Metabolic nuclear magnetic resonance tumor fine diagnosis method

Abstract

The invention relates to a metabolic nuclear magnetic resonance tumor fine diagnosis method, which combines artificial intelligence (AI) growth, hydrogen spectrum nuclear magnetic resonance metabolic marker qualitative and quantitative identification scanning and nuclear magnetic resonance image scanning technologies to form a new composite technology system named as metabolic nuclear magnetic resonance (DX-MRI). The metabonomic nuclear magnetic resonance tumor fine diagnosis method has the advantages that the technology not only has the tumor property characteristic diagnosis core capability of PET-CT, but also has no internal and external radioactive radiation, and comprehensively surpasses the PET-CT in the aspects of automatic and fine diagnosis of tumors.

Description

Technical Field

[0001] This invention relates to a refined diagnostic method for metabolic tumors using nuclear magnetic resonance imaging. Background Technology

[0002] Currently, the core clinical imaging equipment in hospitals includes CT, PET-CT, and MRI, which are essential tools for diagnosing diseases.

[0003] Advantages of CT scans: 1. Fast scanning: CT scans are generally faster than MRI, making them suitable for emergency situations and patients who cannot remain still for extended periods. 2. High-density resolution: CT provides high-quality cross-sectional images, which is advantageous for displaying hard tissue structures such as bones and lungs. 3. Wide range of applications: CT can be used to examine multiple body parts, including the head, chest, abdomen, and bones. 4. Image reconstruction capabilities: CT images can be reconstructed in multiple planes using computer software.

[0004] Disadvantages of CT scans: 1. Radiation: CT scans use X-rays, posing a radiation risk, especially for patients who require frequent CT scans, and there is a risk of inducing tumors. 2. Poor soft tissue contrast: Compared to MRI, CT is less effective at displaying soft tissue structures, particularly in areas such as the nervous system and pelvic cavity.

[0005] Advantages of PET-CT: PET-CT is a medical imaging technology that combines positron emission tomography (PET) and computed tomography (CT). 1. Tumor Characterization: PET-CT has a significant advantage in the characteristic diagnosis of tumor characteristics. 2. Whole-Body Examination: PET-CT can perform a whole-body examination, helping to comprehensively understand the disease. 3. Accuracy: PET-CT combines anatomical and functional imaging, improving diagnostic accuracy, especially in tumor staging, treatment evaluation, and metastasis detection.

[0006] Disadvantages of PET-CT scans: 1. Radiation issues: PET-CT involves external radiation and requires the injection of radioactive tracers, posing an internal radiation risk and creating a relative clinical contraindication. 2. Difficulty in diagnosing certain tumors: PET-CT has difficulty diagnosing tumors in hollow organs (such as gastrointestinal tumors), especially early-stage tumors. 3. High cost: PET-CT equipment is expensive, and the examination fee is also high.

[0007] Advantages of MRI examination: 1. Radiation-free: MRI uses magnetic fields and radiofrequency pulses, which do not involve ionizing radiation and pose no radiation damage to the human body. 2. High soft tissue resolution: MRI can clearly display soft tissue structures. 3. Multi-parameter imaging: MRI can provide information on various physical parameters, such as proton density, T1 value, and T2 value, which is helpful for the diagnosis of lesions. 4. Multi-planar imaging: MRI can perform cross-sectional imaging in any direction, which is particularly important for examining lesions in the spinal canal and heart diseases.

[0008] Disadvantages of MRI: 1. Sensitivity to metals: MRI is sensitive to metallic objects; individuals with metallic objects in their bodies should not undergo MRI. 2. Higher cost: MRI examinations are generally more expensive than CT scans. 3. Relatively lower spatial resolution: Compared to CT, MRI has lower spatial resolution, especially when displaying small structures. 4. Lack of tumor-specific diagnostic capability: Compared to PET-CT, MRI lacks the core capability to differentiate the characteristic nature (benign or malignant) of tumors.

