CN116740064B - Nuclear magnetic resonance tumor region extraction method - Google Patents
Nuclear magnetic resonance tumor region extraction method Download PDFInfo
- Publication number
- CN116740064B CN116740064B CN202311014005.5A CN202311014005A CN116740064B CN 116740064 B CN116740064 B CN 116740064B CN 202311014005 A CN202311014005 A CN 202311014005A CN 116740064 B CN116740064 B CN 116740064B
- Authority
- CN
- China
- Prior art keywords
- voxel
- judged
- central
- metabolite
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 206010028980 Neoplasm Diseases 0.000 title claims abstract description 51
- 238000005481 NMR spectroscopy Methods 0.000 title claims abstract description 24
- 238000000605 extraction Methods 0.000 title claims abstract description 15
- 239000002207 metabolite Substances 0.000 claims abstract description 155
- 238000001228 spectrum Methods 0.000 claims abstract description 86
- 238000000034 method Methods 0.000 claims abstract description 51
- 239000000126 substance Substances 0.000 claims description 39
- 238000006073 displacement reaction Methods 0.000 claims description 21
- DRBBFCLWYRJSJZ-UHFFFAOYSA-N N-phosphocreatine Chemical group OC(=O)CN(C)C(=N)NP(O)(O)=O DRBBFCLWYRJSJZ-UHFFFAOYSA-N 0.000 claims description 10
- CVSVTCORWBXHQV-UHFFFAOYSA-N creatine Chemical compound NC(=[NH2+])N(C)CC([O-])=O CVSVTCORWBXHQV-UHFFFAOYSA-N 0.000 claims description 10
- 229960003624 creatine Drugs 0.000 claims description 5
- 239000006046 creatine Substances 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 4
- -1 tetramethyl amine Chemical class 0.000 claims description 2
- 238000013075 data extraction Methods 0.000 abstract description 3
- 238000012545 processing Methods 0.000 abstract description 2
- 230000003595 spectral effect Effects 0.000 description 30
- 238000004364 calculation method Methods 0.000 description 12
- 238000004611 spectroscopical analysis Methods 0.000 description 7
- 238000000655 nuclear magnetic resonance spectrum Methods 0.000 description 6
- 230000033228 biological regulation Effects 0.000 description 5
- 238000005259 measurement Methods 0.000 description 5
- 238000010606 normalization Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 238000002595 magnetic resonance imaging Methods 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000000630 rising effect Effects 0.000 description 2
- 208000003174 Brain Neoplasms Diseases 0.000 description 1
- QNAYBMKLOCPYGJ-REOHCLBHSA-N L-alanine Chemical compound C[C@H](N)C(O)=O QNAYBMKLOCPYGJ-REOHCLBHSA-N 0.000 description 1
- JVTAAEKCZFNVCJ-UHFFFAOYSA-M Lactate Chemical compound CC(O)C([O-])=O JVTAAEKCZFNVCJ-UHFFFAOYSA-M 0.000 description 1
- 125000003047 N-acetyl group Chemical group 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 235000004279 alanine Nutrition 0.000 description 1
- 230000006538 anaerobic glycolysis Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000747 cardiac effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000006911 enzymatic reaction Methods 0.000 description 1
- 230000023266 generation of precursor metabolites and energy Effects 0.000 description 1
- 230000034659 glycolysis Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30096—Tumor; Lesion
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention discloses a nuclear magnetic resonance tumor region extraction method, and belongs to the technical field of image data processing; the method comprises the following steps: acquiring a region of interest in a magnetic resonance image; acquiring the relative concentration gradient coincidence degree of each voxel to be judged; acquiring connectivity outlier factors of each voxel to be judged according to the average link distance of each voxel to be judged; acquiring the outlier degree of each voxel to be judged according to the connectivity outlier factor of each voxel to be judged; judging whether the region of interest is added or not according to the outlier degree of each voxel to be judged to obtain the nuclear magnetic resonance tumor region. According to the invention, through judging the metabolite similarity of adjacent voxels at the boundary of the region of interest and expanding the region of interest according to the spectrum similarity, the spectrum data extraction of the tumor region of interest is more accurate.
Description
Technical Field
The invention relates to the technical field of image data processing, in particular to a nuclear magnetic resonance tumor region extraction method.
Background
In the process of extracting tumor region data in nuclear Magnetic Resonance Spectroscopy (MRS), the MRS signal of interest needs to be selected from a large number of MRS signals for analysis, in the process of extracting, a region of interest (VOI) in the process of multi-voxel spectrum acquisition needs to be determined through MRI images first, and then the multi-voxel spectrum acquisition of the region of interest is performed by using a nuclear magnetic resonance spectrometer through a chemical shift method (CSI). In the Magnetic Resonance Spectroscopy (MRS) extraction of voxels of a region of interest, it is necessary to extract spectroscopic data of tumor voxels in the region of interest, thereby performing tumor extraction through spectroscopic data of a tumor region. In the actual tumor region data extraction process, because the voxels are three-dimensional space structures, similarity judgment is needed to be carried out on voxel spectrums around the region of interest in the extraction process, so that the tumor region is extracted more accurately. In the process of judging the similarity of the voxel spectrums of the adjacent areas of the voxels of the region of interest and the tumor area, firstly, nuclear magnetic resonance spectrums of the voxels of the VOI area are required to be acquired, then, adjacent voxel spectrums outside the edges of the VOI area are detected outwards according to the central voxel in the VOI area, the similarity of spectral lines of the spectrums is judged, and similar voxels are added into the VOI area according to the acquired similarity degree, so that complete MRS data of the tumor area are extracted.
