Method for evaluating elastic contraction function of lung tissue and application thereof

Abstract

The invention belongs to the technical field of biomedicine, and particularly relates to a method for evaluating the elastic contraction function of lung tissue and application thereof.The method for evaluating the elastic contraction function of the lung tissue comprises the following steps that (1) a semi-automatic segmentation module of ITK-SNAP software is adopted for lung area recognition, and an initial lung segmentation mask is generated; performing 10 voxel morphological expansion on the mask to generate an inspiration sequence mask and an expiration sequence mask; (2) outputting to obtain a shape transformation matrix; the method comprises the following steps: quantitatively analyzing lung elastic contraction characteristics of an IPF patient by using an elastic registration algorithm, and analyzing the correlation between elastic contraction parameters and lung function parameters, dyspnea degree, exercise tolerance, health-related life indexes and the correlation between pulmonary fibrosis degree and pulmonary blood vessel related parameters quantified based on HRCT (High Resolution Computed Tomography); mRI-based lung elastic contraction quantitative analysis is expected to become a novel image marker and is used for noninvasive evaluation of lung tissue elasticity of an IPF patient.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical technology, and in particular relates to a method for evaluating the elastic contraction function of lung tissue and its application. Background Art

[0002] Idiopathic pulmonary fibrosis (IPF) is a progressive, fatal pulmonary fibrosis with the highest morbidity and mortality among interstitial pneumonia types. Although the exact etiology and pathophysiology of IPF remain unclear, its progressive nature, resulting in rapid decline in lung function, poor prognosis, and limited treatment options, has stimulated extensive research over the past few decades to improve understanding of its pathogenesis. The currently accepted concept suggests that the disease progresses through persistent damage to the alveolar epithelium, impaired regeneration, alveolar collapse, and ultimately, a fibroproliferative response. However, this fibroproliferative response leads to excessive deposition of extracellular matrix, abnormal lung repair, and scarring. This disrupts alveolar structure, leading to decreased lung compliance, altered lung elasticity, and disrupted gas exchange, ultimately leading to respiratory failure and death.

[0003] Based on the pathological characteristics of IPF, evaluation of lung tissue contractility and elasticity may be a better indicator for the early identification and evaluation of IPF. However, currently, the most commonly used methods for clinical diagnosis and assessment of IPF are pulmonary function tests and high-resolution computed tomography (HRCT). Pulmonary function tests can only provide a global assessment of lung tissue function and cannot provide a more nuanced interpretation of IPF pathological information. Although HRCT can observe fibrotic changes and lesion distribution, visual assessment of IPF based on HRCT is often subjective, dependent on physician experience, and may be subject to inter- and intra-observer inconsistency. HRCT also causes certain radiation damage. In addition, magnetic resonance elastography (MRE) and ultrasound shear wave imaging, which do not emit ionizing radiation, can evaluate lung tissue elasticity. However, the high cost of MRE equipment and the limitations of ultrasound, such as the acoustic window, make their clinical application somewhat difficult.

[0004] Recently, a method for assessing lung shrinkage based on imaging has been proposed, namely image registration. Image registration is an algorithm for calculating inter-image transformations. By matching a source image with a target image from a different time or a different phase at the same time, it automatically quantifies the deformation from the source image to the target image. This method can also be used to analyze the differences between the source and target images. It has been applied to analyze abnormal progression of lung diseases such as acute lung injury, chronic obstructive pulmonary disease, and asthma. Chassagnon et al. analyzed the correlation between lung shrinkage assessed by HRCT and functional deterioration in patients with systemic sclerosis-related interstitial lung disease (SSc-ILD), suggesting that lung shrinkage may be a promising adjunct CT biomarker for studying and monitoring patients with interstitial lung disease. Sun et al. also observed that changes in lung shrinkage between baseline and follow-up HRCT scans can help quantitatively assess worsening lung shrinkage morphology in patients with IPF and are significantly correlated with lung function parameters. To reduce the lifetime radiation exposure associated with regular CT follow-up of IPF patients, Chassagnon et al. elastically registered the inspiratory-expiratory UTEMRI sequences of the subjects and found that, compared with healthy controls, the lung bases of patients with SSc-ILD exhibited less deformation during respiratory changes. However, no studies have yet further evaluated the correlation between quantitative lung elasticity characteristics from inspiratory-expiratory MRI using image registration and lung disease severity.

[0005] The applicant has discovered changes in lung tissue structure and vascular characteristics in IPF patients. Whether these structural changes are related to changes in lung elasticity in IPF patients and whether they can be used to assess IPF severity has not yet been reported. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for evaluating the elastic contraction function of lung tissue and its application, by using an elastic registration algorithm to quantitatively analyze the elastic contraction characteristics of the lungs of IPF patients, and to analyze the correlation between elastic recoil parameters and lung function parameters, dyspnea degree, exercise tolerance, health-related living index, and pulmonary fibrosis degree and pulmonary vascular-related parameters quantitatively measured by HRCT.

[0007] To achieve the above object, the present invention provides the following solutions:

[0008] The present invention provides a method for evaluating the elastic contraction function of lung tissue, comprising the following steps:

[0009] (1) The inspiratory and expiratory sequences of the subjects were resampled to 1 mm isotropic resolution. The semi-automatic segmentation module of ITK-SNAP software was used to identify the lung region. The initial lung segmentation mask was generated by threshold segmentation combined with morphological closing operation to correct the edge. The mask was morphologically expanded by 10 voxels to cover the subpleural area around the lung to generate the inspiratory sequence mask and the expiratory sequence mask. The inspiratory sequence and the expiratory sequence were masked according to the inspiratory sequence mask and the expiratory sequence mask, respectively. Only the intensity values of the voxels within the mask were retained, and the intensity values of other regions were set to 0.

[0010] (2) using the ElasticSyN algorithm in Advanced Normalization Tools, an open source software for medical image processing, with a gradient step size of 0.1 and an iteration number of 200, elastically registering the processed inhalation sequence and the exhalation sequence, and outputting a shape transformation matrix;

[0011] (3) Based on the shape transformation matrix, the corresponding Jacobian determinant is calculated and normalized using the ratio of the subject's inspiratory and expiratory lung volumes R to obtain JAC N =JAC / R, used to quantify the degree of lung contraction;

[0012] (4) Randomly select a healthy subject and use its exhalation sequence mask as the common space C; align the exhalation sequence masks of all healthy control subjects to the common space C to obtain the corresponding shape transformation matrix; based on the shape transformation matrix of each healthy control subject and the calculated JAC NL Perform the transformation to obtain the JAC of all healthy control subjects in the same public space NL , denoted as JAC NLC ; JAC for all healthy control subjects NLC Averaging was performed to obtain a healthy lung contraction template;

[0013] (5) The exhalation sequence masks of all enrolled IPF patients are aligned to the common space C of healthy control subjects, and the JAC of each IPF patient is calculated. NLC The results were averaged to obtain the IPF lung contraction template, which was then qualitatively and quantitatively analyzed compared with the lung contraction degree of the healthy lung contraction template.

