the FF atlas), calcified plaque (through the R2 * atlas) and IPH staging (through T1W-TSE / R2 * fusion), and the assessment accuracy is high. The defects that traditional MRI qualitative analysis cannot distinguish IPH periods are overcome; and moreover, the inherent T1 / T2 * deviation in chemical shift imaging is eliminated by the Dixon-based FF / R2 * calculation, and the quantitative analysis precision is improved.

Description

Technical Field

[0001] The present invention belongs to the field of medical information technology, and specifically relates to a method based on FF and R2 * Atlas-based carotid artery plaque and vascular inflammation assessment method and system. Background Art

[0002] Carotid atherosclerosis is the second leading cause of recurrent ischemic stroke. Pathological studies have confirmed that vulnerable plaques characterized by intraplaque hemorrhage (IPH), a lipid-rich necrotic core (LRNC), and calcification are strong predictors of downstream ischemic stroke. Furthermore, proinflammatory cytokines released by inflamed vessels disrupt perivascular adipose tissue (PVAT) homeostasis, further exacerbating plaque vulnerability and being associated with adverse clinical outcomes. Current analysis using carotid plaque imaging technology has the following problems:

[0003] (1) Insufficient quantification of plaque composition analysis. For example, conventional high-resolution magnetic resonance imaging (MRI) relies on morphological features (such as signal intensity) to qualitatively analyze plaque composition. It cannot accurately quantify plaque components such as LRNC and IPH, and there has been no attempt to characterize PVAT.

[0004] (2) Lack of collaborative analysis of multimodal data. Existing quantitative magnetic resonance imaging atlas technologies often use isolated parameter analysis such as T1 or T2 mapping, without combining them with collaborative analysis of multimodal data, resulting in an incomplete comprehensive assessment of plaque composition. In addition, existing systems do not integrate inflammatory indicators such as PVAT with the spatial distribution of plaque components, making it difficult to assess the impact of the inflammatory microenvironment on plaque stability.

[0005] Therefore, establishing an accurate assessment system that can simultaneously quantify plaque composition and vascular inflammation has important clinical value for implementing targeted anti-inflammatory treatment and formulating individualized stroke prevention and treatment plans. Summary of the Invention

[0006] The present invention provides a novel method based on FF and R2 * Atlas-based carotid artery plaque and vascular inflammation assessment method and system.

[0007] The technical solutions of the present invention are as follows:

[0008] The present invention provides a method based on FF and R2 * The atlas' carotid plaque and vascular inflammation assessment methods include:

[0009] S1: Imaging of the target area of ​​the carotid artery based on a three-dimensional six-echo Dixon sequence, and the resulting fat fraction map and R2 * Relaxation spectra, recorded as FF spectra, R2 * Atlas;

[0010] S2: For each FF atlas layer, after identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial plane, an adaptive ellipse is generated along the outer wall of the vessel based on the morphologically constrained ellipse growing algorithm, which serves as the inner boundary. The ellipse automatically expanded outward based on the inner boundary serves as the outer boundary.

[0011] Calculate the total number of pixels within the inner boundary whose fat fraction is not less than the first value to obtain the volume of the fat-rich necrotic core;

[0012] Calculate the average fat fraction of all pixels in the annular area between the inner and outer boundaries whose fat fraction is not less than the first value as the fat attenuation index;

[0013] S3: Imaging of the target area of ​​the carotid artery based on T1-weighted turbo spin echo sequence, combined with R2 * Spectrum, calculated T1-weighted turbo spin echo sequence is greater than the first preset signal and is less than R2 * After the second preset signal area, the volume of acute intraplaque hemorrhage is obtained by multi-slice T1-weighted turbo spin echo imaging;

[0014] S4: For each layer R2 * After identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial position, an elliptical area that fits the outer wall of the blood vessels was delineated and R2 within the elliptical area was calculated. * The total pixel value of which the value is greater than the second value is used to obtain the volume of the calcified plaque;

[0015] S5: Calculate the risk score according to the formula: risk score = 0.2 × volume of lipid-rich necrotic core + 0.5 × volume of acute intraplaque hemorrhage + 0.1 × volume of calcified plaque + 0.2 / fat attenuation index for assessment.

[0016] The step S2 further includes calculating the total number of pixels in the annular region between the inner and outer boundaries whose fat fraction is not less than a first value as the area of ​​perivascular adipose tissue, and obtaining the volume of the perivascular adipose tissue from the multi-layer FF atlas.

[0017] Before S2, it also included:

[0018] The carotid bifurcation level was used as the central level, and 7 levels above and below the central level were included, totaling 15 levels to cover the 1.5 cm area above and below the carotid bifurcation for imaging.

[0019] S2 calculates the total number of pixels within the inner boundary whose fat fraction is not less than the first value, and obtains the volume of the fat-rich necrotic core, specifically:

[0020] The total pixel value within the inner boundary whose fat fraction is not less than the first value is taken as the area of ​​the fat-rich necrotic core, and the volume of the fat-rich necrotic core is obtained from the multi-layer FF atlas.

[0021] S4 calculates R2 within the elliptical area * The total pixel value whose value is greater than the second value is used to obtain the volume of the calcified plaque, specifically:

[0022] Place R2 in the elliptical area * The total pixel value with a value greater than the second value is regarded as the area of ​​the calcified plaque, which is obtained by multi-layer R2 * The volume of calcified plaque was obtained by mapping.