To achieve the above objective, those skilled in the art make a similarity determination of the magnetic resonance spectral lines of two voxels by Dynamic Time Warping (DTW). However, in the similarity judging process of the voxel spectrum, because the traditional sequence similarity judgment is carried out through subintervals with fixed lengths, the similarity judgment error can be caused by the metabolite peak deviation of spectral lines in the judging process.
Disclosure of Invention
In order to solve the problem that in the prior art, in the similarity judging process of voxel spectrums, because the traditional sequence similarity judgment is carried out through subintervals with fixed length, the similarity judgment is wrong because of metabolite peak deviation of spectral lines in the judging process, the invention provides a nuclear magnetic resonance tumor region extraction method.
The invention aims to provide a nuclear magnetic resonance tumor region extraction method, which comprises the following steps of:
acquiring a region of interest in a magnetic resonance image; acquiring a central voxel in a region of interest; taking voxels in the region outside the region of interest as voxels to be judged; acquiring the relative concentration and spectrum data of each voxel;
acquiring initial DTW distances between a central voxel and each voxel to be judged by using all data points in spectrum data of the central voxel and each voxel to be judged;
optimizing the initial DTW distance between the central voxel and each voxel to be judged according to the positions of the metabolite data points in the spectrum data of the central voxel and each voxel to be judged to obtain the optimized DTW distance between the central voxel and each voxel to be judged;
acquiring the relative concentration of voxels from the center voxel to each voxel to be judged to form a relative concentration sequence, and acquiring a differential sequence of the relative concentration sequence;
according to the number of voxels contained between the central voxel and each voxel to be judged, the number of voxels with continuous gradient change of the relative concentration of the voxels contained between the central voxel and each voxel to be judged, and the differential sequence of the relative concentration sequence, acquiring the relative concentration gradient coincidence degree of each voxel to be judged;
acquiring the average link distance of each voxel to be judged based on a connectivity outlier factor algorithm according to a voxel sequence formed by voxels contained between a central voxel and each voxel to be judged, the relative concentration gradient coincidence degree of each voxel to be judged and the optimized DTW distance;
acquiring connectivity outlier factors of each voxel to be judged according to the average link distance of each voxel to be judged; acquiring the outlier degree of each voxel to be judged according to the connectivity outlier factor of each voxel to be judged;
judging whether the region of interest is added or not according to the outlier degree of each voxel to be judged to obtain the nuclear magnetic resonance tumor region.
In an embodiment, the optimized DTW distance is obtained according to the following steps:
adjusting the position of a metabolite data point in the spectrum data of each voxel to be judged to be at the same position as a corresponding metabolite data point in the spectrum data of the central voxel, and acquiring a chemical displacement deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel and a distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel; obtaining chemical displacement deviation factors of different metabolite data points of each voxel to be judged relative to the central voxel and distance difference factors of different metabolite data points of each voxel to be judged relative to the central voxel;
optimizing the data point distance of the same metabolite according to the chemical displacement deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel and the distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel; acquiring the optimized DTW distance of the same metabolite data points;
optimizing the distances of different metabolite data points according to the chemical displacement deviation factors of different metabolite data points of each voxel to be judged relative to the central voxel and the distance difference factors of different metabolite data points of each voxel to be judged relative to the central voxel, and obtaining the optimized DTW distances of different metabolite data points;
the optimized DTW distance for the same metabolite data point and the optimized DTW distance for a different metabolite data point are taken as the optimized DTW distances.
In one embodiment, the chemical shift deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel is obtained according to the following steps:
when the positions of the metabolite data points in the spectrum data of each voxel to be judged are adjusted, acquiring the chemical displacement difference of the positions of the metabolite data points in the spectrum data of each voxel to be judged relative to the corresponding metabolite data points in the spectrum data of the central voxel; normalized chemical shift difference was used as a chemical shift deviation factor.
In one embodiment, the distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel is obtained according to the following steps:
when the position of a metabolite data point in the spectrum data of each voxel to be judged is adjusted, acquiring the dynamic time regular distance difference of the metabolite data point in the spectrum data of each voxel to be judged relative to the corresponding metabolite data point in the spectrum data of the central voxel before adjustment and after adjustment; taking the normalized dynamic time-warping distance difference as a distance difference factor.
In one embodiment, the relative concentration of each voxel is obtained according to the following steps:
index the metabolite concentration in each voxelAs the relative concentration of each voxel; wherein Cho represents the concentration of the metabolite being tetramethylamine; cr represents the concentration of creatine phosphate and creatine.
In an embodiment, the relative concentration of the voxels included between the center voxel and each voxel to be determined forms a relative concentration sequence, comprising:
acquiring a voxel sequence formed by a central voxel and voxels contained between each voxel to be judged;
taking a sequence formed by the relative concentration corresponding to each voxel in the voxel sequence as a relative concentration sequence;
the voxel sequence is formed by all voxels on the connecting line from each voxel to be judged to the central voxel.
In one embodiment, the nmr tumor region is extracted according to the following steps:
acquiring the interest degree of each voxel to be judged according to the outlier degree of each voxel to be judged;
setting a threshold value;
when the interest degree of the voxel to be judged is greater than a threshold value, adding the voxel to be judged into the region of interest;
and judging each voxel to be judged in sequence, and adding all voxels to be judged which are larger than a threshold value into the region of interest to obtain the nuclear magnetic resonance tumor region.