[0009] In summary, whole-body anatomical imaging with CT is a fundamental means of diagnostic imaging. PET-CT, building upon CT anatomical imaging, has been upgraded to possess functional imaging capabilities for diagnosing the nature of tumors. However, it also inherits the principles and limitations of CT imaging, particularly the unresolved risks of external radiation from CT and internal radiation from injecting radioactive isotopes. MRI offers similar whole-body anatomical imaging diagnostic capabilities to CT while eliminating the external radiation drawback, but it lacks the functional imaging capabilities for diagnosing the nature of tumors. Figure 1 This is a comparison table of the technical performance of CT, PET-CT, MRI, and DX-MRI. Summary of the Invention

[0010] Given the inherent limitations of both PET-CT and MRI, we propose a novel diagnostic technology: combining artificial intelligence, qualitative and quantitative identification of metabolic biomarkers using proton NMR spectroscopy, and NMR imaging techniques to form a new composite system, named Metabolic Magnetic Resonance Imaging (DX-MRI). This technology possesses the core diagnostic capabilities of PET-CT in identifying the characteristics of tumors, while eliminating internal and external radiation exposure. Furthermore, it comprehensively surpasses PET-CT in terms of automated and refined tumor diagnosis. Metabolic Magnetic Resonance Imaging (DX-MRI) has the potential to fully replace CT, PET-CT, and MRI in terms of anatomical imaging, functional imaging, and intelligent diagnosis.

[0011] To achieve the above objectives, the technical solution adopted in this invention is: a refined diagnostic method for tumors using metabolic nuclear magnetic resonance, including obtaining tumor marker characteristic data using proton nuclear magnetic resonance metabolomics technology.

[0012] Furthermore, tumor marker characteristic data are fed into artificial intelligence models for growth.

[0013] Furthermore, we will build intelligent metabolic magnetic resonance imaging equipment to achieve precise and automated diagnosis of all tumors.

[0014] This invention also provides a method for refined diagnosis of metabolic tumors using nuclear magnetic resonance imaging, which achieves this through a three-step macroscopic pathway:

[0015] 1) Obtain tumor marker characterization data using proton nuclear magnetic resonance metabolomics technology;

[0016] 2) Feature data is used to feed the growth of artificial intelligence models;

[0017] 3) Build intelligent metabolic magnetic resonance imaging equipment to achieve precise and automated diagnosis of all tumors.

[0018] Furthermore, step 1 specifically includes the following process, comprising four steps.

[0019] Four progressive steps to detect and obtain markers:

[0020] 1) High-resolution, high-sensitivity proton NMR spectroscopy detection of body fluids (saliva, urine, serum, tissue extracts, etc.);

[0021] 2) High-resolution magic-angle rotational nuclear magnetic resonance spectroscopy of ex vivo lesion tissue;

[0022] 3) Localized magnetic resonance spectroscopy detection of lesions in isolated organs;

[0023] 4) Localized spectroscopy detection of lesions in living tissue using nuclear magnetic resonance.

[0024] Furthermore, the macroscopic path described includes three stages.

[0025] Three consecutive stages enable patients to confirm a tumor diagnosis under outpatient DX-MRI:

[0026] 1) A. Find the biomarkers based on the answer (tumor tissue with confirmed postoperative pathology);

[0027] 2) Bridging phase: Bridging from A to B;

[0028] 3)B. Find the answer based on the landmarks.

[0029] Furthermore, step 1 specifically includes the following steps:

[0030] 101) Serum samples were obtained through clinical sample collection and separation;

[0031] 102) Proton NMR (1H NMR) experiment, spectrum preprocessing and principal component analysis;

[0032] 103) Results: Principal component analysis (PCA) was used to perform multivariate statistical modeling on the 1H NMR spectra of serum samples. Potential biomarkers for pancreatic cancer detection were obtained.