[0023] After calculating the total number of pixels within the inner boundary whose fat fraction is not less than the first value, S2 further includes extracting the fat fraction value and constructing a three-dimensional model of the fat-rich necrotic core for visual display;

[0024] S4 calculates R2 within the elliptical area * After the total pixel value is greater than the second value, it also includes extracting R2 * The three-dimensional model of calcified plaque was constructed for visual display.

[0025] After calculating the total pixel value in the annular area between the inner and outer boundaries where the fat fraction is not less than the first value, the method further includes extracting the fat fraction value and constructing a three-dimensional model of perivascular fat tissue for visual display.

[0026] The present invention also provides a method based on FF and R2 * Atlas' carotid plaque and vascular inflammation assessment system includes:

[0027] Imaging module: Based on the three-dimensional six-echo Dixon sequence, the target area of ​​the carotid artery is imaged, and the fat fraction map and R2 are obtained. * Relaxation spectra, recorded as FF spectra, R2 * Atlas;

[0028] Fat-rich necrotic core and fat attenuation index quantification module: For each FF atlas layer, the outer wall contours of the bilateral internal carotid arteries in the transverse axial plane are identified. An adaptive ellipse is generated along the outer wall of the vessel using a morphologically constrained ellipse growing algorithm, which serves as the inner boundary. The ellipse automatically expanded outward based on the inner boundary serves as the outer boundary.

[0029] Calculate the total number of pixels within the inner boundary whose fat fraction is not less than the first value to obtain the area of ​​the fat-rich necrotic core, and obtain the volume of the fat-rich necrotic core from the multi-layer FF atlas;

[0030] Calculate the average fat fraction of all pixels in the annular area between the inner and outer boundaries whose fat fraction is not less than the first value as the fat attenuation index;

[0031] Acute intraplaque hemorrhage quantification module: imaging of the target area of ​​the carotid artery based on T1-weighted turbo spin echo sequence, combined with R2 * Spectrum, calculated T1-weighted turbo spin echo sequence is greater than the first preset signal and is less than R2 * After the second preset signal area, the volume of acute intraplaque hemorrhage is obtained by multi-slice T1-weighted turbo spin echo imaging;

[0032] Calcified plaque quantification module: for each layer R2 * After identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial position, an elliptical area that fits the outer wall of the blood vessels was delineated and R2 within the elliptical area was calculated. * The total pixel value with a value greater than the second value is used to obtain the area of ​​the calcified plaque, and the volume of the calcified plaque is obtained from the multi-layer R2* map;

[0033] Evaluation module: used to calculate the risk score and perform evaluation according to the formula: risk score = 0.2×volume of fat-rich necrotic core + 0.5×volume of acute intraplaque hemorrhage + 0.1×volume of calcified plaque + 0.2 / fat attenuation index.

[0034] Beneficial effects

[0035] The present invention generates T1W-TSE and FF / R2 by a single MRI scan * The atlas can simultaneously quantitatively evaluate vascular inflammation and plaque components characterized by PVAT, such as LRNC (through FF atlas), calcified plaque (through R2 * atlas) and IPH staging (by T1W-TSE / R2 * fusion), overcomes the shortcomings of traditional MRI qualitative analysis and inability to distinguish IPH stages; and based on Dixon's FF / R2 * Computationally eliminates the inherent T1 / T2 difference in chemical shift imaging * deviation and improve the accuracy of quantitative analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A is a schematic diagram of the segmentation of perivascular adipose tissue (PVAT) and lipid-rich necrotic core (LRNC), and B is a schematic diagram of the segmentation of calcified plaque.

[0037] Figure 2 Flowchart of the validation test.

[0038] Figure 3Figure 2 shows the multiparametric imaging features of an adult patient with asymptomatic right carotid atherosclerosis, where A is T1-weighted turbo spin echo (T1W-TSE) imaging, B is fat fraction (FF) atlas, C is color-coded FF atlas, D is CT angiography (CTA) axial image, and E is R2 * Relaxation spectrum, F is the color-coded R2 * Relaxation spectrum.

[0039] Figure 4 Figure 2 shows the multiparametric imaging features of an adult patient with symptomatic left carotid artery stenosis, where A is an axial T1-weighted turbo spin echo (T1W-TSE) image, B is a fat fraction (FF) map, C is a color-coded FF map, D is an axial CT angiography (CTA) image, and E is an R2 * Relaxation spectrum, F is the color-coded R2 * Relaxation spectrum.

[0040] Figure 5 A is the correlation analysis between the FF value of PVAT in the FF atlas and the Hounsfield unit value of the CTA axial image, and B is R2 * Correlation analysis between the calcification area of ​​relaxation maps and the calcification area of ​​CTA.

[0041] Figure 6 A visual diagram. DETAILED DESCRIPTION

[0042] The following examples are intended to illustrate the present invention rather than to further limit the present invention.

[0043] The present invention provides a method based on FF and R2 * The atlas' carotid plaque and vascular inflammation assessment methods include:

[0044] S1: Imaging of the target area of ​​the carotid artery based on a three-dimensional six-echo Dixon sequence, and the resulting fat fraction map and R2 * Relaxation spectra, recorded as FF spectra, R2 * Atlas.