In one embodiment, the central voxel of the region of interest is the voxel with the highest average concentration of the metabolite in all spectral data of the region of interest.
Further, the threshold is set to 0.7.
Further, the region of interest is obtained by human labeling.
The beneficial effects of the invention are as follows: according to the nuclear magnetic resonance tumor region extraction method provided by the invention, the initial DTW distance between the central voxel and each voxel to be judged is optimized through the position of the metabolite data point in the spectrum data of the central voxel and each voxel to be judged, so that the optimized DTW distance between the central voxel and each voxel to be judged is obtained, and the data point in the traditional DTW dynamic regulation process is prevented from being in the correspondence of different metabolite subintervals. The DTW distance can ensure the corresponding relation between data points through the optimization of the distance, so that the DTW distance between the spectral line to be detected and the nearest spectral line in the region of interest is obtained; and secondly, comparing the similarity obtained by distance normalization with the traditional similarity obtained by obtaining the gradient coincidence degree of the relative concentration of each voxel to be judged, judging whether the voxel to be judged is placed in the region of interest or not, and correcting the distance in the similarity judging process by the metabolite concentration information in the magnetic resonance data of the tumor region, thereby obtaining the accurate similarity between the magnetic resonance spectrum data of the voxel to be judged and the voxels of the region of interest. Finally, combining the relative concentration gradient coincidence degree of each voxel to be judged and the optimized DTW distance, and obtaining the outlier degree of each voxel to be judged through a connectivity outlier factor algorithm, so that similarity judgment between voxel spectrum curves is more accurate, and expanding judgment is carried out on a region of interest through spectrum similarity, so that a more accurate tumor region is extracted in nuclear magnetic resonance detection.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a flowchart illustrating general steps of an embodiment of a method for extracting a nuclear magnetic resonance tumor region according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention aims to optimize the determination of neutron intervals in the similarity judgment process through the space gradual change condition of metabolite peaks in the spectrums of voxels of a tumor region of interest, so that the similarity judgment between voxel spectrum curves is more accurate, similar voxels are gathered together to be used as the region of interest through spectrum similarity until no voxel to be judged can be added into the region of interest, the judgment is finished, and the finally obtained region of interest is used as an extracted tumor region, so that a more accurate tumor region is extracted in nuclear magnetic resonance detection.
The invention aims at a scene that when spectrum data extraction is carried out on a tumor region of interest in Magnetic Resonance Imaging (MRI), because the condition of inaccurate image exists in artificial labeling of the region of interest, similarity judgment is further carried out on voxels to be judged outside the region of interest, and accurate extraction of the tumor region is carried out through characterization of the voxels in a spectrum curve. In the similarity judging process of the voxel spectrum, because the traditional sequence similarity judgment is carried out through subintervals with fixed lengths, the situation of similarity judgment error caused by metabolite peak deviation of spectral lines in the judging process can be caused.
The invention provides a nuclear magnetic resonance tumor region extraction method, which is shown in figure 1 and comprises the following steps:
s1, acquiring a region of interest in a magnetic resonance image; acquiring a central voxel in a region of interest; taking voxels in the region outside the region of interest as voxels to be judged; acquiring the relative concentration and spectrum data of each voxel;
in this embodiment, the region of interest is determined from the MRI image, typically by manual labeling, and the voxels outside the region of interest are determined based thereon. I.e. the region of interest mentioned in this embodiment is regarded as a tumor region.
In this embodiment, the device used in acquiring the spectroscopic data is a Philips 3.0T magnetic resonance scanner, using a standard 8-channel head coil. Scanning by adopting MRI detection and T2WI tumor transverse position thin layers conventionally, primarily positioning a tumor region by combining images, and determining a field of view (FOV) and a region of interest (VOI) of multi-voxel magnetic resonance spectroscopy scanning; the multi-voxel magnetic resonance spectrum data adopts a 3D-PRESS sequence, and the parameters are TR:2000ms, TE 32ms and 90 degrees of reverse rotation. For each voxel of the magnetic resonance spectrum acquired by the region of interest, quantification and baseline calibration were performed by lcmode.
In this embodiment, the relative concentration of each voxel is obtained according to the following steps:
index the metabolite concentration in each voxelAs the relative concentration of each voxel; wherein Cho represents the concentration of the metabolite being tetramethylamine; cr represents the concentration of creatine phosphate and creatine.
In the spectrum data of the region of interest, the peak value of the different metabolites gradually decreases from the voxel with the highest concentration as the center, so that the voxel with the highest average concentration of the metabolites in all the spectrum data is used as the center voxel; taking voxels in the region outside the region of interest as voxels to be judged; in the case of performing the similarity determination of nuclear magnetic resonance spectra, the objective is to extract the complete spectrum data of the tumor region, so that the similarity is mainly determined by the similarity determination of several metabolite peaks that are most important for the determination of the tumor, and the determination needs to be performed by the similarity of the same metabolite peaks in the nuclear magnetic resonance spectra of different voxels. In this embodiment, the spectrum data of each voxel to be determined is generally represented by a spectrum curve or a spectral line.
S2, acquiring initial DTW distances between the central voxel and each voxel to be judged by using all data points in the spectrum data of the central voxel and each voxel to be judged;
optimizing the initial DTW distance between the central voxel and each voxel to be judged according to the positions of the metabolite data points in the spectrum data of the central voxel and each voxel to be judged to obtain the optimized DTW distance between the central voxel and each voxel to be judged;
in this embodiment, the spectrum data of the central voxel corresponds to a time sequence, and the spectrum data of each voxel to be determined corresponds to a time sequence, so that the initial distance between the central voxel and all data points in the spectrum data of each voxel to be determined is obtained based on a dynamic time warping algorithm (DTW).
