Construction method and system of liver fat non-invasive quantitative evaluation model based on PCD-CT

Abstract

The application provides a method and system for constructing a liver fat noninvasive quantitative evaluation model based on PCD-CT. The model construction method comprises the following steps: 1: recruit a number of healthy volunteers and divide them into a training queue and a verification queue; 2: perform PCD-CT routine scanning on the training queue and perform PCD-CT scanning under different acquisition conditions on the verification queue; reconstruct the CT scanning images to obtain standardized CT images; 3: perform water and fat iterative decomposition and echo asymmetry least square estimation sequence scanning on the training queue to obtain MRI-PDFF images; 4: perform image post-processing and analysis in the region of interest in the standardized CT images and the MRI-PDFF images respectively, and calculate the average value and the standard deviation of the standardized CT and the MRI-PDFF; and 5: obtain a liver fat noninvasive quantitative evaluation model by using linear least square regression. The application realizes precise evaluation of intrahepatic fat under different acquisition conditions.

Description

Technical Field

[0001] This invention relates to the fields of medicine and biotechnology, particularly the field of medical device technology innovation, and relates to the ability of photon counting CT to accurately and stably perform non-invasive quantitative assessment of liver fat content under different scanning schemes. Background Technology

[0002] Intrahepatic steatosis is the histopathological feature of metabolic-associated fatty liver disease (MAFLD), currently the most common chronic liver disease worldwide and a common characteristic of other liver diseases. It is frequently associated with obesity, metabolic syndrome, and diabetes. Early steatosis is often asymptomatic, but as fat content increases, changes in hepatic metabolism can occur, potentially progressing to steatohepatitis, and subsequently cirrhosis and hepatocellular carcinoma. This leads to increased disability rates and healthcare costs. Therefore, accurate and reliable quantification of intrahepatic fat content is crucial for the diagnosis, treatment evaluation, and monitoring of MAFLD.

[0003] Liver biopsy remains the gold standard for diagnosing and grading intrahepatic fat content. However, as an invasive procedure, the sample location of a puncture biopsy is limited, making it impossible to comprehensively assess the state of intrahepatic steatosis. Furthermore, liver biopsy has limitations such as sampling error, inter-observer variability, and low patient acceptance. Therefore, non-invasive imaging methods that can quantify the entire liver and provide dynamic follow-up for assessing intrahepatic fat content are particularly important.

[0004] Currently, chemical shift imaging-based 3D multi-echo gradient echo magnetic resonance imaging (CSE-MRI) is considered the gold standard for non-invasive quantitative assessment of liver fat content. It can be obtained with a single breath-hold and generates a proton density fat fraction (PDFF) image of the entire liver, corrected for confounding factors, within 20 seconds, serving as a biomarker for fat content. MRI-PDFF is a quantitative imaging biomarker capable of accurate and reproducible quantitative assessment of liver fat and has been applied in numerous clinical trials. However, in clinical practice, many patients (such as elderly or obese patients) cannot hold their breath even within the appropriate acquisition time, resulting in respiratory motion-related ghosting on the images and leading to measurement bias and errors in PDFF. Furthermore, non-invasive quantitative MRI of liver fat content is not feasible when the patient has contraindications to MRI, such as metallic implants or claustrophobia. Therefore, developing a new imaging method to replace the non-invasive gold standard PDFF is of significant clinical importance.

[0005] Currently, plain abdominal CT scans are widely used clinically, and the decay of CT values ​​is closely related to intrahepatic fat content. In traditional energy integral detector (EID) CT, researchers have demonstrated a high linear correlation between CT values ​​and MRI-PDFF. Based on this linear relationship, a clinical CT-MRI conversion equation has been derived, making CT values ​​a promising new indicator for non-invasive quantitative assessment of intrahepatic fat content. However, this equation only applies to acquisitions performed at 120 kVp, and the stability of CT values ​​is highly dependent on the scanning tube voltage. Changing the tube voltage will also change the CT value. In clinical scanning practice, besides using the conventional 120 kVp, the tube voltage is not fixed under different circumstances. For example, 80 kVp is appropriately selected for scanning children to reduce radiation dose, while a high-voltage tube such as 140 kV is more suitable for scanning obese patients. This greatly limits the application of quantifying fat content based on plain CT values. In addition, some studies have achieved the quantification of intrahepatic fat content based on the three-substance separation method of dual-energy CT; however, the results of these studies are only focused on enhanced CT scans and do not show a significant advantage compared to plain abdominal CT values.