[0045] In order to achieve more accurate coverage, a high-field magnetic resonance imaging system (Siemens MAGNETOM Vida 3.0T MRI) with a dedicated mode for vascular wall imaging was used in actual operation. A 64-channel head and neck combined phased array coil was used. The imaging protocol used a three-dimensional six-echo Dixon sequence as a quantitative sequence to simultaneously obtain fat fraction (FF) and R2 *Relaxation rate dual-parameter atlas. Key acquisition parameters have been carefully designed to achieve a high spatial / temporal resolution balance, including: repetition time (TR) = 8.9 ms; echo time (TE) = 1.33 / 2.56 / 3.79 / 5.02 / 6.25 / 7.48 ms (ΔTE = 1.23 ms); field of view (FOV) = 300×300 mm 2 ;Average=12;Matrix=256×256;Bandwidth=1150Hz / pixel;Acquisition resolution

[0046] =0.75×0.75mm 2 , reconstructed to 0.6×0.6mm 2 Slice thickness = 2 mm; Number of slices = 36 (this sequence generates 36 slices per scan, with the target number of slices subsequently analyzed, for example, only 15 of these slices); Flip angle = 4°; Total scan time = 5:14 minutes. Water-fat offset is maximized at approximately 0.4 pixels. A 72-mm saturation zone located below the imaging volume effectively suppresses inflow artifacts from circulating blood.

[0047] The final processing generates five parameters: 1) water phase; 2) lipid phase; 3) FF spectrum (FF‰); 4) T2 * Relaxation spectrum; 5) R2 * Relaxation spectrum (R2 * =1 / T2 * ). Among them, FF spectrum and R2 * The relaxation spectrum was used for subsequent analysis.

[0048] S2: For each FF atlas layer, after identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial plane, an adaptive ellipse is generated along the outer wall of the vessel based on the morphologically constrained ellipse growing algorithm, which serves as the inner boundary. The ellipse automatically expanded outward based on the inner boundary serves as the outer boundary.

[0049] Calculating the total number of pixels within the inner boundary whose fat fraction is not less than a first value (e.g., 10%) to obtain the volume of the lipid-rich necrotic core (LRNC);

[0050] The average fat fraction of all pixels with a fat fraction not less than 10% in the annular area between the inner and outer boundaries was calculated as the fat attenuation index (FAI).

[0051] Before data analysis, the multi-layer images generated by S1 were screened. Preferably, the carotid bifurcation level was used as the central level, and 7 levels above and below the central level were included, totaling 15 levels to cover the 1.5 cm area above and below the carotid bifurcation for analysis. This range includes the common distribution area of ​​carotid plaques.

[0052] Preferably, the S2 calculates the total pixel value within the inner boundary with a fat fraction not less than 10% to obtain the volume of the fat-rich necrotic core (LRNC), specifically:

[0053] The total pixel value with a fat fraction of not less than 10% within the inner boundary was taken as the area of ​​the fat-rich necrotic core, and the volume of the fat-rich necrotic core (LRNC) was obtained from the multi-layer FF atlas.

[0054] Furthermore, after calculating the total pixel value of the fat fraction within the inner boundary that is not less than 10%, S2 further includes extracting the fat fraction value and constructing a three-dimensional model of the fat-rich necrotic core for visual display.

[0055] In addition, the S2 also includes: calculating the total pixel value with a fat fraction of not less than 10% in the annular area between the inner and outer boundaries as the area of ​​perivascular adipose tissue, and obtaining the volume of perivascular adipose tissue (PVAT) from the multi-layer FF atlas.

[0056] Furthermore, after calculating the total pixel value of the fat fraction not less than 10% in the annular area between the inner and outer boundaries, the method further includes extracting the fat fraction value and constructing a three-dimensional model of perivascular fat tissue for visual display.

[0057] Regarding the use of FF atlas, the feature extraction of PVAT and LRNC and the construction of 3D models are achieved through the following process. The specific implementation process is shown in Matlab code Step 2-5 in Table 1.

[0058] 1) Retrieve the FF atlas sequence (each sequence contains 36 layers of images);

[0059] 2) In each FF atlas image, the outer wall contours of the bilateral internal carotid arteries in the transverse axial position were identified, and the region of interest (ROI) was segmented using a self-designed semi-automatic morphologically constrained ellipse growing algorithm to extract lipids in the internal carotid artery wall and PVAT features around the internal carotid artery. Figure 1 As shown in A, the inner boundary is a manually drawn adaptive ellipse generated along the outer wall of the blood vessel ( Figure 1 Middle A, red) [code: haz1 (right), hax4 (left)], the outer boundary is an ellipse that automatically expands 5mm based on the inner boundary ( Figure 1 Middle A, green) [coded: haz2 (right), hax5 (left)];

[0060] The application first designed a semi-automatically generated ellipse that fits the outer contour of the blood vessel wall as the inner boundary to segment the lipid-rich necrotic core from the perivascular fat. Second, an automatically generated ellipse that extends radially outward by 5 mm from the inner boundary was designed as the outer boundary to segment the perivascular fat from the distal adipose tissue. Because the perivascular adipose tissue is continuous with the surrounding tissue, these two steps manually defined the inner and outer boundaries of the perivascular fat to achieve vascular adipose tissue segmentation.