It should be noted that, in the magnetic resonance spectrum lines, metabolite peaks at different chemical shift (PPM) positions represent corresponding metabolites, and in the calculation process of the DTW distance, since there are conditions corresponding to different metabolite peaks, the same metabolite peak needs to be corrected in the dynamic normalization process, so that the DTW distance can correctly measure the similarity of two lines in the similarity measurement process. Since the peak of the metabolite is shifted due to physical displacement or the like, the metabolite cannot be extracted in a fixed interval. And extracting the whole range of the wave crest through an adaptive window for the wave crest in the spectrum curve.
In this embodiment, the specific process of extracting the peak in the spectrum curve of each voxel to be judged in the whole range through the adaptive window is as follows: firstly, judging from the leftmost side of a spectral line (the position with highest chemical shift (PPM)) to the right by using the right line of a window as a starting point; second, the baseline corrected spectral line will appear as a baseline stable spectral line, and for each PPM value corresponding spectral signal intensity, it is detected whether it is rising compared to the previous data point. Then, when a fluctuation rise is detected, the window is cut off until the regression baseline is reached, and the data corresponding to the spectral line selected by the window is recorded as the firstThe +.>A line fluctuation window. That is to say, the metabolite peaks in the spectral line are framed as +.>The +.>A fluctuation window; meanwhile, taking data corresponding to a window of each frame selected wave crest as a subinterval of spectral lines; finally, all fluctuations in the spectral line are detected (only fluctuations rising from the base line are considered).
For this purpose, a plurality of fluctuation windows are obtained in the nuclear magnetic resonance spectrum of each voxel to be judged, and for each fluctuation, the metabolite represented by the fluctuation needs to be determined by the chemical shift (PPM) position of the fluctuation. In brain tumors, the following are mainly identified: cho:tetramethyl amine, at 3.2ppm. Cr: creatine phosphate and creatine, at 3.0ppm. NAA: n-acetyl groups were located near 2.0 ppm. Ala: alanine, a metabolite of intermediary metabolism, was located at 1.4ppm. LA: lactate, an index that indirectly reflects anaerobic glycolysis or glycolysis due to abnormal enzymatic reactions, is at 1.3ppm. And determining the metabolite corresponding to the fluctuation through the nearest neighbor feature PPM of the position of the fluctuation peak value in each window. Obtain the firstNuclear magnetic resonance spectrum of each voxel +.>Metabolites corresponding to the respective fluctuation windows. And determining a one-to-one correspondence relationship through the correspondence relationship of the metabolites for the two nuclear magnetic resonance spectrum lines of the similarity to be compared.
In this embodiment, the initial DTW distance is optimized according to the relative positional relationship between the position of the metabolite data point in the spectrum data of each voxel to be determined and the metabolite data point in the spectrum data of the central voxel, so as to obtain the optimized DTW distance.
Specifically, the optimized DTW distance is obtained according to the following steps:
adjusting the position of a metabolite data point in the spectrum data of each voxel to be judged to be at the same position as a corresponding metabolite data point in the spectrum data of the central voxel, and acquiring a chemical displacement deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel and a distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel; obtaining chemical displacement deviation factors of different metabolite data points of each voxel to be judged relative to the central voxel and distance difference factors of different metabolite data points of each voxel to be judged relative to the central voxel;
optimizing the data point distance of the same metabolite according to the chemical displacement deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel and the distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel; acquiring the optimized DTW distance of the same metabolite data points;
optimizing the distances of different metabolite data points according to the chemical displacement deviation factors of different metabolite data points of each voxel to be judged relative to the central voxel and the distance difference factors of different metabolite data points of each voxel to be judged relative to the central voxel, and obtaining the optimized DTW distances of different metabolite data points;
the optimized DTW distance for the same metabolite data point and the optimized DTW distance for a different metabolite data point are taken as the optimized DTW distances.
In this embodiment, after the obtained correspondence between the metabolite peaks in the central voxel line and each voxel line to be determined, since there is a difference in chemical shift between the metabolite peaks, it is necessary to register the metabolite peaks when performing similarity comparison between them. For the corresponding fluctuation window in the spectrum data of the central voxel and the voxel to be judged, namely corresponding identical metabolite wave peaks in the spectrum data. It should be noted that, the two corresponding fluctuation windows correspond to two subsequences in the spectrum data, and the two corresponding fluctuation windows correspond to two subintervals at the same time; and obtaining the nearest corresponding relation of the data points in the two subsequences through DTW dynamic regulation. And adjusting the peak data point of the metabolite peak of the voxel to be judged to be at the same position as the peak of the metabolite peak of the central voxel, and filling the two subintervals into equal length by filling the baseline numerical values at the two sides of the peak data point after adjustment. And acquiring the DTW distances of the two fluctuations in the subinterval after registration. The PPM distance for adjusting the peak position is the difference of curve chemical displacementSub-interval sequence DTW distance difference before and after registration +.>Obtaining chemical shift deviation factor of magnetic resonance spectrum of voxel to be detected according to PPM difference between each corresponding subinterval and DTW distance difference before and after registration>Difference factor from subinterval>。
Specifically, the chemical shift deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel is obtained according to the following steps: when the positions of the metabolite data points in the spectrum data of each voxel to be judged are adjusted, acquiring the chemical displacement difference of the positions of the metabolite data points in the spectrum data of each voxel to be judged relative to the corresponding metabolite data points in the spectrum data of the central voxel; normalized chemical shift difference was used as a chemical shift deviation factor.
The distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel is obtained according to the following steps: when the position of a metabolite data point in the spectrum data of each voxel to be judged is adjusted, acquiring the dynamic time regular distance difference of the metabolite data point in the spectrum data of each voxel to be judged relative to the corresponding metabolite data point in the spectrum data of the central voxel before adjustment and after adjustment; taking the normalized dynamic time-warping distance difference as a distance difference factor.
In the present embodiment, the chemical shift deviation factorDifference factor from subinterval>The calculation formula is as follows:
in the method, in the process of the invention,representing +.f. in each voxel to be judged>A chemical shift deviation factor of each fluctuation window relative to the center voxel; />Representing +.f. in each voxel to be judged>A distance difference factor of each fluctuation window relative to the center voxel;metabolite data points in the spectroscopic data representing each voxel to be judged (th +.>A fluctuation window) is located with respect to the corresponding metabolite data point (th +.>A fluctuation window); />Metabolite data points in the spectroscopic data representing each voxel to be judged (th +.>A fluctuation window) before adjustment and after adjustment corresponding metabolite data points in the spectral data relative to the central voxel (th->A fluctuation window); />Representing a normalization function; the first +.>After two different factors between each fluctuation window, as the distance between different metabolite peaks is calculated in the existing calculation or the same relative distance is calculated in the dynamic regulation process through the distance matrix, the distance between the data points is larger when the data points are in different subintervals and the distance between the data points is corrected according to the deviation of the peaks when the data points are in the same subintervals because the distance between the different metabolite peaks is corrected through the chemical displacement deviation factor and the subinterval distance difference factor.
Specifically, optimizing the data point distance of the same metabolite according to the chemical displacement deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel and the distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel; acquiring the optimized DTW distance of the same metabolite data points;
and optimizing the distances of different metabolite data points according to the chemical displacement deviation factors of different metabolite data points of each voxel to be judged relative to the central voxel and the distance difference factors of different metabolite data points of each voxel to be judged relative to the central voxel, and obtaining the optimized DTW distances of different metabolite data points.
In this embodiment, the data point distance of the same metabolite is optimized according to the chemical shift deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel and the distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel; in particular for the optimal distance of data points between a central voxel and sub-intervals of the same metabolite in the voxel to be determinedThe calculation formula is as follows:
for middle partOptimized distance of data points between subintervals of different metabolites in cardiac voxels and voxels to be determinedThe calculation formula is as follows:
in the method, in the process of the invention,indicate->First>Initial DTW distance of data points from data points in the same metabolite in the center voxel; />Representing the +.f in each voxel to be judged>The first +.in the spectral line of each fluctuation window and center voxel>The individual fluctuation windows correspond to all fluctuation windows between the metabolite and the current metabolite; />Representing +.f. in each voxel to be judged>A chemical shift deviation factor of each fluctuation window relative to the center voxel; />Representing +.f. in each voxel to be judged>Each wave window phaseA distance difference factor for the center voxel; />An optimized DTW distance representing a data point between a central voxel and a subinterval of the same metabolite in the voxel to be determined. Therefore, the peak data point of the metabolite peak of the voxel to be judged is adjusted to be positioned at the same position as the peak of the metabolite peak of the central voxel, the DTW distance between the data point in the voxel to be judged and the data point in the central voxel is optimized, and the optimized DTW distance is obtained, namely, the corresponding distance between the data points in the two spectral lines in the DTW dynamic regulation process is calculated and corrected by the chemical displacement factor and the distance difference factor, so that the data points in different metabolite subintervals in the traditional DTW dynamic regulation process are prevented from being corresponding. The DTW distance can ensure that the corresponding relation between data points is kept in the corresponding relation of subintervals through the correction of the distance, so that the DTW distance between the to-be-judged voxel and the spectral line in the central voxel is obtained, and the subsequent spectral line similarity measurement can be carried out according to the DTW distance.
S3, acquiring the relative concentration of voxels from the center voxel to each voxel to be judged to form a relative concentration sequence, and acquiring a differential sequence of the relative concentration sequence;
according to the number of voxels contained between the central voxel and each voxel to be judged, the number of voxels with continuous gradient change of the relative concentration of the voxels contained between the central voxel and each voxel to be judged, and the differential sequence of the relative concentration sequence, acquiring the relative concentration gradient coincidence degree of each voxel to be judged;
specifically, the relative concentration of the voxels from the center voxel to each voxel to be judged forms a relative concentration sequence, which comprises the following steps: acquiring a voxel sequence formed by a central voxel and voxels contained between each voxel to be judged; taking a sequence formed by the relative concentration corresponding to each voxel in the voxel sequence as a relative concentration sequence; the voxel sequence is formed by all voxels on the connecting line from each voxel to be judged to the central voxel.