[0033] Furthermore, step 2 specifically includes the following steps:

[0034] 201) Obtain tissue samples from patients and rat models and store them at -80°C to prevent the samples from being affected by temperature changes;

[0035] 202) Tissue samples were acquired using magic-angle rotating proton nuclear magnetic resonance (HR-MAS1H NMR);

[0036] 203) 1H NMR spectral processing was performed to obtain a two-dimensional data matrix for data analysis;

[0037] 204) Statistical analysis was conducted to summarize differential metabolites;

[0038] 205) Metabolic correlation and pathway analysis;

[0039] 206) As a result, based on the above biological tests and multiple analyses, potential biomarkers for pancreatic ductal carcinoma (PDAC) were finally identified from pancreatic tissue.

[0040] Furthermore, the unique features and innovations of tumor marker acquisition technology.

[0041] 1) Employing proton nuclear magnetic resonance (NMR) technology: high-resolution and high-sensitivity proton NMR technology for body fluids, high-resolution "magic angle rotation" NMR technology for tissues, and in vivo magnetic resonance localization spectroscopy technology, combining the systematic nature of highly sensitive, high-throughput, and in-situ quantitative metabolomics analysis methods with the overall complexity of animal and human systems;

[0042] 2) Integrate physical detection, cluster analysis, analytical chemistry, and medical analysis techniques. Based on high-throughput physical detection and cluster index analysis, and utilizing analytical chemistry and medical analysis techniques, combined with multivariate data analysis and mathematical modeling, this approach aims to address critical challenges in early clinical diagnosis.

[0043] 3) Integrate biomedical understanding at different levels, such as animal body fluids, tissues, and the whole body, as well as human body fluids and tissues, with their metabolomic characteristics. Extend accurate animal information and data modeling to mathematical modeling for the early diagnosis of human diseases.

[0044] 4) Integrate basic scientific research on the physiological effects of animal and human diseases with the data modeling needs for early diagnosis of diseases in clinical medicine.

[0045] Compared to existing technologies, this invention offers the following advantages: Both PET-CT and MRI have significant limitations. We propose a novel diagnostic method: combining artificial intelligence, qualitative and quantitative identification scanning of metabolic biomarkers using hydrogen spectroscopy and nuclear magnetic resonance imaging (NMR), to form a new composite technology system. This technology possesses the core diagnostic capabilities of PET-CT in identifying the characteristics of tumors, without internal or external radioactive radiation, and achieves a comprehensive surpassing of PET-CT in automated and refined tumor diagnosis. Metabolic Magnetic Resonance Imaging (DX-MRI): DX-MRI has the potential to completely replace CT, PET-CT, and MRI in anatomical imaging, functional imaging, and intelligent diagnosis, possessing enormous market potential. The successful construction of intelligent DX-MRI equipment will enable automated and refined diagnosis of whole-body tumors using artificial intelligence, significantly improving the application depth of imaging technology in the early and staged diagnosis of whole-body tumors, which is of great significance to public health. Attached Figure Description

[0046] Figure 1 This is a comparison table of the technical performance of DX-MRI compared to CT, PET-CT, and MRI.

[0047] Figure 2 This is the research strategy of metabolomics in implementing the technology of this invention.

[0048] Figure 3 This invention describes the process for analyzing blood samples, urine samples, and tissue extracts using metabolomics methods.

[0049] Figure 4 This invention describes the process of establishing animal models and analyzing human and animal model tissue samples using metabolomics methods.

[0050] Figure 5 This is an article publishing the implementation technology of this invention. Figure 6 This invention relates to the proposed DX-MRI intelligent device. Detailed Implementation

[0051] To make the above features and advantages of the present invention more apparent and understandable, specific embodiments are described below in conjunction with the accompanying drawings for detailed explanation.

[0052] like Figure 2 As shown, biomarkers were obtained by proton NMR spectroscopy:

[0053] 1) such as Figure 3 As shown, high-resolution, high-sensitivity proton NMR spectroscopy detection of serum was performed. The specific steps are as follows:

[0054] 101) Clinical sample collection: Blood samples were collected by medical staff. A total of 40 cases were included, comprising a pancreatic cancer study group and a healthy control group. General characteristics were similar. Blood was collected in the morning on an empty stomach, allowed to stand at 4°C for 1 hour, and then centrifuged at 1000 rpm for 5 minutes to obtain serum samples. These samples were then stored at -80°C for later use. All patients' diagnoses were verified by postoperative pathological examination.