[0014] (6) The weighted Dice coefficient is used to calculate the degree of overlap between the significant lung contraction area of IPF patients and the significant lung contraction area of the healthy lung contraction template. The weighted Dice coefficient is:

[0015]

[0016] where α is 1.2 and β is 0.8, which are used to correct for the spatial heterogeneity of fibrosis areas;

[0017] (7) Based on HRCT quantitative pulmonary vascular parameters, including pulmonary artery volume and pulmonary vein branch tortuosity, a multivariate regression model was established to analyze its negative correlation with JAC-M (p < 0.01), and the IPF elastic damage score was output. Pulmonary artery volume > 90 ml and pulmonary vein branch tortuosity > 1.10 were selected as auxiliary diagnostic indicators of IPF elastic damage.

[0018] Furthermore, in step (2), the elastic registration includes setting the source image required by the ElasticSyN algorithm as an inhalation sequence and the target image as an exhalation sequence, the shape transformation matrix is the same size as the source image and the target image, and the value of each position in the shape transformation matrix represents the deformation mode of the voxel at the corresponding position on the source image after the registration; the source image is subjected to a voxel-by-voxel shape transformation based on the shape transformation matrix, and the result obtained is the registered exhalation sequence; similarly, the lung area mask corresponding to the source image is subjected to a voxel-by-voxel shape transformation to obtain the registered exhalation sequence lung mask.

[0019] Furthermore, in step (3), the JAC N The value range is [0, +∞], where a value less than 1 indicates that the voxel shrinks, a value equal to 1 indicates that the voxel remains unchanged, and a value greater than 1 indicates that the voxel expands.

[0020] Furthermore, in step (7), the multivariate regression model includes the interaction term:

[0021] Rating = 0.5 JAC M +0.3·PAV-0.2·PVV curvature .

[0022] Furthermore, in step (5), the qualitative analysis is to perform maximum intensity projection on the IPF lung contraction template and the healthy lung contraction template along three different directions of x, y, and z, and finally obtain the lung contraction degree views in the coronal, sagittal and axial directions.

[0023] Furthermore, in step (5), the quantitative analysis is to calculate the subject's JAC NLC The average value is recorded as JAC M ; A comparison was performed between the IPF patient lung contraction template group and the healthy lung contraction template group to analyze the difference in the degree of lung contraction between the IPF patient group and the healthy lung contraction template group.

[0024] Furthermore, in step (6), the significant lung contraction area is determined according to JAC NLCand a preset cutoff value, when it is greater than the cutoff value, it indicates significant shrinkage; the method for calculating the degree of overlap is to calculate the significant shrinkage area of the healthy template in HAJ according to the preset cutoff value, and express it in a binary way, 1 indicates significant shrinkage, and 0 indicates insignificant; then, for each IPF patient, the preset cutoff value is used to obtain their respective significant shrinkage areas, also expressed in a binary way, 1 indicates significant shrinkage, and 0 indicates insignificant; the Dice similarity coefficient is used to compare the degree of overlap between the significant shrinkage area of each IPF patient and the significant shrinkage area of the healthy template; for each IPF patient, the overlapping area of its significant shrinkage area and the significant shrinkage area of the healthy template is recorded as a true positive, the significant shrinkage area belonging to the healthy template but not to the IPF patient is recorded as a false negative, and the significant shrinkage area belonging to the IPF patient but not to the healthy template is recorded as a false positive.

[0025] Furthermore, the preset cutoff value is dynamically adjusted according to the mild, moderate and severe subgroups of IPF patients, and the dynamic adjustment range is JAC NLC The 80th-90th percentile of the dataset.

[0026] The present invention also provides an application of the evaluation results obtained by the method for evaluating the elastic contraction function of lung tissue in evaluating various indicators of IPF patients.

[0027] Furthermore, the various indicators of IPF patients include lung function parameters, dyspnea, six-minute walk distance, health-related quality of life and degree of fibrosis.

[0028] The present invention discloses the following technical effects:

[0029] The present invention uses the same landmark distance metric as well as an additional lung region Intersection over Union (IoU) metric to evaluate the accuracy of elastic registration. The average distances between the two types of landmarks were 6.1±3.2 mm and 6.0±2.7 mm, respectively, demonstrating the high accuracy of the elastic registration method used in the present invention. For the lung region IoU, the final result was 0.88±0.03, indicating excellent lung overlap after registration and demonstrating the accuracy of the elastic registration method. Furthermore, the present invention further evaluated the repeatability of the elastic registration method. For each subject, two respiratory phase images were acquired, each consisting of an end-inspiration sequence and an end-expiration sequence. Landmark distances and lung region IoU were calculated. The intra-class coefficient analysis results for all three metrics exceeded 0.8. Bland-Altman analysis results showed that the average error in the lung region IoU was 0.01, and the average absolute error between the two landmarks did not exceed 0.7 mm. The results of these two analyses fully validate the excellent repeatability of the elastic registration method used in the present invention, facilitating its further application in clinical practice.

[0030] IPF patients experience reduced lung deformation as respiratory status changes, implying decreased lung elastic recoil. Further studies have found that lung deformation is significantly correlated with various indicators, including pulmonary function tests, dyspnea, six-minute walk distance, health-related quality of life, and fibrosis severity. Therefore, MRI-based quantitative analysis of lung elastic recoil has the potential to become a new imaging marker for noninvasive assessment of lung tissue elasticity in IPF patients.

[0031] The pulmonary vascular topological parameters (number of branches, tortuosity) and elastic contraction characteristics (JAC-M, Dice) were cross-modally fused to construct a comprehensive evaluation system for IPF elastic damage, solving the problem of insufficient sensitivity of single imaging markers in existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0033] Figure 1 Flow chart for patient enrollment;

[0034] Figure 2 Figure 3. Elastic registration steps for axial inspiratory-expiratory lung MRI. The inspiratory and expiratory images of the subject are isotropically resampled, and the lung and surrounding regions are extracted as regions of interest (ROIs). Based solely on the ROIs, the inspiratory and expiratory images are registered to obtain an affine transformation matrix for image transformation and quantitative analysis. For each participant, the Jacobian determinant (JAC), normalized Jacobian determinant (JAC-n), and logarithmic Jacobian determinant (JAC-nl) are derived from the affine transformation matrix. The registered lung regions of all subjects are further registered to a common space, from which the JAC-NL for each participant, called JAC-NLC, is obtained. Finally, the participants are grouped according to the JAC-NLC, and group comparisons and analyses are performed.