It should be noted that the number of the substrates,by optimizing the distance calculation between the center voxel and the data point of each voxel to be judged, after the distance between the magnetic resonance spectrum lines corresponding to the voxels is obtained, the measurement of the similarity can be further optimized through the metabolite concentration characteristic of the tumor region in the scene for the voxels outside the region of interest, namely all the voxels to be judged, within the field of view. In the conventional sequence similarity measurement, the degree of similarity between sequences needs to be obtained by normalizing the distances between all sequences. But only the sequence distance between the spectral lines is considered in the determination of this degree of similarity. In the magnetic resonance spectrum line, the extraction of the tumor area also needs to be based on the metabolite concentration in the magnetic resonance spectrum corresponding to the voxels. And optimizing the similarity degree of voxels obtained by the distance between spectral lines through the coincidence degree of the concentration gradient process, so as to obtain a more accurate tumor region. In this embodiment, first, the voxel with the highest metabolite concentration in the region of interest is obtained by the metabolite concentration determination method, all voxels on the connecting line of the voxel with the highest metabolite concentration and the voxel to be determined are obtained, and the distance between each voxel and the previous voxel on the magnetic resonance spectrum line is obtained from the voxel with the highest metabolite concentration, that is, the optimized DTW distance is used. Secondly, using the corresponding spectral line of each voxel to be judged as the metabolite concentration indexAnd determining the coincidence degree of the gradient concentration as the relative concentration of each voxel to be judged. And finally, adjusting the sequence distance in the connectivity outlier detection process of the distance in the sequence through the concentration gradient coincidence degree, so as to obtain the accurate outlier degree of the voxel to be judged, and obtaining the interest degree of the voxel by subtracting the outlier degree from 1.
In this embodiment, the calculation formula of the relative density gradient coincidence degree of each voxel to be judged is as follows:
in the method, in the process of the invention,indicate->The relative concentration gradient coincidence degree of each voxel to be judged; />Representing center voxel to->The number of voxels included between the voxels to be judged; />Representing center voxel to->The number of voxels with continuous gradient change of the relative concentration of the voxels contained between the voxels to be judged is determined by the method of the +.>The voxels to be judged judge the signs of the relative concentration differences to the central voxels, and the number of continuous signs is used as the number of voxels with continuous gradient changes of the relative concentration; />Representing the value of the t element in the differential sequence; />Representing the number of elements in the differential sequence; />Indicate->The relative concentration difference value of each voxel to be judged and the previous voxel;representing the normalization function. For this purpose, the degree of gradient of the gradient process is measured by the number of voxels that are continuously negative, for the concentration values in the sequence, a differential sequence is obtained by differencing the concentration sequence, for the differential sequence it represents the concentration difference between every two voxels, and the degree of coincidence of the gradient state of the metabolite concentration in the magnetic resonance spectrum corresponding to the voxel is measured by the difference value of the voxel to be judged and the preceding voxel and the average difference of the concentration differential sequence.
It should be noted that, for the metabolite concentration information in the magnetic resonance spectrum corresponding to the voxels in the concentration sequence, whether the voxel sequence stops gradual change is judged by the sign of the differential sequence of the concentration value sequence, and whether the position of the voxel to be judged has gradual change stability is judged according to the difference between the differential value of the voxel to be judged and the differential mean value of the sequence, so as to determine the gradual change coincidence degree of the voxel to be judged. For this purpose, a gradient compliance is obtained from the metabolite concentration information of the voxels to be determined in the sequence of the magnetic resonance spectrum data. Compared with the traditional similarity degree judgment obtained through distance normalization, whether the voxel to be judged is placed in the region of interest or not can be corrected through metabolite concentration information in the tumor region magnetic resonance data, so that the accurate similarity degree between the voxel to be judged and the voxel of the region of interest can be obtained.
S4, acquiring an average link distance of each voxel to be judged based on a connectivity outlier factor algorithm according to a voxel sequence formed by voxels contained between a central voxel and each voxel to be judged, the relative concentration gradient coincidence degree of each voxel to be judged and the optimized DTW distance;
acquiring connectivity outlier factors of each voxel to be judged according to the average link distance of each voxel to be judged; acquiring the outlier degree of each voxel to be judged according to the connectivity outlier factor of each voxel to be judged;
it should be noted that, after the gradient coincidence degree of the metabolite concentration in the voxel magnetic resonance spectrum sequence is obtained, the path distance of each data point may be corrected in the process of obtaining the connectivity-based outlier degree of the voxel to be judged in the voxel sequence. The interest degree of the voxel to be judged can be judged by the outlier degree of the voxel in the voxel sequence, and when the outlier degree of the magnetic resonance spectrum line corresponding to the voxel based on connectivity in the voxel sequence is higher, the spectrum line is not placed in the region of interest, namely the interest degree is lower.
For calculation of the outlier factor (COF) of connectivity, the number of voxels in the voxel sequence is taken as the K value in the COF calculation process, i.e. each voxel performs calculation of the outlier degree through all voxels. And the optimized DTW distance between every two voxels in the voxel sequence is used as the path distance. In the course of calculating the outlier factor by means of the path distance, however, the higher the degree of coincidence of the corresponding concentration gradient of a voxel, the lower the distance between the spectral line of that voxel and the line of the preceding voxel should be corrected because of the coincidence of the concentration with the gradient process. The path distance to the spectral line is corrected by the gradient coincidence level, so that the outlier factor in the spectral line sequence for each voxel is more accurate.