[0006] In summary, existing non-invasive imaging techniques for quantifying intrahepatic fat content have certain limitations. Summary of the Invention

[0007] To overcome the aforementioned technical deficiencies, the first aspect of this invention provides a method for constructing a non-invasive quantitative assessment model of liver fat based on PCD-CT, comprising:

[0008] Step S1: Recruit several healthy volunteers who meet the inclusion criteria and randomly assign them to different groups based on tube voltage and radiation dose. Then, further divide the groups of healthy volunteers into a training cohort and a validation cohort. (Subsequently, the training cohort will undergo routine PCD-CT scans, while the validation cohort will undergo PCD-CT scans under different acquisition conditions.)

[0009] Step S2: Perform PCD-CT scans on the training queue, and then reconstruct the CT scan images to obtain standardized CT images;

[0010] Step S3: Perform iterative decomposition of water and fat and scanning of echo asymmetric least squares estimation sequences on the training queue to obtain MRI-PDFF images;

[0011] Step S4: Select several regions of interest of the same size at the same location in the standardized CT image and MRI-PDFF image respectively for image post-processing and analysis, and calculate the mean and standard deviation of the standardized CT and MRI-PDFF from all regions of interest;

[0012] Step S5: Using R statistical software, linear least squares regression was employed to obtain a non-invasive quantitative assessment model for liver fat.

[0013] MRI-PDFF(%)=-0.58xCT(HU)+43.13

[0014] Among them, MRI-PDFF is a magnetic resonance imaging assessment of proton density fat fraction, and its value is a percentage; standardized

[0015] CT refers to standardized CT scans obtained through PCD-CT technology, and the Hounsfield unit used is HU.

[0016] Further, in step S1, all volunteers were randomly assigned to 6 groups, each with a scanning protocol having different tube voltages and dose levels. The groupings were as follows: tube voltages were divided into three groups based on the commonly used clinical procedures of 90kVp, 120kVp, and 140kVp; each tube voltage group was further divided into a standard dose group and a low dose group based on the radiation dose.

[0017] Further, in step S2, a first-generation photon-counting detector CT was used, equipped with two photon-counting detectors. The scanning scheme applied tube voltages of 90kVp, 120kVp, and 140kVp to the 90kVp, 120kVp, and 140kVp groups, respectively. The image quality IQ level for the low-dose group was set to 90, and the IQ level for the standard-dose group was set to 145. Automatic tube current modulation optimized for non-contrast enhancement was used. The collimator width was 144*0.4m, the rotation time was 0.25s, and the pitch coefficient was 0.8. All images were reconstructed on the cross-section using the soft tissue reconstruction kernel Br40, employing quantum iterative reconstruction with an intensity level of 3, a slice thickness of 1mm, and an interval of 0.7mm.

[0018] Further, in step S3, MRI-PDFF measurements were performed on the volunteers using a 3.0T MRI (Architect, GE Healthcare) with the following scanning parameters: TR = 6.4 ms; first echo time (TE) = 0.9 ms; 6 echoes with an interval TE of 0.8 ms; field of view = 440 × 330 mm; slice thickness = 3 mm; flip angle = 3°; matrix = 160 × 160 × 50; number of excitations = 0.5; scan time = 18 s.

[0019] Furthermore, in step S4, for each volunteer, three puncture sites with an area of ​​2-3 cm² were placed in the left lobe, right anterior lobe, and right posterior lobe of the liver, respectively. 2Circular regions of interest were selected, avoiding major blood vessels and bile ducts, liver capsule, artifacts, and gas in the lungs or gastrointestinal tract; post-processing of standardized CT images was performed using syngo.via, version VB60, Siemens Healthineers; post-processing of MRI-PDFF images was performed using AW 4.7, GE Healthcare.

[0020] A second aspect of this application provides a PCD-CT-based non-invasive quantitative assessment system for liver fat, characterized in that it comprises:

[0021] An input module is used to acquire standardized CT values, which are obtained based on PCD-CT scanning and reconstruction of CT scan images;

[0022] A quantitative assessment module is used to evaluate MRI-PDFF results according to the following non-invasive quantitative assessment model for liver fat:

[0023] MRI-PDFF(%)=-0.58xCT(HU)+43.13

[0024] Among them, MRI-PDFF is a magnetic resonance imaging assessment of proton density fat fraction, and its value is a percentage; standardized CT is a standardized CT obtained through PCD-CT technology, and the Hounsfield unit used is HU.