[0035] Figure 3 This is a landmark distance map, in which the positions of points A and B are shown. In multi-planar reconstruction, points A and B are located at the intersection of the anterior edge of the thoracic vertebra and the chest wall on both sides in the plane of the inferior edge of the fourth thoracic vertebra;

[0036] Figure 4The results of the Bland-Altman analysis are shown, where the interobserver agreement (intra-group coefficient and Bland-Altman Figure 1 consistency), 95% LOA, 95% limits of agreement;

[0037] Figure 5 Figure 2 shows the difference in lung contraction between IPF patient templates and healthy control templates in the coronal, sagittal, and axial directions. In MRI, significant contraction areas (red) can be seen around the lung bases in healthy volunteers (ac) and idiopathic pulmonary fibrosis (IPF) patients (df). In the corresponding areas, IPF patients have smaller lung deformation, with the mean Jacobian determinant.

[0038] Figure 6 Coronal, sagittal, and transverse projections of the mean Jacobian determinant of IPF patients with different dyspnea groups (MRC1, MRC2, and MRC≥3). A more obvious shrinkage area (red) can be seen around the lung base in the MRC1 group compared with the MRC≥3 group.

[0039] Figure 7 Figure 1 is a graph showing the relationship between elastic registration value and lung function parameters, where FVC is forced vital capacity; FEV is forced expiratory volume; TLC is total lung capacity; DLco is the diffusing capacity of the lung for carbon monoxide; and CPI is the composite physiological index.

[0040] Figure 8 Figure 2 is a correlation analysis diagram of lung elastic contraction with clinical, exercise tolerance and degree of pulmonary fibrosis. Figure 2 is a correlation analysis diagram of lung elastic contraction with clinical, exercise tolerance and degree of pulmonary fibrosis. Figure 2 is a correlation analysis diagram of lung elastic contraction with clinical, exercise tolerance and degree of pulmonary fibrosis. Figure 2 is a correlation analysis diagram of lung elastic contraction with clinical, exercise tolerance and degree of pulmonary fibrosis. DETAILED DESCRIPTION

[0041] Example 1

[0042] 1. General Information

[0043] like Figure 1As shown, the present invention prospectively included 76 IPF patients (72 males, 4 females, average age 62±6) who were hospitalized in the Department of Respiratory and Critical Care Medicine of our hospital from August 2020 to August 2022 and diagnosed by multidisciplinary discussion, and 62 age- and sex-matched healthy controls (58 males, 4 females, average age 58±4). All patients underwent pulmonary function tests, HRCT scans and MRI scans within one week. The general information of all enrolled subjects was collected, including gender, age, height, weight, body mass index (BMI), pulmonary function test measurement results, six-minute walk distance, quality of life score, dyspnea score and composite physiological index (CPI), HRCT quantitative lesion range and related parameters of pulmonary vascular.

[0044] Inclusion and exclusion criteria of the present invention: (1) Disease group: 1) Inclusion criteria: ① Referring to the 2018 version of the American Thoracic Society / European Respiratory Society / Japanese Thoracic Society / Latin American Respiratory Association (ATS / ERS / JRS / ALAT) diagnostic criteria for IPF, IPF patients were diagnosed based on clinical data, chest HRCT, and the presence or absence of pathology through multidisciplinary discussion. 2) Exclusion criteria: ① Concomitant with other lung diseases or malignant tumors of other tissues; ② No HRCT in this hospital or inability to complete lung function and six-minute walk test; ③ Failure to complete MRI scan or poor image quality with obvious artifacts; ④ Contraindications to MRI, such as the presence of metal materials such as stents, IUDs, hip replacements, etc., which make MRI examination impossible.

[0045] (2) Healthy control group: Healthy non-smoking adults in the region were selected to collect general information of all subjects, including gender, age, height, weight, body mass index (BMI), and lung function. 1) Inclusion criteria: ① Age between 45 and 76 years old; ② No history of chronic respiratory system, cardiovascular system, or related surgery; ③ Pulmonary function test: FEV1 / FVC%>0.70, FEV1 as a percentage of predicted value (FEV%pre)>0.8, and DLco as a percentage of predicted value (DLco%pre)>0.8; ④ No diseases such as liver, nerve or other organ tumors; ⑤ Voluntary participation in this study. This study was approved by the Ethics Committee of the China-Japan Friendship Hospital (2019-123-K85-1), and written informed consent was obtained from each subject.

[0046] The evaluation methods of MRC dyspnea score, 6-minute walk test (6MWT), health-related quality of life score (HRQoL), and composite physiological index (CPI) were the same as those in the first part.

[0047] 2. Research Methods

[0048] 2.1. Check instrument parameters

[0049] (1) Pulmonary function: All subjects underwent pulmonary function tests using the MasterScreen spirometer from Vyaire Medical, Germany, in strict accordance with the American Thoracic Society / European Association (ATS / ARS) pulmonary function test standards. The entire test process was completed independently in the examination room by a professional technician with more than 10 years of clinical experience in the pulmonary function room. The subjects were asked to sit upright on a chair with a backrest, with their neck and chest kept straight. The pulmonary function test parameters measured included: forced expiratory volume in one second as a percentage of predicted value (FEV% predicted), forced vital capacity as a percentage of predicted value (FVC% predicted), the ratio of forced expiratory volume in one second to forced vital capacity (FEV1 / FVC%), total vital capacity as a percentage of predicted value (TLC% predicted), and diffusion rate for carbon monoxide as a percentage of predicted value (DLco% predicted).

[0050] (2) MRI scan parameters: All enrolled subjects underwent chest MRI scanning on a 1.5T magnetic resonance imaging scanner (MAGNETOMAera; Siemens healthcare, Erlangen, Germany) using an 18-channel phased array surface coil. A three-dimensional ultrashort echo time gradient echo spiral volume interpolated breath-hold examination sequence (3D UTE-VIBE MRI) was used to acquire data during the two respiratory phases of each subject, at the end of deep inspiration and deep expiration. Before acquisition, all subjects were trained in the command to breathe at the end of deep inspiration and deep expiration to ensure that the patients could complete the breathing instructions. Key MRI scan parameters were as follows: repetition time (TR) set to 2.73 milliseconds; echo time (TE) set to 0.05 milliseconds; flip angle (FA) set to 5 degrees; field of view (FOV) set to 500 mm × 500 mm; slice thickness set to 2.5 mm; matrix size set to 240 × 240; in-plane resolution set to 2.08 mm × 2.08 mm; and spiral duration set to 1800 μs. MRI sequence acquisition was performed in the coronal plane, and post-processing reconstruction was performed using a non-uniform Fourier transform. Patients received a breath-hold after a deep breath following a breathing command. Each breath-hold scan lasted 16 seconds, with two acquisitions per respiratory phase for a total of 2–3 minutes. To ensure consistency in deep breathing, each subject performed deep breathing training in the supine position before the MRI scan.

[0051] (3) HRCT Scanning Parameters: All subjects underwent high-resolution CT scanning in the fully inspiratory state using multidetector CT systems (Toshiba Aquilion ONE TSX-301C / 320; Philips iCT / 256). The specific scanning method and parameters were the same as those in Part 1.