[0055] 102) For NMR experiments, spectrum preprocessing, and principal component analysis, serum samples were dissolved in phosphate D20 buffer (0.2M, pH 7.4, SIGMA, USA) at a volume ratio of 2:1 to prepare a final volume of 600 μL. TSP (2,2,3,3-deuterotrimethylsilylpropionic acid) was added as an internal standard (P < 0.00). Samples were loaded into 5 mm NMR tubes and the pH was measured at room temperature using a pH meter (METTLER TOLEDO DELTA 320pH), with a value of 7.4 ± 0.1. All NMR data for this experiment were obtained from Broker... The data was acquired using a DRX-500 nuclear magnetic resonance spectrometer (iH observation frequency 500.13MHz). The specific method for acquiring nuclear magnetic spectrum is described in the serum nuclear magnetic spectrum scanning of Xu Jingjing et al. [1] at Xiamen University. Phase correction and baseline correction were performed on all acquired 2D spectra. Spectrum preprocessing and principal component analysis: The spectra were manually phase-adjusted, baseline-corrected and peak-aligned using MestRe-C2.3 software and self-developed software. The serum samples were integrated in segments at intervals of Δδ = 0.02 in the δ0.2 to 4.6 region. The integrated data were normalized to form a data matrix, and the data matrix was statistically analyzed using the principal component analysis method.

[0056] 103) Results: Multivariate statistical modeling was performed on the 1H NMR spectra of serum samples using PCA. The PCA score plots showed a clear distinction in the metabolome between pancreatic cancer patients and healthy controls. Furthermore, loading plot analysis revealed that this distinction was primarily attributed to metabolites such as isoleucine, triglycerides, leucine, creatinine, lactate, 3-hydroxybutyrate, 3-hydroxyisovaleric acid, and trimethylamine-N-oxide (TMAO), changes of which may serve as biomarkers for pancreatic cancer detection.

[0057] 2) such as Figure 4 The image shows high-resolution magic-angle rotation nuclear magnetic resonance spectroscopy of isolated pancreatic tissue. The specific steps are as follows:

[0058] 201) Tissue sample collection from patients and rat models: In our study, pancreatic tissue from rats was collected within 15 minutes of sacrifice, and pancreatic tissue from patients was collected within 30–60 minutes from the onset of intraoperative ischemia, depending on the surgical time. A small portion of each specimen was stained with hematoxylin and eosin and histologically verified by an experienced pathologist. The remaining portions were then rapidly frozen in liquid nitrogen and stored at -80°C to prevent sample deterioration due to temperature changes.

[0059] 202) Tissue samples were acquired using HR-MAS 1H NMR. For HR-MAS 1H NMR spectroscopy, tissue samples were removed from a -80°C freezer and thawed at room temperature. Each sample, weighing approximately 20 mg, was placed in a 4 mm diameter zirconia rotor. The rotor was simultaneously filled with physiological saline prepared from heavy water to maintain osmotic pressure and provide fill lock. The rotor was then sealed with a cap to expel excess heavy water from the gaps. HR-MAS NMR experiments were performed on a Bruker 600 MHz NMR spectrometer equipped with a triple-field resonance (1H / 13C / 31P) high-resolution MAS probe, operating at a 1H frequency of 600.13 MHz. The rotation speed of the tissue samples was set to 3.0 kHz, and the experimental temperature was controlled at 288 K (temperature limitation due to hardware constraints) at a magic angle (54.7°). Sampling was performed using a CPMG (Carr-Purcell-Meiboom-Gill) spin-echo pulse sequence (RD-90°-(τ-180°-τ)n-ACQ) to suppress the water signal. Spectral acquisition had a spin-spin relaxation delay of 2nτ = 350 ms, during which the water signal was irradiated. Typically, 128 scans were collected onto 16K data points, with a spectral width of 12 kHz, a relaxation time of 2.0 seconds, and an acquisition time of 1.33 seconds.