[0025] The output module is used to output MRI-PDFF results and qualitative assessment results corresponding to the MRI-PDFF results according to the fatty liver diagnosis rules.

[0026] Furthermore, the criteria for determining fatty liver are as follows: when the MRI-PDFF result is less than 5%, the volunteer is assessed as having no fatty liver; when the MRI-PDFF result is 5%-15%, the volunteer is assessed as having mild fatty liver; when the MRI-PDFF result is 15%-25%, the volunteer is assessed as having moderate fatty liver; and when the MRI-PDFF result is greater than 25%, the volunteer is assessed as having severe fatty liver.

[0027] Compared with existing technologies, the above technical solution has the following advantages:

[0028] This application discloses a non-invasive quantitative assessment method and system for liver fat based on PCD-CT, enabling accurate assessment of intrahepatic fat under different acquisition conditions. PCD-CT is a novel CT technology that uses a photon-counting detector (PCD) to directly convert detected X-ray photons into electrical signals, measuring the specific energy of each photon. By using energy thresholds, it achieves inherent energy spectrum separation and outputs fixed single-energy images, improving spatial resolution and spectral information for material decomposition. Standardized CT attenuation images, i.e., standardized CT values, can be acquired under different scanning schemes (different tube voltages, different radiation doses, etc.). Therefore, the standardized CT values ​​obtained through PCD-CT can reliably and accurately quantify intrahepatic fat content at different tube voltages or radiation dose levels. PCD technology can remove electronic background noise, ensuring the stability of CT values ​​in low-dose scans. This application presents the first liver fat quantitative assessment system based on a novel photon-counting detector CT (PCD-CT). During the design process, MRI-PDFF was used as a reference standard to verify the accuracy of standardized CT values ​​generated by PCD-CT in liver fat quantification, and to obtain a conversion formula between standardized CT values ​​and MRI-PDFF. This method overcomes the limitations of existing methods for measuring intrahepatic fat. Liver biopsy, as an invasive procedure, cannot provide whole-liver, dynamic monitoring of intrahepatic fat changes; MRI-PDFF is currently the gold standard for non-invasive quantitative assessment, but its results are affected by respiratory artifacts, and patients with contraindications to MRI cannot be measured. The PCD-CT-based intrahepatic fat assessment system can achieve the conversion between standardized CT values ​​and MRI-PDFF, enabling accurate assessment of intrahepatic fat even for patients who cannot undergo MRI. Furthermore, CT operation is convenient and quick, allowing for dynamic monitoring of whole-liver intrahepatic fat and evaluation of treatment efficacy. This application derives a CT-MRI fat quantification conversion formula based on the linear correlation between standardized CT values ​​and PDFF under different CT acquisition conditions (different tube voltages and different radiation doses). Furthermore, its accuracy is verified under different CT acquisition conditions (different tube voltages and different radiation doses). The fat fraction CTFF value obtained by substituting the standardized CT values ​​into the formula shows a high degree of consistency with the actual PDFF. Although previous researchers have demonstrated a linear relationship between CT values ​​and MRI-PDFF in energy integral detector CT, this equation only holds true under specific scanning conditions (120 kVp), which significantly limits its widespread clinical application.The CT-MRI fat quantification conversion formula in this application can achieve stability and accuracy of measurement results under different acquisition conditions, making it suitable for different clinical application scenarios (such as appropriately selecting 80kVp for children to reduce radiation dose, and selecting higher voltages such as 140kV for obese patients). It has broad applicability, effectiveness, feasibility, and safety in clinical practice, providing the possibility for its widespread use in actual clinical operation. Attached Figure Description

[0029] Figure 1 A diagram illustrating the number of volunteers to be recruited and the grouping principle.

[0030] Figure 2 Standardized CT image (left) from PCD-CT and MRI-PDFF image (right) from chemical shift magnetic resonance imaging of a 33-year-old male. Three regions of interest (ROIs) were placed at corresponding locations in the left lobe, right anterior lobe, and right posterior lobe of the liver.

[0031] Figure 3 The correlation between standardized CT values ​​and MRI-PDFF values ​​is shown in the training and validation cohorts, respectively. The scatter plot of PDFF values ​​versus CT values ​​in the training cohort (156 ROIs) with PDFF as the reference standard shows a linear correlation (r = 0.936, P < 0.001). This cohort was used to derive the linear transformation formula: MRI-PDFF(%) = -0.58 × CT(HU) + 43.13.