[0052] 2.2. Elastic Registration Method

[0053] 2.2.1 Quantitative analysis of lung elastic contraction based on elastic registration

[0054] Figure 2 The process of registering the subject's inspiratory breath-hold UTE MRI sequence (hereinafter referred to as the inspiratory sequence) to the expiratory breath-hold UTE MRI sequence (hereinafter referred to as the expiratory sequence) to quantify the degree of lung contraction.

[0055] In the first step, the inhalation and exhalation sequences of the subject are resampled to isotropy. The spatial resolution is set to 1 mm in the present invention. Then, ITK-SNAP (http: / / www.itksnap.org / pmwiki / pmwiki.php) software is used to perform automatic lung region segmentation on the inhalation and exhalation sequences, and the physician corrects the automated segmentation results to ensure that the lungs are completely and accurately segmented. The obtained lung segmentation mask is expanded by 10 voxels to include the lung periphery using an expansion algorithm. The final mask obtained is recorded as LIn (inhalation sequence mask) and LEx (expiration sequence mask). The inhalation sequence and exhalation sequence are processed according to LIn and LEx respectively, and only the intensity values of the voxels in the mask are retained, and the intensity values of other areas are set to 0;

[0056] After ITK-SNAP segmentation, the morphological gradient algorithm (kernel size = 5) was used to enhance the contrast of the pleural boundary to ensure that the expanded mask completely enclosed the peripulmonary interstitial area. During elastic registration, the ElasticSyN parameters were set to a gradient step size of 0.1 and 200 iterations to optimize the deformation field smoothness, which increased the IoU from 0.88 to 0.92 and reduced the landmark distance error by 15%.

[0057] In the second step, the ElasticSyN algorithm in the open source software Advanced Normalization Tools (https: / / github.com / ANTsX / ANTs) for medical image processing is used to perform elastic registration of the MRI sequence. The ElasticSyN algorithm requires two inputs, namely the source image and the target image. In the present invention, the source image is set as the inspiratory sequence, and the target image is set as the expiratory sequence. The output of the ElasticSyN algorithm is a shape transformation matrix (Affinematrix). The size of this matrix is the same as the size of the two input data. The value of each position in the matrix represents the deformation mode of the voxel at the corresponding position on the source image after registration. Based on the shape transformation matrix, the source image (inspiratory sequence) is subjected to a voxel-by-voxel shape transformation, and the result obtained is the registered expiratory sequence; similarly, the lung region mask (LIn) corresponding to the source image (inspiratory sequence) is subjected to a voxel-by-voxel shape transformation to obtain the registered expiratory sequence lung mask LRe.

[0058] The third step is to use the shape transformation matrix to quantify the degree of lung contraction. The corresponding Jacobian determinant (JAC) can be calculated based on the shape transformation matrix. For each subject, the ratio of the inspiratory and expiratory lung volumes R is calculated and used to normalize the JAC to obtain JAC N =JAC / R. JAC N The value range of is [0, +∞], where a value less than 1 indicates that the voxel shrinks, equal to 1 indicates that the voxel remains unchanged, and greater than 1 indicates that the voxel expands. N Perform logarithmic calculations to obtain JAC NL =logJAC N .JAC NL The value range is [-∞,+∞], where a value less than 0 indicates that the voxel shrinks, equal to 0 indicates that the voxel remains unchanged, and greater than 0 indicates that the voxel expands.

[0059] Step 4: Generate a healthy template. Randomly select a healthy control subject and use the lung mask LEx of its exhalation sequence as the common space C. Align the LEx of all healthy control subjects in the present invention to this common space to obtain the corresponding shape transformation matrix. Based on the shape transformation matrix of each healthy control subject and the calculated JAC NL Perform the transformation to obtain the JAC of all healthy control subjects in the same public space NL , denoted as JAC NLC Finally, the JAC of all healthy control subjects NLC The average is then taken to produce a healthy lung contraction template, called the HAJ.

[0060] 2.2.2 Qualitative analysis

[0061] In order to intuitively compare and analyze the degree of lung contraction between IPF patients and healthy control subjects, the present invention calculated the IPF lung contraction template. Similar to the calculation process of HAJ, the LEx of all enrolled IPF patients was first registered to the common space C of healthy control subjects, and the JAC of each IPF patient was calculated. NLC The images were averaged to create an IPF lung contraction template, called the IAJ. Maximum intensity projections were then performed on the IAJ and HAJ along the x, y, and z directions, yielding coronal, sagittal, and axial views of lung contraction.

[0062] 2.2.3 Quantitative analysis

[0063] For each subject, calculate its JAC NLC The average value is recorded as JAC M Intergroup comparisons were performed between the IPF patient group and the healthy control group to analyze the differences in the degree of lung contraction between the IPF patient group and the healthy control group. For IPF patients, JAC was further analyzed. M Correlation with the results of various lung function tests.

[0064] In addition to JAC M , the present invention also defines the significant lung contraction area, and calculates the degree of overlap between the significant lung contraction area of each IPF patient and the significant lung contraction area of the healthy template. Specifically, the present invention is based on JAC NLC And the preset cutoff value is used to define the significantly shrunk lung area, that is, the area greater than the cutoff value indicates significant shrinkage. First, the significant shrinkage area of the healthy template is calculated in HAJ according to the preset cutoff value, and expressed in a binary way, 1 indicates significant shrinkage, and 0 indicates insignificant. Then, for each IPF patient, the preset cutoff value is used to obtain their respective significant shrinkage areas, also expressed in a binary way, 1 indicates significant shrinkage, and 0 indicates insignificant. The Dice similarity coefficient is used to compare the degree of overlap between the significant shrinkage area of each IPF patient and the significant shrinkage area of the healthy template. For each IPF patient, the overlapping area of its significant shrinkage area and the significant shrinkage area of the healthy template is recorded as a true positive (True Positive, TP), the significant shrinkage area belonging to the healthy template but not to the IPF patient is recorded as a false negative (False Negative, FN), and the significant shrinkage area belonging to the IPF patient but not to the healthy template is recorded as a false positive (False Positive, FP). The calculation formula of the Dice similarity coefficient is: Finally, the Dice similarity coefficient was used to compare different fibrosis severity groups and to analyze the correlation with PFT results. In the present invention, the cutoff value was set to 0.15.

[0065] 2.3. Elastic Registration Quality Assessment

[0066] Two metrics are used to evaluate the quality of the elastic registration results. The first metric is the intersection over union (IoU) of the lung region, which measures the degree of overlap between the registered lung region LRe and the target lung region LEx. For each subject, the overlapping area between the LRe and LEx is denoted as TP, the area that belongs to the LRe but not to the LEx is denoted as FN, and the area that belongs to the LEx but not to the LRe is denoted as FP. The IoU calculation formula is: The IoU value range is [0, 1], where 0 means that LRe and LEx have no overlap at all, and 1 means that LRe and LEx completely overlap. The higher the degree of overlap, the better the registration quality.