[0061] The radial expansion of 5 mm outward along the inner boundary is determined by developing the perivascular tissue through FF atlas. The present study found that the 5 mm expansion outward along the outer wall of the blood vessel can include the perivascular adipose tissue to the greatest extent and exclude the distant adipose tissue.

[0062] 3) Generate bilateral masks based on the inner and outer boundaries [inner boundary: BW1 (right), BW4 (left); outer boundary: BW2 (right), BW5 (left); annular ROI between the inner and outer boundaries: BW3 = BW2-BW1 (right), BW6 = BW5-BW4 (left)];

[0063] 4) LRNC feature extraction: The total number of pixels within the inner boundary [BW1 (right), BW4 (left)] with FF threshold ≥ 100‰ (10%) [counter-FF-ROI(1,i) (right); counter-FF-ROI(4,i), (left)] was used to quantify the LRNC area within the internal carotid artery wall on both sides of each layer ( Figure 1 Middle A);

[0064] 5) LRNC 3D model construction: extract the FF values ​​of all pixels within the inner boundary with a FF threshold ≥ 100‰ (10%) and construct the 3D model [img3D1new (right); img3D4new (left)];

[0065] 6) PVAT feature extraction: All pixels within the annular ROI between the inner and outer boundaries [BW3 (right), BW6 (left)] with a FF threshold ≥100‰ (10%) [counter-FF-ROI(3,i) (right); counter-FF-ROI(6,i), (left)] are identified as the PVAT area; the FAI of the PVAT is calculated by the average FF value of all pixels above the threshold within the annular ROI [Mean-ROI3-value (right); Mean-ROI6-value (left)].

[0066] 7) PVAT 3D model construction: The FF values ​​of all pixels with a FF threshold ≥ 100‰ (10%) within the annular ROI were extracted to construct the PVAT 3D model [img3D3new (right); img3D6new (left)].

[0067] Table 1 Matlab code

[0068]

[0069]

[0070]

[0071]

[0072] S3: Imaging of the target area of ​​the carotid artery based on T1-weighted turbo spin echo sequence, combined with R2 * Atlas, calculate the T1-weighted turbo spin echo sequence greater than the first preset signal (such as greater than 1.5 times the muscle signal) and is less than R2 * The second preset signal (such as <100s -1 ), that is, the area of ​​the region with high signal in T1-weighted turbo spin echo sequence and low signal in R2*, and the volume of acute intraplaque hemorrhage (IPH) is obtained by multi-slice T1-weighted turbo spin echo imaging.

[0073] In addition to the S1 three-dimensional six-echo Dixon sequence, the present invention also uses a lipid-saturated 2D black blood T1-weighted turbo spin-echo (T1W-TSE) sequence as a supplementary sequence for anatomical localization of the carotid artery wall and detection of acute intraplaque hemorrhage (acute IPH). The configuration is: TE = 19 milliseconds; TR = 1010 milliseconds; bandwidth = 264 Hz / pixel; acquisition resolution 0.66 × 0.66 mm 2 , reconstructed to 0.4×0.4mm 2 ; Slice thickness = 2 mm; Number of slices = 24; Flip angle = 120°; Total scan time = 5:27 min.

[0074] Since R2 * The value is linearly related to the iron and calcium content in the tissue, so this technology is used to quantitatively analyze chronic IPH. According to the previous diagnostic criteria, T1W-TSE showed high signal and R2 * Plaque components with low signal intensity on the atlas were classified as acute IPH, while those with high signal intensity on both modalities were classified as chronic IPH. * Chronic IPH was excluded from the analysis.

[0075] The dual sequence design of the present invention has the following advantages:

[0076] 1) Temporal synchronization: spatial registration of anatomical sequences (T1W-TSE) and quantitative sequences (Dixon) is achieved through the scan planner;

[0077] 2) Parameter complementation: T1W-TSE provides morphological benchmarks (vascular wall and plaque components), and Dixon sequence provides metabolic quantitative data (characterization of PVAT and LRNC by FF atlas; characterization of PVAT and LRNC by R2 * Atlas for IPH staging and calcification burden measurement);

[0078] 3) Artifact suppression: A shared motion correction module is used to eliminate the effects of patient displacement.

[0079] S4: For each layer R2 * After identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial position, an elliptical area that fits the outer wall of the blood vessels was delineated and R2 within the elliptical area was calculated. * The value is greater than the second value (such as 100s -1 ) to obtain the volume of the calcified plaque.

[0080] Preferably, the S4 calculates R2 within the elliptical area * Value greater than 100s -1 The total pixel value of is used to obtain the volume of the calcified plaque, specifically:

[0081] Place R2 in the elliptical area * Value greater than 100s -1 The total pixel value is used as the area of ​​calcified plaque, which is obtained by multi-layer R2 * The volume of calcified plaque was obtained by mapping.

[0082] Furthermore, the S4 calculates R2 in the elliptical area * Value greater than 100s -1 After the total pixel value, it also includes extracting R2 * The three-dimensional model of calcified plaque was constructed for visual display.

[0083] Combine T1W-TSE with R2 in S3 * After the atlas excludes chronic IPH, in R2 * The calcified plaque features are extracted from the atlas and a 3D model is constructed. The specific implementation process is shown in the Matlab code Step 6-8 in Table 1 above.