Specifically, for the average link distance of each voxel to be judged, for the firstThe average link distance calculation method after optimizing each voxel to be judged is as follows:
in the method, in the process of the invention,represents the>A voxel; />Representing a data point index in the local SBN path; />Indicate->A plurality of paths; />Indicate->Path length of each path, namely optimized DTW distance between spectral lines; />Indicate->The degree of conformity of the relative concentration gradation of the individual elements; />Representing +.>It is taken as the first data point in the SBN path according to the corresponding +.>I.e. +.>Average link distance in neighborhood range; the k neighborhood range is the whole data in the sequence. The method is characterized in that the path distance, namely the optimized DTW distance, is corrected by the degree of coincidence of metabolite concentration information in magnetic resonance spectrum lines corresponding to voxels in the sequence to gradual change characteristics of the region of interest on the original basis through a calculation mode of average link distance in COF outlier factors. Therefore, in the process of measuring the outlier factors, voxels meeting the conditions, which are caused by large distance difference between single spectral lines but concentration of which meets the gradual change of a tumor area, can be prevented from being judged as being incapable of adding the region of interest. For this purpose, the magnetic resonance corresponding to each voxel is usedThe metabolite concentration coincidence degree in the spectrum lines corrects the path length to the voxel in the sequence outlier measurement process, and compared with the average link distance in the traditional sequence, the path distance can be stretched according to the concentration coincidence degree, so that the voxel with the concentration coincidence gradual change can be considered to be added into the region of interest when the distance between the spectrum lines is larger. In this embodiment, after the average link distance is obtained, the average link distance is used to further calculate the connectivity outlier factor of each voxel to be determined corresponding to each voxel to be determined in the sequence, and the connectivity outlier factor of each voxel to be determined is used to normalize to obtain the outlier degree of each voxel to be determined.
S5, judging whether the region of interest is added or not according to the outlier degree of each voxel to be judged to obtain the nuclear magnetic resonance tumor region.
Specifically, the nuclear magnetic resonance tumor area is extracted according to the following steps: acquiring the interest degree of each voxel to be judged according to the outlier degree of each voxel to be judged; setting a threshold value; when the interest degree of the voxel to be judged is greater than a threshold value, adding the voxel to be judged into the region of interest; and sequentially judging each voxel to be judged, adding all voxels to be judged which are larger than a threshold value into the region of interest, and obtaining the nuclear magnetic resonance tumor region with the central voxel as the center.
In this embodiment, after obtaining the outlier degree of each voxel to be determined, the method includes the steps ofSubtracting the outlier degree obtains its interest degree, i.e. an indication of whether it can be added to the region of interest. And for the judgment standard, taking the average value of the interest degrees of all voxels except the voxel to be judged as a threshold value, adding the voxel into the interest region when the interest degree of the voxel to be judged is higher than the threshold value, and taking the finally obtained interest region as an extracted tumor region, so that a more accurate tumor region is extracted in nuclear magnetic resonance detection. In one embodiment of the invention, the threshold is set to 0.7.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (9)
1. The nuclear magnetic resonance tumor region extraction method is characterized by comprising the following steps of:
acquiring a region of interest in a magnetic resonance image; acquiring a central voxel in a region of interest; taking voxels in the region outside the region of interest as voxels to be judged; the relative concentration of the tetramethyl amine relative to the creatine phosphate and the creatine in each voxel and spectrum data are obtained;
acquiring initial DTW distances between a central voxel and each voxel to be judged by using all data points in spectrum data of the central voxel and each voxel to be judged;
optimizing the initial DTW distance between the central voxel and each voxel to be judged according to the positions of the metabolite data points in the spectrum data of the central voxel and each voxel to be judged to obtain the optimized DTW distance between the central voxel and each voxel to be judged;
acquiring the relative concentration of voxels from the center voxel to each voxel to be judged to form a relative concentration sequence, and acquiring a differential sequence of the relative concentration sequence;
according to the number of voxels contained between the central voxel and each voxel to be judged, the number of voxels with continuous gradient change of the relative concentration of the voxels contained between the central voxel and each voxel to be judged, and the differential sequence of the relative concentration sequence, acquiring the relative concentration gradient coincidence degree of each voxel to be judged;
acquiring the average link distance of each voxel to be judged based on a connectivity outlier factor algorithm according to a voxel sequence formed by voxels contained between a central voxel and each voxel to be judged, the relative concentration gradient coincidence degree of each voxel to be judged and the optimized DTW distance;
acquiring connectivity outlier factors of each voxel to be judged according to the average link distance of each voxel to be judged; acquiring the outlier degree of each voxel to be judged according to the connectivity outlier factor of each voxel to be judged;
judging whether a region of interest is added to obtain a nuclear magnetic resonance tumor region according to the outlier degree of each voxel to be judged;
the optimized DTW distance is obtained according to the following steps:
adjusting the position of a metabolite data point in the spectrum data of each voxel to be judged to be at the same position as a corresponding metabolite data point in the spectrum data of the central voxel, and acquiring a chemical displacement deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel and a distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel; obtaining chemical displacement deviation factors of different metabolite data points of each voxel to be judged relative to the central voxel and distance difference factors of different metabolite data points of each voxel to be judged relative to the central voxel;
optimizing the data point distance of the same metabolite according to the chemical displacement deviation factor of the same metabolite data point of each voxel to be judged relative to the central voxel and the distance difference factor of the same metabolite data point of each voxel to be judged relative to the central voxel; acquiring the optimized DTW distance of the same metabolite data points;
optimizing the distances of different metabolite data points according to the chemical displacement deviation factors of different metabolite data points of each voxel to be judged relative to the central voxel and the distance difference factors of different metabolite data points of each voxel to be judged relative to the central voxel, and obtaining the optimized DTW distances of different metabolite data points;
the optimized DTW distance for the same metabolite data point and the optimized DTW distance for a different metabolite data point are taken as the optimized DTW distances.