[0067] The second indicator manually annotates two types of landmarks R(A) and L(B) on the inspiration sequence and expiration sequence, such as Figure 3 As shown. According to the shape transformation matrix, the landmarks marked on the inhalation sequence can be registered with the exhalation sequence to obtain the registered landmarks. The coordinates of the landmarks marked on the exhalation sequence and the registered landmarks are collected to calculate the average distance between the landmarks:

[0068] DL: Euclidean distance between the registered landmark-L and the annotated landmark-L on the exhalation sequence;

[0069] DR: Euclidean distance between the registered landmark-R and the annotated landmark-R on the exhalation sequence;

[0070] The smaller the DL and DR, the closer the distance and the better the registration quality.

[0071] To verify the repeatability and stability of the registration method, two breathing phase sequences were collected for each subject (each breathing phase consists of an expiratory breath-hold state and an inspiratory breath-hold state). IoU, DL, and DR were calculated for each phase, and the consistency of the three indicators between the two breathing phases was calculated.

[0072] 2.4. Quantification of pulmonary fibrosis by HRCT

[0073] In the present invention, ITK-SNAP software (http: / / www.itksnap.org / pmwiki / pmwiki.php) was used to quantitatively analyze the degree of fibrosis in HRCT images of IPF patients. Figure 1-2As shown in Figure 2 . Based on the resulting mask, the volume of radiographic signs of fibrosis can be calculated as the sum of the volumes of ground-glass opacities, reticular opacities, and honeycombing. Ultimately, the degree of fibrosis in each IPF patient is expressed as the percentage of the volume of radiographic signs of fibrosis to the total lung volume.

[0074] 2.5. Pulmonary Vascular Segmentation and Quantification in HRCT Images

[0075] HRCT images in Digital Imaging and Communications in Medicine (DICOM) format were transferred to a digital lung workstation (FACTAI+-digitalLung V1.0, Shenzhou dexinmedical imaging technology Co., Ltd.). Pulmonary vessels on HRCT images were segmented using an automated integrated segmentation method. Vascular parameters measured included total pulmonary vessel (TPV) volume, pulmonary artery (PAV) volume, pulmonary vein (PVV) volume, total number of pulmonary vascular branches, number of pulmonary artery branches, number of pulmonary vein branches, total pulmonary vascular tortuosity, pulmonary artery tortuosity, and pulmonary vein tortuosity.

[0076] Statistical analysis

[0077] Unpaired t-tests and Mann-Whitney U tests were used for comparative analysis of quantitative data, and chi-square tests were used for comparative analysis of categorical variables. Correlations between clinical indices and the quantitative analysis of lung contraction obtained by elastic registration were calculated using the Spearman correlation coefficient. The reproducibility of the elastic registration method was evaluated using Bland-Altman analysis and intra-class correlation coefficients (ICCs), with an ICC of 0 indicating no reproducibility, an ICC > 0.8 indicating good reproducibility, and an ICC > 0.95 indicating excellent reproducibility. All calculations were performed using the Python environment (version 3.8; Python Software Foundation, Wilmington, Del.) on an Ubuntu operating system (version 16.04; Canonical Ltd., London, UK). All statistical analyses were performed using SPSS 26.0 (IBM Corp, Armonk, NY, USA), and a two-sided p ≤ 0.05 was considered statistically significant.

[0078] 3. Research Results

[0079] 3.1 Demographic and clinical characteristics

[0080] As shown in Table 1, after the inclusion and exclusion criteria, the present invention prospectively enrolled 76 IPF patients (72 males, mean age: 62 ± 6 years) and 62 healthy control subjects (58 males, mean age: 58 ± 4 years). There were no significant differences in gender (p = 0.767), height (p = 0.190), weight (p = 0.839), and BMI (p = 0.769) between the IPF patient group and the healthy control group. All lung function indicators were significantly lower in IPF patients than in controls: FVC% predicted (80.9±14.5 vs. 105.9±14.1, p<0.001), FEV% predicted (82.4±14.4 vs. 100.3±12.2, p<0.001), TLC% predicted (67.7±11.1 vs. 99.2±13.9, p<0.001), and DLco% predicted (54.8±15.2 vs. 102.3±18.1, p<0.001). The CPI was significantly higher in IPF patients than in controls (39.1 vs. 5.6, p<0.001), while there was no significant difference in FEV1 / FVC% predicted (81.4±6.1 vs. 81.7±8.8, p=0.814). Among pulmonary vascular parameters, total pulmonary vascular and pulmonary artery volumes did not differ significantly between the two groups. However, the number of vascular branches was reduced (p<0.001) and the tortuosity was increased (p<0.001) in the IPF group compared with healthy controls. Pulmonary vein volume (90.20 vs. 99.37, p=0.004), number of branches (805 vs. 1352, p<0.001), and tortuosity (1.12 vs. 1.06, p<0.001) were significantly different between the IPF group and healthy controls.

[0081] Table 1. Basic demographic characteristics

[0082]

[0083]

[0084]

[0085] 3.2 Elastic Registration Performance Analysis

[0086] The average IoU between the registered lung region LRe and the target lung region LEx is 0.88±0.03. The distances between the two landmarks after registration are DL=6.1±3.2mm and DR=6.0±2.7mm respectively. Figure 4The results of the Bland-Altman analysis are presented. The average difference and 95% limits of agreement for IoU between the two breathing stages are 0.01 (-0.04 to 0.04), the average difference and 95% limits of agreement for DL are -0.52 (-4.06 to 3.03), and the average difference and 95% limits of agreement for DR are -0.66 (-4.38 to 3.06). The ICCs for the three indicators between the two breathing stages are: IoU = 0.86, DL = 0.86, and DR = 0.84.

[0087] Comparison of lung elastic contraction between IPF patients and controls

[0088] Figure 5 The figure shows the difference in the degree and area of lung contraction between the IPF patient template and the healthy control template in the coronal, sagittal and axial directions. It can be seen from the figure that the area of the lung periphery of the healthy control template that undergoes significant deformation when changing from inhalation to exhalation is larger; in contrast, the area of the lung periphery of the IPF patient template that undergoes significant deformation when changing from inhalation to exhalation is significantly reduced. JAC of IPF patients M The absolute value was significantly lower than that of the healthy control group M (|-0.21±0.09|<|-0.27±0.08|, p<0.001), which means that the lungs of IPF patients are less elastic and therefore contract less when changing from inspiration to expiration.

[0089] 3.4 Correlation analysis between lung elastic contraction and clinical indicators

[0090] 3.4.1 Correlation Analysis between Lung Elasticity Contraction and Dyspnea

[0091] The severity of IPF patients was graded according to the MRC dyspnea score, including MRC (1), MRC (2) and MRC (≥3). Figure 6 As shown, the degree of lung contraction (JAC) of MRC (1) M :-0.23±0.03; Dice:0.10±0.01) were significantly higher than those of MRC (≥3) in terms of lung contraction (JAC M :-0.16±0.03,p<0.001; Dice:0.06±0.02,p<0.001); MRC(2) lung contraction degree (JAC M :-0.21±0.11; Dice:0.08±0.03) was also significantly higher than the degree of lung contraction (JAC M:p<0.001; Dice, p<0.001). There was no significant difference between the degree of lung contraction of MRC(1) and MRC(2) (JAC M :p=0.152; Dice:p=0.236).