[0084] 1) Retrieve R2 * Atlas sequence;

[0085] 2) Identify the outer wall contours of the bilateral internal carotid arteries in the transverse axial plane and semi-automatically delineate an elliptical ROI that fits the outer wall of the blood vessels ( Figure 1 Middle B) [code: haz1 (right), hax4 (left)];

[0086] 3) Generate masks of bilateral internal carotid arteries [BW1 (right); BW4 (left)];

[0087] 4) Use threshold segmentation technology to exclude low R2 in ROI * The value of mixed tissue (vascular wall and blood flow in the cavity, etc.) interference is eliminated to achieve accurate segmentation of calcified plaques. The threshold parameter optimization (R2 * >100s -1 ).

[0088] 5) Calcification feature extraction: R2 within ROI * >100s -1 The total number of pixels [counter-R2-ROI(1,i)(right);

[0089] counter-R2-ROI(4,i), (left)], used to quantify the calcification area within the internal carotid artery wall on both sides of each layer;

[0090] 6) Calcified plaque 3D model construction: Extract R2 within ROI * >100s -1 R2 of all pixels * Value, build 3D model [img3D1new (right); img3D4new (left)].

[0091] S5: Calculate the risk score according to the formula: risk score = 0.2 × volume of lipid-rich necrotic core + 0.5 × volume of acute intraplaque hemorrhage + 0.1 × volume of calcified plaque + 0.2 / fat attenuation index for assessment.

[0092] The evaluation method of the present invention has the following advantages:

[0093] 1) Multi-parameter efficiency: a single MRI scan generates T1W-TSE and FF / R2 * The atlas can simultaneously quantitatively assess vascular inflammation and plaque components characterized by PVAT, such as LRNC (via FF atlas), calcified plaque (via R2 * atlas) and IPH staging (by T1W-TSE / R2 * fusion), overcoming the shortcomings of traditional MRI qualitative analysis and its inability to distinguish IPH;

[0094] 2) Quantitative accuracy: based on Dixon's FF / R2 * Computationally eliminates the inherent T1 / T2 difference in chemical shift imaging * deviation;

[0095] 3) 3D dynamic interaction: intuitively displays the spatial relationship between plaque composition and surrounding fat inflammation, assisting clinical decision-making;

[0096] 4) Clinical accessibility: The 5-minute acquisition time facilitates integration into routine carotid artery protocols.

[0097] Verification test

[0098] 1. Scanning repeatability verification test

[0099] The internationally accepted ICC standard was used to evaluate the repeatability of the quantitative evaluation of perivascular adipose tissue (PVAT) using FF atlas in this application, such as Figure 2 The first phase of the study showed that:

[0100] 1) Experimental Design: Five healthy volunteers (aged 36.2 ± 11.9 years, 40% male) were recruited and scanned twice independently under standardized conditions (ambient temperature 22 ± 1°C, equipment calibration error <0.5%).

[0101] 2) Scanning protocol: The key acquisition parameters of the six-echo Dixon sequence described in S1 of this application (TR = 8.9 ms, TE = 1.33-7.48 ms, etc.) were used;

[0102] 3) Region of interest (ROI) standardization: an area of ​​2 mm was defined within the PVAT at the carotid bifurcation. 2 Circular ROI (such as Figure 3 (B and D, orange circles in the middle), ensure that the spatial positioning error is less than 0.1mm;

[0103] 4) Statistical model: A two-way random-effects interclass correlation coefficient (ICC) model was used to evaluate the measurement consistency of FF based on paired regions of interest between the initial and second scans of PVAT.

[0104] The validation results showed that the FF measurement value of PVAT had excellent repeatability between the two scans (ICC=0.982, 95%CI:0.931-0.996, P<0.001), confirming that the quantitative evaluation of perivascular adipose tissue (PVAT) by the present application using FF atlas meets the stability requirements of clinical testing.

[0105] 2. Comparative verification test

[0106] By comparing with the traditional detection method of CT angiography (CTA), the accuracy of the present application in detecting phenotypic transformation of perivascular adipose tissue (PVAT) and calcified plaques was verified. Figure 2 This was shown in a Phase II study.

[0107] 2.1 Subject Cohort

[0108] From January to October 2024, patients meeting the following criteria were consecutively recruited:

[0109] 1) Carotid artery atherosclerotic plaque confirmed by ultrasound examination;

[0110] 2) presence of ≥2 risk factors for atherosclerosis (men aged 50 years or older or women aged 60 years or older, hypertension, diabetes, dyslipidemia, coronary heart disease, and current smoking);

[0111] 3) All patients underwent CTA and MRI examinations, with the time interval between them not exceeding 14 days.

[0112] Exclusion criteria include:

[0113] 1) Previous carotid artery revascularization;

[0114] 2) the presence of other vascular diseases (moyamoya disease, dissection, vasculitis, etc.);

[0115] 3) history of neck radiotherapy;

[0116] 4) Severe disturbance of consciousness;

[0117] 5) Contraindications for MRI.

[0118] Finally, 2 of the 27 cases were excluded: one patient was excluded due to poor imaging quality caused by motion artifacts, and the other patient was excluded due to insufficient scan coverage. A total of 25 patients (age 62.3±11.3 years; 17 males) with carotid atherosclerotic plaques diagnosed by carotid ultrasound were included in the study cohort.