[0032] Figure 4 A scatter plot showing PDFF values ​​versus CT values ​​for the validation queue (1069 ROIs) demonstrates a good linear correlation. The black dashed line represents... Figure 2 CT-MRI fitting lines in the training cohort.

[0033] Figure 5 These are Bland–Altman plots of standardized CT values, fat fractions (CTFF and PDFF). Detailed Implementation

[0034] The advantages of the present invention are further illustrated below with reference to the accompanying drawings and specific embodiments. Those skilled in the art should understand that the following detailed description is illustrative rather than restrictive and should not be construed as limiting the scope of protection of the present invention.

[0035] Example 1

[0036] This implementation provides a process for constructing a non-invasive quantitative assessment model for liver fat based on PCD-CT, as detailed below:

[0037] Step 1: Volunteer Recruitment

[0038] Between September and December 2023, 500 healthy adults and older adults (aged 65 years and older) were recruited consecutively. Inclusion criteria were healthy individuals able to sign informed consent forms. Exclusion criteria included pregnant women, individuals with metal implants, a history of chronic liver disease, and suspected hepatic iron overload (R2). * People with a heart rate >127Hz or those who cannot tolerate MRI examinations due to claustrophobia.

[0039] All volunteers were randomly assigned to 6 groups, each with different tube voltages and dose levels. The groupings were as follows: tube voltages were divided into three groups based on the clinically commonly used 90kVp, 120kVp, and 140kVp. Each tube voltage group was further divided into a standard dose scanning group and a low dose scanning group based on different image quality (IQ) levels. The IQ level for the low dose scanning group was set to 90, and the IQ level for the standard dose scanning group was set to 145.

[0040] Specifically, such as Figure 1 As shown, a total of 500 volunteers were included, of whom 88 were excluded for the following reasons: incomplete MRI data (n=44), severe respiratory artifacts on MRI (n=34), localized intrahepatic lesions (n=4), and suspected intrahepatic iron overload (n=6). Furthermore, due to unreliable fat quantification caused by respiratory artifacts, data from 11 ROIs were excluded from 9 volunteers. Finally, 412 volunteers (1225 ROIs) were analyzed, with a median age of 47 years (range 18–74 years). All volunteers were randomly assigned to 6 groups, using different tube voltages (i.e., 90, 120, and 140 kVp) and dose levels (standard dose and low dose). To avoid bias, the data were further divided into training and validation cohorts. The 53 volunteers (156 ROIs) in the 120 kVp standard dose subgroup served as the training cohort. The remaining 359 volunteers (1069 ROIs) served as the validation cohort. There were no significant differences in sex, age, and BMI among the 6 groups.

[0041] Step 2: Photon Counting Detector (PCD CT) Scanning Protocol and Reconstruction

[0042] The first-generation photon-counting detector CT scanner (NAEOTOM Alpha, Siemens Healthineers, Germany) was used, equipped with two photon-counting detectors. Scanning protocols applied tube voltages of 90kVp, 120kVp, and 140kVp for the 90kVp, 120kVp, and 140kVp groups, respectively. The image quality IQ level was set to 90 for the low-dose group and 145 for the standard-dose group, using automated tube current modulation optimized for non-contrast enhancement (CARE Dose4D, Siemens Healthcare). Other parameters included a collimator width of 144 x 0.4 m, a rotation time of 0.25 s, and a pitch factor of 0.8.

[0043] All images were reconstructed in cross-sections using a soft tissue reconstruction kernel (Br40) with quantum iterative reconstruction at an intensity level of 3. Slices were 1 mm thick with a spacing of 0.7 mm.

[0044] Step 3 MRI Scanning Protocol

[0045] The median time interval between CT and MRI examinations was 0 days (range 0–22 days).

[0046] MRI-PDFF (Iterative PDFF) measurements, i.e., scans using an IDEAL-IQ sequence of water and fat iterative decomposition and echo asymmetric least squares estimation, were performed on volunteers using a 3.0T MRI (Architect, GE Healthcare) to assess intrahepatic fat content. Outputs upon scan completion included water, fat, R2*, and proton density fat fraction (PDFF) maps. The R2* map was used to assess the presence of iron deposition, and the fat fraction was used to assess the degree of intrahepatic fat accumulation. Scan parameters were as follows: TR 6.4 ms; first echo time (TE) 0.9 ms; 6 echoes with an interval TE of 0.8 ms; field of view 440 × 330 mm; slice thickness 3 mm; flip angle 3°; matrix 160 × 160 × 50; number of excitations 0.5; scan time 18 s.