[0060] 203) NMR spectral processing followed by transfer of NMR spectra to MestReNova software (version: 9.0.1, Mestrelab Research SL, Spain). All free induction decays were Fourier transformed with a FT size of 32K and a linewidth broadening factor of 0.3Hz. Spectral phase and baseline correction were manually adjusted, and then referenced to endogenous lactate at 1.33ppm. Each 1H NMR spectrum (0.50–9.00ppm) was integrated at 0.005ppm intervals after removing peaks for residual water (4.7–5.2ppm), and the entire spectrum was then normalized to obtain a two-dimensional data matrix for data analysis.

[0061] 204) Statistical analysis: In order to analyze the metabolic profile of tissues through pattern recognition technology, the processed spectral data were analyzed using SIMCA software (V14.1, Umetrics AB, Umeå, Sweden). Principal component analysis (PCA), an unsupervised pattern analysis method, was performed first to provide an overview of the sample distribution and identify possible outliers. Orthogonal partial least squares discriminant analysis (OPLS-DA), a supervised pattern recognition method, was then used to distinguish paired samples to the maximum extent and further screen for characteristic metabolites corresponding to PDAC. OPLS separates the systematic variation in X into two parts: one part is the component associated with Y (predicted) (t[1]P in our case) and the other part is the component unrelated to Y (orthogonal) (t[2]O in our case). This improves the interpretability of the model. In our case, only one predictive component and one orthogonal component were set. The model parameters R2 and Q2 describe the model’s fitting ability and predictive ability, respectively. The model was cross-validated using permutation tests (number of permutations = 200) and CV-ANOVA (cross-validation analysis of variance). Differential metabolites between the PDAC group and the corresponding control were identified using Pearson correlation coefficients (r) and projected importance (VIP) values ​​from multivariate statistical analysis (OPLS-DA) and p-values ​​for each metabolite from univariate statistical analysis (one-way ANOVA using Student's t-test). Enhanced four-dimensional volcano plots were displayed using MATLAB (R2014b, Mathworks, USA) to reflect correlation coefficients, VIP, p-values, and fold changes from the relative concentration of each metabolite, summarizing the differential metabolites.

[0062] 205) Metabolic Correlation and Pathway Analysis: To understand the metabolic correlations among differentially metabolites and the perturbed metabolic pathways involved in PDAC, cluster analysis of characteristic metabolites from PDAC patients and rats was performed using Heml software (version 1.0.3.7, Zhang Zhang, Beijing Institute of Genomics, China). Pathway topology analysis of differentially metabolites between paired groups was performed using the Kyoto Encyclopedia of Genetics and Genomes (KEGG) (http: / / www.kegg.jp / ) and the MetaboAnalyst 4.0 online service (http: / / www.metaboanalyst.ca / MetaboAnalyst). Finally, based on the characteristic metabolites of PDAC patients and SD rats, a metabolic network containing a series of correlations between metabolic responses was constructed using PathVisio software (version 3.3.0).

[0063] 206) Results: Based on the above biological tests and multiple analyses, potential biomarkers for PDAC were ultimately identified from pancreatic tissue. Accordingly, relevant metabolic pathways associated with the occurrence of PDAC were identified using the MetaboAnalyst online service, where each point represents an associated metabolic pathway, and the size of the point represents the influence value of that metabolic pathway. In our case, pathways with a p-value less than 0.01 were screened as potential target pathways. To better illustrate the metabolic differences between PDAC patients and SD rats, their respective metabolic networks were constructed based on characteristic metabolites, including correlations between a range of metabolic responses to PDAC. Essentially, the associated metabolic disorders are concentrated in the metabolism of glycerol phospholipids, glycolysis or gluconeogenesis, galactose metabolism, glycine, serine and threonine metabolism, taurine and hypotaurine metabolism, methane metabolism, and the synthesis and degradation of ketone bodies in PDAC patients, while the metabolic disorders in PDAC rats involve the biosynthesis of aminoacyl-tRNA, valine, leucine and isoleucine biosynthesis, the metabolism of amino acids including histidine, alanine, aspartic acid, glutamate and glutamine, galactose metabolism, glycerol phospholipid metabolism, biotin metabolism, and taurine and hypotaurine metabolism.