[0119] 2.2 Scanning method

[0120] First, the carotid artery is scanned using the scanning equipment and parameter settings of this application;

[0121] First, CTA scanning was used. Specifically: CTA examinations were performed using a third-generation dual-source CT scanner (SOMATOM Force; Siemens Healthineers, Erlangen, Germany). CTA studies were performed in spiral scanning mode, covering the area from the aortic arch to the cranial vertex. 60 to 80 ml of contrast agent (Omnipaque-350, GE Healthcare) was injected at a rate of 4 ml / s, followed by a 40 ml saline flush via an electric syringe. After an attenuation threshold of 100 HU was detected in the aortic arch for 5 seconds, Bolus tracking was used to initiate acquisition. The carotid CTA scanning parameters were as follows: tube voltage 100 kV, pitch 1.0, reconstruction layer thickness 0.5 mm, reconstruction layer interval 0.5 mm, and rotation time 350 ms.

[0122] 2.3 Statistical analysis

[0123] All statistical analyses were performed using SPSS Statistics version 26.0 (IBM Corp., NY). For the atherosclerosis cohort, bivariate correlations between imaging markers were assessed using the Pearson correlation coefficient (r), including: 1) PVAT FF values ​​on the FF atlas and Hounsfield unit (HU) measurements on co-registered CTA axial images; 2) R2 * Relaxation mapping and CTA-based threshold segmentation quantified calcified plaque area. All correlation analyses were corrected for multiple comparisons using Bonferroni correction, and statistical significance was defined as P < 0.05 (two-sided).

[0124] 2.4 PVAT detection consistency analysis

[0125] Figure 3 Multiparametric imaging features of an adult patient with asymptomatic right carotid atherosclerosis.

[0126] T1W-TSE( Figure 3 Middle (A) shows annular thickening of the right carotid artery wall with focal intraplaque hyperintensity near the stellate calcification (arrow).

[0127] In the atherosclerosis cohort, the average FF value of FF atlas mapping was used to characterize the lipid-water phase transition in the carotid PVAT and compared with the CTA attenuation value. A 2 mm spherical area was delineated within the PVAT contour at the carotid bifurcation and at the level of maximum plaque thickness. 2 ROI to generate the average FF value, such as Figure 3 Figure B. FF profile showing metabolic activity of perivascular adipose tissue (PVAT) at the carotid bifurcation, normalized to 2 mm 2 The average FF value within the region of interest (ROI, outlined in orange) was 965±42‰.

[0128] The FF maps based on the threshold (threshold greater than 50‰) color-coded the FF values ​​in the carotid plaque (the area between the red and green lines) and the perivascular adipose tissue (the area outside the green line), as shown in Figure 2. Figure 3 As shown in C.

[0129] Through multi-index registration anchored by carotid bifurcation contour and plaque / PVAT morphological features, the Hounsfield unit (HU) measurement of PVAT at the corresponding position was obtained from the corresponding spatially matched CTA level, such as Figure 3 The corresponding axial CT angiography (CTA) image confirms the presence of multifocal calcified plaques, and the low-density lesions spatially correspond to the high-signal areas on T1W-TSE (arrows), suggesting intraplaque hemorrhage. 2The average Hounsfield Unit (HU) value within the ROI was -97±5.

[0130] The above results show that in the atherosclerosis cohort, the FF value of PVAT analyzed by FF spectrum in this application is consistent with the PVAT attenuation value detection based on Hounsfield units.

[0131] 2.5 Consistency Analysis of Calcified Plaque Detection

[0132] R2 * Relaxation spectrum ( Figure 3 Middle E) shows the magnetic susceptibility effect within the T1W-TSE high signal area (arrow), which is a characteristic manifestation of chronic intraplaque hemorrhage (chronic IPH), that is, T1W-TSE high signal and R2 * High signal.

[0133] In R2 * After excluding chronic bleeding on the relaxation spectrum, the threshold (>100s -1 ) Semi-automatic segmentation of calcified plaques. On CTA, the area of ​​calcified plaques was quantified as the number of pixels above the threshold by thresholding, and R2 was detected. * Consistency with CTA in quantification of calcified plaques.

[0134] like Figure 3 As shown in F, according to R2 * Calcified plaques were color-coded by values ​​(threshold: R2 * >100s -1 , excluding the chronic IPH area) were consistent with the CTA calcification contours in spatial position.

[0135] Figure 4 Multiparametric imaging features of an adult patient with symptomatic left carotid artery stenosis.

[0136] T1W-TSE( Figure 4 Middle (A) shows a vulnerable plaque in the left carotid artery, characterized by a discontinuous fibrous cap (arrowhead) and adjacent areas of T1W-TSE hyperintensity, suggesting fresh intraplaque hemorrhage (IPH, arrowhead), consistent with acute plaque rupture. Small amounts of calcification are observed in the plaque shoulder (asterisk).

[0137] like Figure 4 As shown in center B, the FF map depicts the characteristics of perivascular adipose tissue (PVAT).

[0138] The fat segmentation after threshold optimization (threshold>50‰) color-coded the plaque microenvironment. Figure 4 Middle C shows the area with low lipid content within the plaque (between the red and green lines) and PVAT associated with inflammation (area outside the green line).

[0139] Figure 4 A mid-D axial CT angiography (CTA) image demonstrates the presence of a noncalcified, low-density plaque (arrow) with punctate calcifications (asterisks), which is anatomically consistent with the high-risk features identified on T1W-TSE.