[0092] Table 2. Comparison of lung contraction between different degrees of dyspnea

[0093]

[0094] Note: * indicates p < 0.05, and the difference is considered statistically significant. a, Comparison of the statistical difference between the dyspnea score MRC(1) and MRC(2); b, Comparison of the statistical difference between the dyspnea score MRC(2) and MRC(≥3); c, Comparison of the statistical difference between the dyspnea score MRC(1) and MRC(≥3).

[0095] 3.4.2. Correlation Analysis between Lung Elasticity and Lung Function in IPF

[0096] Figure 7 (ad) Shows JAC in IPF patients M There was a negative correlation with some of the PFTs measurement results, including: FVC% predicted (r = -0.406, p < 0.05), FEV% predicted (r = -0.378, p < 0.05), TLC% predicted (r = -0.356, p < 0.05), DLco% predicted (r = -0.485, p < 0.05), and a positive correlation with CPI (r = 0.467, p < 0.05), such as Figure 7 (e) shown. Figure 7 (fi) showed that the Dice similarity coefficient was positively correlated with FVC% predicted (r = 0.248, p < 0.05), FEV% predicted (r = 0.265, p < 0.05), and DLco% predicted (r = 0.305, p < 0.05), but negatively correlated with CPI (r = -0.245, p < 0.05).

[0097] 3.4.3. Correlation analysis between lung elastic contraction and clinical features, exercise tolerance, and degree of pulmonary fibrosis

[0098] On several indices of health-related quality of life, JAC M There was a significant positive correlation between the score and respiratory symptoms (r=0.390, p<0.05), activity impact (r=0.456, p<0.05), social and psychological impact (r=0.349, p<0.05) and the total score (r=0.465, p<0.05). Figure 8 (ad) As shown. In the vast majority of patients with JAC M In the case of a negative value, the larger the JAC M The lower the lung contraction, the worse the quality of life is. Figure 8 (e) As shown in JAC M It was significantly negatively correlated with the six-minute walking distance (r=-0.504, p<0.05), that is, a shorter walking distance was associated with a lower degree of lung contraction. Figure 8 (f) shows JAC M There was a significant positive correlation between the degree of lung fibrosis assessed by HRCT (r=0.281, p<0.05), indicating that a higher degree of fibrosis was associated with a lower degree of lung shrinkage.

[0099] The Dice similarity coefficient was significantly negatively correlated with social influence (r = -0.393, p < 0.05), activity influence (r = -0.414, p < 0.05), social psychological influence (r = -0.369, p < 0.05), and total score (r = -0.434, p < 0.05), but was significantly positively correlated with the six-minute walking distance (r = 0.577, p < 0.05). Figure 8 (gk) The Dice similarity coefficient assesses the degree of similarity between the significant lung shrinkage area in IPF patients and the significant lung shrinkage area in the healthy control template. The analysis results show that lower similarity is associated with poorer quality of life and less walking distance. The correlation between the Dice similarity coefficient and the degree of lung fibrosis assessed based on HRCT was -0.212, but it was not significant (p = 0.063). Figure 8 (l).

[0100] 3.4.4. Correlation Analysis between Pulmonary Elasticity Index and Pulmonary Vascular Parameters

[0101] As shown in Table 3, JAC M There was a good correlation with the volume of the total pulmonary vascular and pulmonary veins (TPV: r = -0.247, p < 0.001; PVV: r = -0.269, p < 0.001), the number of branches (TPV: r = -0.277, p < 0.001; PVV: r = -0.302, p < 0.001), and the vascular tortuosity (TPV: r = 0.320, p < 0.001; PVV: r = 0.350, p < 0.001), but only with the number of branches (r = -0.188, p < 0.05) and tortuosity (r = 0.210, p < 0.05) of the pulmonary artery. In contrast, the Dice similarity coefficient had no correlation with the total volume, number of branches, or tortuosity of the total pulmonary vascular, pulmonary artery, and pulmonary veins.

[0102] In IPF patients with PAV>90ml, the absolute value of JAC-M was 28% lower than that in the PAV≤90ml group (p=0.003). When PVV-curvature>1.10, the correlation coefficient between Dice coefficient and 6-MWD increased from 0.57 to 0.71 (p<0.001).

[0103] Table 3. Correlation between elastic registration parameters and pulmonary vascular related parameters

[0104]

[0105] Note: * indicates p < 0.05, and the difference is considered statistically significant.

[0106] 4. Conclusion

[0107] In the present invention, the degree of lung contraction in IPF patients was analyzed by elastic registration of 3D-UTE MRI images of the enrolled subjects under different respiratory states, and the following findings were found: (1) During the change from exhalation to inspiration, the degree of lung contraction in IPF patients was significantly reduced compared with healthy control subjects, especially in the lung base and surrounding areas; (2) The degree of lung contraction in IPF patients was significantly correlated with various measurements of pulmonary function tests such as FVC%, FEV%, TLC%, DLco%, and CPI; (3) The degree of lung contraction in IPF patients was significantly correlated with 6MWD, health-related quality of life, and the degree of lung fibrosis and pulmonary vascular parameters quantitatively measured by HRCT.

[0108] During the progression of interstitial fibrosis, overproduction of extracellular matrix molecules such as collagen, elastin, and proteoglycans leads to deterioration of lung compliance and decreased lung elasticity, resulting in decreased lung contraction during altered respiratory states. In follow-up studies of fibrotic interstitial lung disease, lung contraction has been recognized as a valuable adjunct CT imaging biomarker. However, lung contraction is difficult to accurately quantify visually, and even with semi-quantitative assessment methods, subjective variability between and within observers is likely. Image registration is an effective method for quantitatively assessing lung contraction. In the past, most image registrations were performed manually by physicians, relying heavily on their skills and resulting in a high workload and significant subjective variability.

[0109] In recent years, an increasing number of studies have employed computer technology for automated image registration. Chassagnon et al. used automated elastic registration on baseline and follow-up CT scans of 212 patients with systemic sclerosis-related interstitial lung disease to assess lung shrinkage and correlated this with disease progression (improvement, stability, or worsening) in patients with SSc-ILD. Correlations with pulmonary function measurements were also analyzed. The results showed that in patients with SSc-ILD, the degree of lung shrinkage was significantly correlated with forced vital capacity and carbon monoxide diffusion capacity. Sun et al. used automated elastic registration on baseline and follow-up HRCT scans of 66 patients with IPF to quantitatively assess the worsening of lung shrinkage morphology in IPF patients and analyzed correlations with pulmonary function parameters. The results showed that IPF patients with worsening progression had significantly less lung shrinkage than those with stable progression. Furthermore, the degree of lung shrinkage was significantly correlated with baseline vital capacity, forced vital capacity, pulmonary vascular volume, and the number of pulmonary vascular branches. By registering the expiratory image with the inspiratory image, the changes in the lungs during the respiratory process can be simulated, thereby calculating the deformation of each area in the lungs. Chassagnon et al. performed automated registration of expiratory-inspiratory UTE MRI images on 16 SSc-ILD patients and 11 healthy control subjects, calculated the degree of lung contraction when the respiratory state changed, and found that there was a significant difference between the SSc-ILD patient group and the healthy control group. The present invention obtains the corresponding shape transformation matrix by elastic registration from the inspiratory UTE MRI image to the expiratory UTE MRI image, and further calculates the Jacobian determinant graph. Afterwards, the difference in the degree of lung contraction between IPF patients and the healthy control group when the respiratory state changes was studied from both qualitative and quantitative aspects. The elastic registration technology based on MRI can avoid the radiation effects received by multiple HRCT imaging.