[0064] Artificial intelligence model growth: based on the database obtained in the "first step".

[0065] 1) Continuous testing of clinical pancreatic cancer specimens to obtain sufficient data, continuously feeding and training the AI, and gradually improving accuracy (through mutual verification and relearning between imaging diagnosis and postoperative pathological diagnosis). This leads to the formation of a stable digital AI model for pancreatic cancer diagnosis.

[0066] 2) Taking pancreatic cancer research as a starting point and template, we will obtain metabolic biomarkers for the whole body and the entire tumor spectrum, and form a stable whole body tumor diagnosis AI digital model by feeding AI with big data.

[0067] Guided by the research and development strategy and expected goals, multiple teams were assembled to collaboratively develop and build a DX-MRI intelligent device. This aims to combine the technologies of 1H-NMR and MRI in metabolomics. According to relevant data, both technologies are mature and share the same underlying principles, making their combination feasible. MRI is widely used in clinical practice, and domestic production has flourished, but its technology still cannot surpass that of foreign counterparts. Currently, 1H-NMR in metabolomics is only available with imported equipment and has no clinical application.

[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation of the present invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the technical solutions claimed in the present invention.

Claims

1. A novel diagnostic technology, characterized in that, This includes a new composite technology system called Metabolic Magnetic Resonance Imaging (DX-MRI), which combines artificial intelligence (AI) with hydrogen spectroscopy and nuclear magnetic resonance (NMR) scanning technology to form qualitative and quantitative identification of metabolic biomarkers. Its features include hardware and software consisting of three core functions: anatomical imaging, functional imaging, and intelligent diagnosis.

2. The method for refined diagnosis of metabolic tumors using nuclear magnetic resonance imaging according to claim 1, characterized in that, The device described in the method requires no contrast agents and is free from internal and external radiation during the diagnostic process.

3. The method for refined diagnosis of metabolic tumors using nuclear magnetic resonance imaging according to claim 1, characterized in that, The device implementing the method has the ability to clearly define the nature and stage of tumors, qualitatively and quantitatively analyze tumor markers, and provide intelligent diagnostic capabilities, which can effectively improve the accuracy of various tumor diagnoses.

4. A refined diagnostic method for metabolic tumors using nuclear magnetic resonance imaging, characterized in that, Macro-level path to implementation: 1) Obtain tumor marker characterization data using proton nuclear magnetic resonance metabolomics technology; 2) Feature data is used to feed the growth of artificial intelligence models; 3) Build intelligent metabolic magnetic resonance imaging equipment to achieve precise and automated diagnosis of all tumors.

5. The method for refined diagnosis of metabolic tumors using nuclear magnetic resonance imaging according to claim 4, characterized in that, Path 1 specifically includes the following four steps: 101) High-resolution, high-sensitivity proton nuclear magnetic resonance spectroscopy detection of body fluids (saliva, urine, serum, tissue extracts, etc.); 102) High-resolution magic-angle rotational nuclear magnetic resonance spectroscopy detection of ex vivo lesion tissue; 103) Localized magnetic resonance spectroscopy detection of lesions in isolated organs; 104) Localized spectroscopy detection of in vivo lesions using nuclear magnetic resonance.

6. The method for refined diagnosis of metabolic tumors using nuclear magnetic resonance imaging according to claim 4, characterized in that, Each path specifically includes the following three stages: 401) A. Find the landmark based on the answer; 402) Bridging phase, bridging from A to B; 403)B. Find the answer based on the landmark.

7. The method for refined diagnosis of metabolic tumors using nuclear magnetic resonance imaging according to claims 5 and 6, characterized in that, The method achieves live tumor identification through four steps and three stages: finding the answer (postoperative pathologically confirmed tumor tissue), then bridging, and finally relying on markers to identify live tumors (achieving DX-MRI diagnosis of outpatient tumor patients).

Images