[0140] R2 * Relaxation spectrum ( Figure 4 Middle E) shows a fresh IPH with low signal intensity (arrow) and a calcified component with high signal intensity (asterisk).

[0141] T1W-TSE hyperintensity and R2 * Plaque components with low signal intensity were judged as acute IPH.

[0142] According to R2 * Color-coded calcification deposition (threshold: R2 * >100s -1 ) showed that the spatial position of calcification was completely consistent with that on CTA, such as Figure 4 Middle F.

[0143] Figure 3 、 Figure 4 All show that through R2 * The area of ​​calcified plaque quantified by relaxometry is consistent with the quantitative detection of calcified plaque by CTA.

[0144] 2.6 Verification Results

[0145] The mean PVAT values ​​derived from fat fraction (FF) mapping ranged from 213‰ to 998‰ (mean FF±SD: 610.7±223.1‰), and the corresponding Hounsfield units (HU) values ​​on CTA ranged from -109 to 38 (mean HU±SD: -41.9±38.1). Figure 5 The results showed that there was a strong negative linear correlation between the FF values ​​derived from the FF atlas and the HU values ​​measured on geometrically matched CTA axial images (Pearson correlation coefficient r = -0.868, P < 0.001; Figure 5 Middle A).

[0146] Forty carotid arteries with calcified plaques from 22 patients (age 64.2 ± 9.6 years; 16 males) were included in the analysis of calcium content measurements. Figure 5 Shows R2 * Relaxometry showed high agreement with CTA as the reference standard in quantifying calcified plaque area (Pearson correlation coefficient r = 0.991, P < 0.001; Figure 5 Middle B).

[0147] 3. Visual display

[0148] The evaluation method of the present invention is to simultaneously quantify the PVAT metabolic phenotype and LRNC (by FF profile analysis), calcification load (by R2 * atlas) and IPH staging (by T1W-TSE and R2 * The fusion method provides comprehensive characteristics of carotid plaque composition and inflammatory metabolism, forming a potential comprehensive imaging biomarker platform for atherosclerosis risk stratification and treatment monitoring. Among them, the invention extracts plaque components (LRNC, IPH, calcified plaque) and PVAT segmentation within the coverage range to establish a 3D model. Matlab supports 3D model rotation and cross-section cutting through an interactive visualization platform, and automatically generates plaque component volume ratio, PVAT inflammation score and spatial distribution heat map report ( Figure 6 ), enabling quantitative analysis of multimodal plaque composition and inflammation in carotid atherosclerosis.

[0149] The present invention also provides a method based on FF and R2 * Atlas' carotid plaque and vascular inflammation assessment system includes:

[0150] Imaging module: Based on the three-dimensional six-echo Dixon sequence, the target area of ​​the carotid artery is imaged, and the fat fraction map and R2 are obtained. * Relaxation spectra, recorded as FF spectra, R2 * Atlas;

[0151] Fat-rich necrotic core and fat attenuation index quantification module: For each FF atlas layer, the outer wall contours of the bilateral internal carotid arteries in the transverse axial plane are identified. An adaptive ellipse is generated along the outer wall of the vessel using a morphologically constrained ellipse growing algorithm, which serves as the inner boundary. The ellipse automatically expanded outward based on the inner boundary serves as the outer boundary.

[0152] Calculate the total number of pixels within the inner boundary whose fat fraction is not less than the first value to obtain the volume of the fat-rich necrotic core;

[0153] Calculate the average fat fraction of all pixels in the annular area between the inner and outer boundaries whose fat fraction is not less than the first value as the fat attenuation index;

[0154] Acute intraplaque hemorrhage quantification module: imaging of the target area of ​​the carotid artery based on T1-weighted turbo spin echo sequence, combined with R2 * Spectrum, calculated T1-weighted turbo spin echo sequence is greater than the first preset signal and is less than R2 * After the second preset signal area, the volume of acute intraplaque hemorrhage is obtained by multi-slice T1-weighted turbo spin echo imaging;

[0155] Calcified plaque quantification module: for each layer R2 * After identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial position, an elliptical area that fits the outer wall of the blood vessels was delineated and R2 within the elliptical area was calculated. * The total pixel value of which the value is greater than the second value is used to obtain the volume of the calcified plaque;

[0156] Evaluation module: used to calculate the risk score and perform evaluation according to the formula: risk score = 0.2×volume of fat-rich necrotic core + 0.5×volume of acute intraplaque hemorrhage + 0.1×volume of calcified plaque + 0.2 / fat attenuation index.