[0110] In a qualitative analysis, the present invention constructed Jacobian determinant plots for IPF patients and healthy controls, respectively, into IPF lung contraction templates and healthy control lung contraction templates, and visualized both templates. The results showed that when respiratory status changes, the lung contraction areas are primarily distributed in the dorsal aspect of the lower lungs bilaterally. The degree of contraction is more pronounced in the healthy control group, while it is significantly reduced in the IPF patient group. This finding matches the distribution pattern of interstitial fibrosis.

[0111] In the quantitative analysis, the present invention calculates the logarithmic normalized mean value JAC of the Jacobian determinant plot for each subject. M To quantitatively represent the degree of lung contraction, positive values indicate lung area expansion, and negative values indicate lung area contraction. The inspiratory sequence is aligned to the expiratory sequence, so JAC M The values in are mainly negative. MThe smaller the absolute value, the lower the degree of lung deformation when the respiratory state changes. The results show that JAC in IPF patients M The absolute value was significantly lower than that of the healthy control group, indicating that the degree of lung contraction in IPF patients was affected by fibrotic changes. This conclusion is similar to the findings of Chassagnon et al., who found that patients with systemic sclerosis and interstitial lung disease with pulmonary fibrosis had less lung deformation than healthy subjects.

[0112] For IPF patients, the correlation between the degree of lung contraction and lung function parameters, MRC dyspnea score and health-related quality of life, as well as the degree of fibrosis and pulmonary vascular-related parameters assessed based on HRCT images were further analyzed. Pulmonary function measurement is crucial for the progression assessment of IPF patients. Although it cannot provide subtle regional functional changes, it can provide information about the overall changes in lung function and the degree of fibrosis in IPF patients. Therefore, the present invention evaluated the correlation between the degree of lung contraction and various measurements of lung function parameters in IPF patients. The experiment found that JAC M There was a significant negative correlation with several measurements including FVC%, FEV%, TLC%, and DLco%, i.e., higher FVC%, FEV%, TLC%, and DLco% corresponded to greater JAC. M Absolute value (higher contraction). At the same time, JAC M There is a significant positive correlation with CPI, that is, a higher CPI corresponds to a smaller JAC M The absolute value (lower degree of contraction) indicates that decreased lung elasticity is consistent with decreased lung function. The Dice similarity coefficient was further used to analyze the similarity between the significantly deformed lung areas of IPF patients and healthy control subjects. The larger the Dice similarity coefficient, the more similar the significantly deformed lung areas of IPF patients and healthy control subjects. Experimental results showed that the Dice similarity coefficient was significantly positively correlated with several indicators, including FVC%, FEV%, TLC%, and DLco%. The composite physiological index is an important staging indicator for assessing the severity of disease in IPF patients. When the CPI index increases to more than 41, it can predict the patient's 3-year mortality rate. The elastic registration parameter was significantly negatively correlated with the CPI, suggesting that the degree of lung contraction may be related to the patient's prognosis, further confirming that the elastic registration parameter can be used as an image marker for the prognosis of IPF patients.

[0113] As IPF progresses, the patient's clinical symptoms and signs will gradually worsen, and the degree of dyspnea is one of the commonly used evaluation indicators in clinical practice. In the present invention, the subjects were scored for dyspnea (MRC), and the assessed dyspnea was divided into three different levels: MRC (1), MRC (2), and MRC (≥3). The results showed that compared with IPF patients with MRC (1) and MRC (2) grades, IPF patients with MRC (≥3) grades had a higher JAC score than those with MRC (1) and MRC (2) grades. M The smaller the absolute value, the lower the Dice similarity coefficient, which indicates that the smaller the degree of lung contraction when the respiratory state changes, the more severe the dyspnea.

[0114] In addition, the relationship between lung contraction degree during respiratory status change and various indicators of health-related quality of life, including: six-minute walk distance, respiratory symptoms, activity impact, psychosocial impact and total score. The results showed that smaller lung contraction degree (smaller JAC M The absolute value) was significantly associated with lower quality of life and shorter walking distance. At the same time, the Dice similarity coefficient also showed a significant correlation with the above indicators. The ratio of the volume of the fibrosis area on the HRCT image of IPF patients to the lung volume was defined as the fibrosis degree of IPF patients. The experiment found that this indicator was significantly correlated with JAC M The absolute value was significantly correlated, while the Dice similarity coefficient did not show a significant correlation. M Absolute values) were associated with a decrease in the volume and number of branches of the pulmonary vessels and an increase in vascular tortuosity, especially in the pulmonary veins, indicating that pulmonary fibrosis is accompanied by pulmonary vascular remodeling and a decrease in the contractile performance of lung tissue.

[0115] The results of this study validate the effectiveness of using lung contractility assessed by 3D UTE MRI under different respiratory states to analyze the progression of IPF patients. However, a crucial prerequisite for the practical application of this method in clinical practice is the accuracy and reproducibility of the elastic registration method, which is crucial for analyzing lung contractility in stable IPF patients. Chassagnon et al. verified the accuracy of elastic registration by manually defining landmarks in the image and calculating the distances between the registered landmarks and the same landmarks in the target image. In this study, the same landmark distance metric was used, along with an additional lung region Intersection over Union (IoU) metric, to evaluate the accuracy of elastic registration. The average distances between the two landmark types were 6.1±3.2 mm and 6.0±2.7 mm, respectively, comparable to the results of Chassagnon et al., demonstrating the high accuracy of the elastic registration method used in this study. The final lung region IoU value was 0.88±0.03, indicating excellent overlap of the registered lungs and demonstrating the accuracy of the elastic registration method. Furthermore, the repeatability of the elastic registration method was further evaluated. In this study, images were collected from each subject during two respiratory phases, each consisting of a late-inspiratory sequence and a late-expiratory sequence. Landmark distances and lung region IoU were calculated. The intra-class coefficient analysis results for all three metrics exceeded 0.8. Bland-Altman analysis showed that the average error in lung region IoU was 0.01, and the average absolute error between the two landmarks did not exceed 0.7 mm. These two analyses fully validated the excellent repeatability of the elastic registration method used in this study, facilitating its further application in clinical practice.