[0047] Step 4: Image Post-processing and Analysis

[0048] Post-processing of PCD-CT data was performed on a dedicated Siemens workstation (MM reading, syngo.via, version VB60, Siemens Healthineers). Post-processing of IDEAL-IQ body fat fraction maps was performed using a workstation (AW 4.7, GE Healthcare).

[0049] For each volunteer, three tissue samples with an area of ​​2-3 cm² were placed in the left lobe, right anterior lobe, and right posterior lobe of the liver, respectively. 2The circular region of interest (ROI) is selected. When choosing an ROI, avoid major blood vessels and bile ducts, liver capsule, artifacts, and gas in the lungs or gastrointestinal tract. The size and location of the PDFF ROI should match the ROI on the PCD-CT image as closely as possible. Figure 2 Example of ROI.

[0050] Step 5: Statistical analysis and establishment of a non-invasive quantitative assessment model for liver fat based on standardized CT values ​​obtained through PCD-CT.

[0051] R statistical software was used. Continuous variables were described as mean ± standard deviation or median and interquartile range (IQR). Categorical data were expressed as numbers and percentages. Within-group correlation coefficient (ICC) was used to test the consistency among observers.

[0052] For research purposes, for example, 50% of volunteers in the 120kV standard dose group were randomly assigned to the training cohort to establish the conversion formula, while all other volunteers served as the validation cohort to verify the accuracy of standardized CT fat fraction (CTFF) quantification. A linear least squares regression method was used to establish the CT-MRI fat quantification conversion formula. Figure 3 As shown, in the training cohort, standardized CT and MRI-PDFF showed a strong linear correlation (r = -0.936, P < 0.001), yielding the following CT-MRI fat quantification conversion formula:

[0053] MRI-PDFF(%)=-0.58×CT(HU)+43.13

[0054] The above formula is a non-invasive quantitative assessment model for liver fat based on standardized CT values. MRI-PDFF is the proton density fat fraction assessed by magnetic resonance imaging, expressed as a percentage; standardized CT refers to standardized CT obtained through PCD-CT technology, with units of HU.

[0055] Verification Example

[0056] After establishing a non-invasive quantitative assessment model for liver fat in the above embodiments, the CT values ​​of the validation cohort are substituted into the fat quantitative conversion formula to obtain the CT-based fat fraction CTFF value. The CT measurement results are then converted into CTFF for quantitative accuracy assessment.

[0057] (1) Subgroup and whole group analysis of liver fat content using different CT methods.

[0058] The correlation between CTFF and PDFF was assessed using the Pearson correlation coefficient. In the validation cohort dataset with different tube voltages (90 kVp, 120 kVp, 140 kVp) and dose levels, a strong correlation was also observed between CT values ​​and PDFF. Figure 4The CTFF values ​​in the validation dataset were estimated using this transformation formula.

[0059] Bland-Altman analysis and ICC were used to assess the consistency and reliability between CTFF and PDFF. Furthermore, subgroup analyses were performed at three different tube voltages and two different dose levels. See Table 1.

[0060] Table 1. ICC and Bland-Altman assessments of the consistency between CTFF and PDFF

[0061]

[0062] With a fixed tube voltage, the correlation between CTFF and PDFF was good, with all ICC values ​​greater than 0.930. Considering the standard dose and low dose levels separately, the ICC for the standard dose group was 0.929 (0.917–0.939), and the ICC for the low dose group was 0.948 (0.938–0.957). In all subgroup comparisons, the mean deviation remained below 1%, with the highest observed deviation of 0.88% ± 2.06% in the 90 kVp group.

[0063] When considering the entire validation dataset, the ICC between CTFF and PDFF is 0.938 (0.930-0.945). The overall bias is 0.043% ± 2.29%, very close to zero. Figure 5 Further analysis showed that the average deviation between the two was not correlated with fat content.