Claims

1. A method based on FF and R2 * The atlas carotid artery plaque and vascular inflammation assessment method is characterized by: include: S1: Imaging of the target area of ​​the carotid artery based on a three-dimensional six-echo Dixon sequence, and the resulting fat fraction map and R2 * Relaxation spectra, recorded as FF spectra, R2 * Atlas; S2: For each FF atlas layer, after identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial plane, an adaptive ellipse is generated along the outer wall of the vessel based on the morphologically constrained ellipse growing algorithm, which serves as the inner boundary. The ellipse automatically expanded outward based on the inner boundary serves as the outer boundary. Calculate the total number of pixels within the inner boundary whose fat fraction is not less than the first value to obtain the volume of the fat-rich necrotic core; Calculate the average fat fraction of all pixels in the annular area between the inner and outer boundaries whose fat fraction is not less than the first value as the fat attenuation index; S3: Imaging of the target area of ​​the carotid artery based on T1-weighted turbo spin echo sequence, combined with R2 * Spectrum, calculated T1-weighted turbo spin echo sequence is greater than the first preset signal and is less than R2 * After the second preset signal area, the volume of acute intraplaque hemorrhage is obtained by multi-slice T1-weighted turbo spin echo imaging; S4: For each layer R2 * After identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial position, an elliptical area that fits the outer wall of the blood vessels was delineated and R2 within the elliptical area was calculated. * The total pixel value of which the value is greater than the second value is used to obtain the volume of the calcified plaque; S5: Calculate the risk score according to the formula: risk score = 0.2 × volume of lipid-rich necrotic core + 0.5 × volume of acute intraplaque hemorrhage + 0.1 × volume of calcified plaque + 0.2 / fat attenuation index for assessment.

2. The method according to claim 1 based on FF and R2 * The atlas carotid artery plaque and vascular inflammation assessment method is characterized by: The step S2 further includes calculating the total number of pixels in the annular region between the inner and outer boundaries whose fat fraction is not less than a first value as the area of ​​perivascular adipose tissue, and obtaining the volume of the perivascular adipose tissue from the multi-layer FF atlas.

3. The method according to claim 1 based on FF and R2 * The atlas carotid artery plaque and vascular inflammation assessment method is characterized by: Before S2, it also included: The carotid bifurcation level was used as the central level, and 7 levels above and below the central level were included, totaling 15 levels to cover the 1.5 cm area above and below the carotid bifurcation for analysis.

4. The method according to claim 1 based on FF and R2 * The atlas carotid artery plaque and vascular inflammation assessment method is characterized by: S2 calculates the total number of pixels within the inner boundary whose fat fraction is not less than the first value, and obtains the volume of the fat-rich necrotic core, specifically: The total pixel value within the inner boundary whose fat fraction is not less than the first value is taken as the area of ​​the fat-rich necrotic core, and the volume of the fat-rich necrotic core is obtained from the multi-layer FF atlas.

5. The method according to claim 1 based on FF and R2 * The atlas carotid artery plaque and vascular inflammation assessment method is characterized by: S4 calculates R2 within the elliptical area * The total pixel value whose value is greater than the second value is used to obtain the volume of the calcified plaque, specifically: Place R2 in the elliptical area * The total pixel value with a value greater than the second value is regarded as the area of ​​the calcified plaque, which is obtained by multi-layer R2 * The volume of calcified plaque was obtained by mapping.

6. The method according to claim 1 based on FF and R2 * The atlas carotid artery plaque and vascular inflammation assessment method is characterized by: After calculating the total number of pixels within the inner boundary whose fat fraction is not less than the first value, S2 further includes extracting the fat fraction value and constructing a three-dimensional model of the fat-rich necrotic core for visual display; S4 calculates R2 within the elliptical area * After the total pixel value is greater than the second value, it also includes extracting R2 * The three-dimensional model of calcified plaque was constructed for visual display.

7. The method according to claim 2 based on FF and R2 * The atlas carotid artery plaque and vascular inflammation assessment method is characterized by: After calculating the total pixel value in the annular area between the inner and outer boundaries where the fat fraction is not less than the first value, the method further includes extracting the fat fraction value and constructing a three-dimensional model of perivascular fat tissue for visual display.

8. A method based on FF and R2 * The atlas carotid artery plaque and vascular inflammation assessment system is characterized by: include: Imaging module: Based on the three-dimensional six-echo Dixon sequence, the target area of ​​the carotid artery is imaged, and the fat fraction map and R2 are obtained. * Relaxation spectra, recorded as FF spectra, R2 * Atlas; Fat-rich necrotic core and fat attenuation index quantification module: For each FF atlas layer, the outer wall contours of the bilateral internal carotid arteries in the transverse axial plane are identified. An adaptive ellipse is generated along the outer wall of the vessel using a morphologically constrained ellipse growing algorithm, which serves as the inner boundary. The ellipse automatically expanded outward based on the inner boundary serves as the outer boundary. Calculate the total number of pixels within the inner boundary whose fat fraction is not less than the first value to obtain the volume of the fat-rich necrotic core; Calculate the average fat fraction of all pixels in the annular area between the inner and outer boundaries whose fat fraction is not less than the first value as the fat attenuation index; Acute intraplaque hemorrhage quantification module: imaging of the target area of ​​the carotid artery based on T1-weighted turbo spin echo sequence, combined with R2 * Spectrum, calculated T1-weighted turbo spin echo sequence is greater than the first preset signal and is less than R2 * After the second preset signal area, the volume of acute intraplaque hemorrhage is obtained by multi-slice T1-weighted turbo spin echo imaging; Calcified plaque quantification module: for each layer R2 * After identifying the outer wall contours of the bilateral internal carotid arteries in the transverse axial position, an elliptical area that fits the outer wall of the blood vessels was delineated and R2 within the elliptical area was calculated. * The total pixel value of which the value is greater than the second value is used to obtain the volume of the calcified plaque; Evaluation module: used to calculate the risk score and perform evaluation according to the formula: risk score = 0.2×volume of fat-rich necrotic core + 0.5×volume of acute intraplaque hemorrhage + 0.1×volume of calcified plaque + 0.2 / fat attenuation index.

Images