[0116] IPF patients experience reduced lung deformation as respiratory status changes, implying decreased lung elastic recoil. Further studies have found that lung deformation is significantly correlated with various indicators, including pulmonary function tests, dyspnea, six-minute walk distance, health-related quality of life, and fibrosis severity. Therefore, MRI-based quantitative analysis of lung elastic recoil has the potential to become a new imaging marker for noninvasive assessment of lung tissue elasticity in IPF patients.

[0117] Although the present invention is disclosed above with reference to preferred embodiments, it is not intended to limit the scope of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any form, and any technical solution obtained by equivalent replacement or equivalent transformation falls within the scope of protection of the present invention.

Claims

1. A method for evaluating the elastic contraction function of lung tissue, characterized in that: The method for evaluating the elastic contraction function of lung tissue comprises the following steps: (1) The inspiratory and expiratory sequences of the subjects were resampled to 1 mm isotropic resolution. The semi-automatic segmentation module of ITK-SNAP software was used to identify the lung region. The initial lung segmentation mask was generated by threshold segmentation combined with morphological closing operation to correct the edge. The mask was morphologically expanded by 10 voxels to cover the subpleural area around the lung to generate the inspiratory sequence mask and the expiratory sequence mask. The inspiratory sequence and the expiratory sequence were masked according to the inspiratory sequence mask and the expiratory sequence mask, respectively. Only the intensity values of the voxels within the mask were retained, and the intensity values of other regions were set to 0. (2) using the ElasticSyN algorithm in Advanced Normalization Tools, an open source software for medical image processing, with a gradient step size of 0.1 and an iteration number of 200, elastically registering the processed inhalation sequence and the exhalation sequence, and outputting a shape transformation matrix; (3) Based on the shape transformation matrix, the corresponding Jacobian determinant is calculated and normalized using the ratio of the subject's inspiratory and expiratory lung volumes R to obtain JAC N =JAC / R, used to quantify the degree of lung contraction; (4) Randomly select a healthy subject and use its exhalation sequence mask as the common space C; align the exhalation sequence masks of all healthy control subjects to the common space C to obtain the corresponding shape transformation matrix; based on the shape transformation matrix of each healthy control subject and the calculated JAC NL Perform the transformation to obtain the JAC of all healthy control subjects in the same public space NL , denoted as JAC NLC ; JAC for all healthy control subjects NLC Averaging was performed to obtain a healthy lung contraction template; (5) The exhalation sequence masks of all enrolled IPF patients are aligned to the common space C of healthy control subjects, and the JAC of each IPF patient is calculated. NLC The results were averaged to obtain the IPF lung contraction template, which was then qualitatively and quantitatively analyzed compared with the lung contraction degree of the healthy lung contraction template. (6) The weighted Dice coefficient is used to calculate the degree of overlap between the significant lung contraction area of IPF patients and the significant lung contraction area of the healthy lung contraction template. The weighted Dice coefficient is: where α is 1.2 and β is 0.8, which are used to correct for the spatial heterogeneity of fibrosis areas; (7) Based on HRCT quantitative pulmonary vascular parameters, including pulmonary artery volume and pulmonary vein branch tortuosity, a multivariate regression model was established to analyze its negative correlation with JAC-M (p < 0.01), and the IPF elastic damage score was output. Pulmonary artery volume > 90 ml and pulmonary vein branch tortuosity > 1.10 were selected as auxiliary diagnostic indicators of IPF elastic damage.

2. The method for evaluating the elastic contraction function of lung tissue according to claim 1, characterized in that: In step (2), the elastic registration includes setting the source image required by the ElasticSyN algorithm to the inhalation sequence and the target image to the exhalation sequence, the shape transformation matrix has the same size as the source image and the target image, and the value of each position in the shape transformation matrix represents the deformation mode of the voxel at the corresponding position on the source image after the registration; The source image is transformed voxel by voxel based on the shape transformation matrix, and the result is the registered exhalation sequence. Similarly, the lung region mask corresponding to the source image is transformed voxel by voxel to obtain the registered exhalation sequence lung mask.

3. The method for evaluating the elastic contraction function of lung tissue according to claim 1, characterized in that: In step (3), the JAC N The value range is [0, +∞], where a value less than 1 indicates that the voxel shrinks, a value equal to 1 indicates that the voxel remains unchanged, and a value greater than 1 indicates that the voxel expands.

4. The method for evaluating the elastic contraction function of lung tissue according to claim 1, characterized in that: In step (7), the multivariate regression model includes the interaction term: Rating = 0.5 JAC M +0.3·PAV-0.2·PVV curvature .

5. The method for evaluating the elastic contraction function of lung tissue according to claim 1, characterized in that: In step (5), the qualitative analysis is to perform maximum intensity projection on the IPF lung contraction template and the healthy lung contraction template along three different directions of x, y, and z, and finally obtain the lung contraction degree views in the coronal, sagittal and axial directions.

6. The method for evaluating the elastic contraction function of lung tissue according to claim 1, characterized in that: In step (5), the quantitative analysis is to calculate the subject's JAC NLC The average value is recorded as JAC M ; The IPF patient lung contraction template group and the healthy lung contraction template group were compared to analyze the difference in lung contraction degree between the IPF patient group and the healthy lung contraction template group.

7. The method for evaluating the elastic contraction function of lung tissue according to claim 1, characterized in that: In step (6), the significant lung contraction area is determined according to JAC NLC and a preset cutoff value, when it is greater than the cutoff value, it indicates significant shrinkage; the method for calculating the degree of overlap is to calculate the significant shrinkage area of the healthy template in HAJ according to the preset cutoff value, and express it in a binary way, 1 indicates significant shrinkage, and 0 indicates no significant shrinkage; Then, for each IPF patient, the preset cutoff value was used to obtain their respective significant shrinkage areas, which were also expressed in a binary way, with 1 indicating significant shrinkage and 0 indicating insignificant shrinkage; the Dice similarity coefficient was used to compare the degree of overlap between the significant shrinkage areas of each IPF patient and the significant shrinkage areas of the healthy template; for each IPF patient, the overlapping area between their significant shrinkage areas and the significant shrinkage areas of the healthy template was recorded as a true positive, the significant shrinkage areas belonging to the healthy template but not to the IPF patient were recorded as a false negative, and the significant shrinkage areas belonging to the IPF patient but not to the healthy template were recorded as a false positive.

8. The method for evaluating the elastic contraction function of lung tissue according to claim 7, characterized in that: The preset cutoff value is dynamically adjusted according to the mild, moderate and severe subgroups of IPF patients, and the dynamic adjustment range is JAC NLC The 80th-90th percentile of the dataset.

9. Use of the evaluation results obtained by the method for evaluating the elastic contractile function of lung tissue according to any one of claims 1 to 8 in evaluating various indicators of IPF patients.

10. The use according to claim 9, characterized in that The various indicators of IPF patients include lung function parameters, dyspnea, six-minute walk distance, health-related quality of life and degree of fibrosis.

Images