[0064] (2) CTFF Error Analysis

[0065] To further explore the potential impact on the accuracy of CTFF, the absolute errors between CTFF and PDFF were calculated and analyzed. The Kruskal-Wallis H test was used to compare the differences in absolute errors among the three tube voltage and three fatty degeneration level groups, and the Mann-Whitney U test was used to compare the two dose groups. Furthermore, mixed-effects linear regression was performed to further explore the effects of tube voltage, dose level, and fat fraction (PDFF) on quantitative accuracy.

[0066] CTFF error was defined as the absolute difference between PDFF and CTFF for each ROI, with PDFF used as the reference standard. Mixed-effects multiple linear regression analysis showed that FF error was not affected by tube voltage, dose level, PDFF, or the interaction between tube voltage and dose level (see Table 2).

[0067] Table 2. Linear mixed-effects regression analysis of standardized CT fat fraction CTFF error

[0068] (intercept) 1.861±0.110 <0.001 Tube voltage (kVp) 120 Reference group 140 0.171±0.124 0.169 90 -0.111±0.144 0.441 dose level Standard dose Reference group low dose -0.159±0.145 0.273 PDFF(%) 0.00899±0.00576 0.119 kVp 140kVp: Low dose -0.0943±0.191 0.622 kVp 90kVp: Low dose 0.200±0.204 0.327

[0069] (3) Radiation dose assessment

[0070] The interaction between tube voltage and radiation dose was assessed. At different tube voltages, the radiation dose indices CTDIvol (volume CT dose index), DLP (dose-length product), and effective dose in the low-dose group were significantly lower than those in the standard-dose group (all P < 0.001). Between the standard and low-dose groups, the CTDIvol, DLP, and effective dose in the 90 kVp group were significantly lower than those in the 120 kVp and 140 kVp groups (all P < 0.001), while there was no significant difference in radiation dose between the 120 kVp and 140 kVp groups (all P > 0.5).

[0071] Example 2

[0072] This embodiment provides a PCD-CT-based non-invasive quantitative assessment system for liver fat, which is based on the PCD-CT-based non-invasive quantitative assessment model for liver fat constructed in Embodiment 1. The non-invasive quantitative assessment system for liver fat includes: an input module, a quantitative assessment module, and an output module.

[0073] The input module is used to obtain standardized CT values, which are obtained based on PCD-CT scans and reconstruction of CT scan images.

[0074] The quantitative assessment module is used to evaluate MRI-PDFF results according to the following non-invasive quantitative assessment model for liver fat:

[0075] MRI-PDFF(%)=-0.58xCT(HU)+43.13

[0076] Among them, MRI-PDFF is a magnetic resonance imaging assessment of proton density fat fraction, and its value is a percentage; standardized CT is a standardized CT obtained through PCD-CT technology, and the Hounsfield unit used is HU.

[0077] The output module is used to output MRI-PDFF results and qualitative assessment results corresponding to the MRI-PDFF results according to the fatty liver diagnosis rules.

[0078] The criteria for determining fatty liver are as follows: when the MRI-PDFF result is less than 5%, the volunteer is assessed as having no fatty liver; when the MRI-PDFF result is 5%-15%, the volunteer is assessed as having mild fatty liver; when the MRI-PDFF result is 15%-25%, the volunteer is assessed as having moderate fatty liver; and when the MRI-PDFF result is greater than 25%, the volunteer is assessed as having severe fatty liver.

[0079] Of the 412 volunteers, 226 had normal liver fat content, 121 had mild fatty liver, 50 had moderate fatty liver, and 15 had severe fatty liver. Therefore, the moderate and severe fatty liver groups were combined into a moderate-to-severe fatty liver group. A higher proportion of men exhibited mild fatty liver (P < 0.05) or moderate-to-severe fatty liver (P < 0.05) compared to women. The average BMI across all volunteers was 24.34 [21.59; 26.89]. Higher BMI was associated with higher PDFF (P < 0.001).

[0080] It should be noted that the embodiments of the present invention have better implementability and are not intended to limit the present invention in any way. Any person skilled in the art may use the above-disclosed technical content to change or modify it into equivalent effective embodiments. However, any modifications or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A method for constructing a non-invasive quantitative assessment model of liver fat based on PCD-CT, characterized in that, include: Step S1: Recruit several healthy volunteers who meet the inclusion criteria and randomly assign them to different groups according to tube voltage and radiation dose. Then, further divide the groups of healthy volunteers into training cohorts and validation cohorts. Step S2: Perform PCD-CT scans on the training cohort, then reconstruct the CT scan images to obtain standardized CT images; Step S3: Perform iterative decomposition of water and fat and scanning of echo asymmetric least squares estimation sequences on the training queue to obtain MRI-PDFF images; Step S4: Select several regions of interest of the same size at the same location in the standardized CT image and MRI-PDFF image respectively for image post-processing and analysis, and calculate the mean and standard deviation of the standardized CT and MRI-PDFF from all regions of interest; Step S5: Using R statistical software, linear least squares regression was employed to obtain a non-invasive quantitative assessment model for liver fat. Among them, MRI-PDFF is a magnetic resonance imaging assessment of proton density fat fraction, and its value is a percentage; standardized CT is a standardized CT obtained through PCD-CT technology, and the Hounsfield unit used is HU. In step S1, all volunteers were randomly assigned to 6 groups, each with a scanning protocol that had different tube voltages and dose levels. The groups are as follows: the tube voltage is divided into three groups according to the commonly used clinical procedures of 90kVp, 120kVp, and 140kVp. Each tube voltage group is further divided into a standard dose group and a low dose group according to the radiation dose.

2. The method for constructing a non-invasive quantitative assessment model of liver fat based on PCD-CT as described in claim 1, characterized in that, In step S2, a first-generation photon-counting detector CT was used, equipped with two photon-counting detectors. The scanning scheme applied tube voltages of 90kVp, 120kVp, and 140kVp to the 90kVp, 120kVp, and 140kVp groups, respectively. The image quality IQ level for the low-dose group was set to 90, and the IQ level for the standard-dose group was set to 145. Automatic tube current modulation optimized for non-contrast enhancement was used. The collimator width was 144*0.4m, the rotation time was 0.25s, and the pitch coefficient was 0.

8. All images were reconstructed on the cross-section using the soft tissue reconstruction kernel Br40, with quantum iterative reconstruction, an intensity level of 3, a slice thickness of 1 mm, and an interval of 0.7 mm.

3. The method for constructing a non-invasive quantitative assessment model of liver fat based on PCD-CT as described in claim 2, characterized in that, In step S3, MRI-PDFF was measured on the volunteers using a 3.0T MRI with the following scanning parameters: TR = 6.4 ms; first echo time (TE) = 0.9 ms; 6 echoes with an interval (TE) of 0.8 ms; field of view = 440 × 330 mm; slice thickness = 3 mm; flip angle = 3°; matrix = 160 × 160 × 50; number of excitations = 0.5; and scan time = 18 s.

4. The method for constructing a non-invasive quantitative assessment model of liver fat based on PCD-CT as described in claim 3, characterized in that, In step S4, for each volunteer, three tissue samples with an area of ​​2-3 cm² were placed in the left lobe, right anterior lobe, and right posterior lobe of the liver, respectively. 2 Circular regions of interest were selected, avoiding major blood vessels and bile ducts, liver capsule, artifacts, and gas in the lungs or gastrointestinal tract; post-processing of standardized CT images was performed using syngo.via, version VB60, Siemens Healthineers; post-processing of MRI-PDFF images was performed using AW 4.7, GE Healthcare.

5. A non-invasive quantitative assessment system for liver fat based on PCD-CT, characterized in that, include: An input module is used to acquire standardized CT values, which are obtained based on PCD-CT scanning and reconstruction of CT scan images; A quantitative assessment module is used to evaluate MRI-PDFF results using a non-invasive quantitative assessment model of liver fat constructed according to any one of claims 1 to 4. Among them, MRI-PDFF is a magnetic resonance imaging assessment of proton density fat fraction, and its value is a percentage; standardized CT is a standardized CT obtained through PCD-CT technology, and the Hounsfield unit used is HU. The output module is used to output MRI-PDFF results and qualitative assessment results corresponding to the MRI-PDFF results according to the fatty liver diagnosis rules.

6. The PCD-CT-based non-invasive quantitative assessment system for liver fat as described in claim 5, characterized in that, The criteria for determining fatty liver are as follows: when the MRI-PDFF result is less than 5%, the volunteer is assessed as having no fatty liver; when the MRI-PDFF result is 5%-15%, the volunteer is assessed as having mild fatty liver; when the MRI-PDFF result is 15%-25%, the volunteer is assessed as having moderate fatty liver; and when the MRI-PDFF result is greater than 25%, the volunteer is assessed as having severe fatty liver.

Images