2. The method for extracting nmr tumor regions according to claim 1, wherein the chemical shift deviation factor of the same metabolite data point of each voxel to be determined relative to the central voxel is obtained by:
when the positions of the metabolite data points in the spectrum data of each voxel to be judged are adjusted, acquiring the chemical displacement difference of the positions of the metabolite data points in the spectrum data of each voxel to be judged relative to the corresponding metabolite data points in the spectrum data of the central voxel; normalized chemical shift difference was used as a chemical shift deviation factor.
3. The method for extracting nmr tumor regions according to claim 1, wherein the distance difference factor of the same metabolite data point of each voxel to be determined relative to the central voxel is obtained by:
when the position of a metabolite data point in the spectrum data of each voxel to be judged is adjusted, acquiring the dynamic time regular distance difference of the metabolite data point in the spectrum data of each voxel to be judged relative to the corresponding metabolite data point in the spectrum data of the central voxel before adjustment and after adjustment; taking the normalized dynamic time-warping distance difference as a distance difference factor.
4. The method of claim 1, wherein the relative concentration of each voxel is obtained by:
index the metabolite concentration in each voxelAs the relative concentration of each voxel; wherein Cho represents the concentration of the metabolite being tetramethylamine; cr represents the concentration of creatine phosphate and creatine.
5. The method of claim 1, wherein the relative concentration of voxels included between the center voxel and each voxel to be determined forms a relative concentration sequence, comprising:
acquiring a voxel sequence formed by a central voxel and voxels contained between each voxel to be judged;
taking a sequence formed by the relative concentration corresponding to each voxel in the voxel sequence as a relative concentration sequence;
the voxel sequence is formed by all voxels on the connecting line from each voxel to be judged to the central voxel.
6. The method of claim 1, wherein the nmr tumor region is extracted according to the steps of:
acquiring the interest degree of each voxel to be judged according to the outlier degree of each voxel to be judged;
setting a threshold value;
when the interest degree of the voxel to be judged is greater than a threshold value, adding the voxel to be judged into the region of interest;
and judging each voxel to be judged in sequence, and adding all voxels to be judged which are larger than a threshold value into the region of interest to obtain the nuclear magnetic resonance tumor region.
7. The method according to claim 1, wherein the central voxel of the region of interest is a voxel having a highest average concentration of metabolites in all spectrum data of the region of interest.
8. The nuclear magnetic resonance tumor region extraction method according to claim 6, wherein the threshold is set to 0.7.
9. The method of claim 1, wherein the region of interest is obtained by human labeling.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311014005.5A CN116740064B (en) | 2023-08-14 | 2023-08-14 | Nuclear magnetic resonance tumor region extraction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311014005.5A CN116740064B (en) | 2023-08-14 | 2023-08-14 | Nuclear magnetic resonance tumor region extraction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116740064A CN116740064A (en) | 2023-09-12 |
| CN116740064B true CN116740064B (en) | 2023-10-20 |
Family
ID=87909996
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311014005.5A Active CN116740064B (en) | 2023-08-14 | 2023-08-14 | Nuclear magnetic resonance tumor region extraction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116740064B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1420367A1 (en) * | 2002-11-15 | 2004-05-19 | MeVis GmbH | A method for coloring of voxels and image data processing and visualization system |
| CN106600579A (en) * | 2016-11-30 | 2017-04-26 | 上海联影医疗科技有限公司 | Magnetic resonance image wave spectrum analysis method and device |
| CN109862911A (en) * | 2016-07-13 | 2019-06-07 | 比奥根Ma公司 | The dosage of LINGO-1 antagonist and purposes for treating demyelinating disorder |
| CN110827242A (en) * | 2019-10-23 | 2020-02-21 | 华中科技大学 | Intracranial aneurysm detection method and system based on convolutional neural network |
| KR20200041616A (en) * | 2018-10-12 | 2020-04-22 | 연세대학교 원주산학협력단 | Apparatus and method for brain network analysis using magnetic resonance spectroscopy |
| CN113808090A (en) * | 2021-09-08 | 2021-12-17 | 中国科学院深圳先进技术研究院 | Magnetic resonance image processing method, terminal device and computer storage medium |
| CN113987905A (en) * | 2020-12-18 | 2022-01-28 | 天津市特种设备监督检验技术研究院(天津市特种设备事故应急调查处理中心) | Escalator braking force intelligent diagnosis system based on deep belief network |
| WO2022099303A1 (en) * | 2020-11-06 | 2022-05-12 | The Regents Of The University Of California | Machine learning techniques for tumor identification, classification, and grading |
| CN114496280A (en) * | 2022-01-14 | 2022-05-13 | 西北工业大学 | Multi-element time sequence data enhancement method for wound surface microenvironment data |
| CN115774227A (en) * | 2021-09-07 | 2023-03-10 | 上海联影医疗科技股份有限公司 | Spectral imaging method and magnetic resonance imaging system |
| CN116385438A (en) * | 2023-06-05 | 2023-07-04 | 济南科汛智能科技有限公司 | Nuclear magnetic resonance tumor region extraction method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070249928A1 (en) * | 2006-04-19 | 2007-10-25 | General Electric Company | Method and system for precise repositioning of regions of interest in longitudinal magnetic resonance imaging and spectroscopy exams |
| AU2020334980A1 (en) * | 2019-08-20 | 2022-03-03 | Terran Biosciences, Inc. | Neuromelanin-sensitive MRI for assessing Parkinson's disease |
-
2023
- 2023-08-14 CN CN202311014005.5A patent/CN116740064B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1420367A1 (en) * | 2002-11-15 | 2004-05-19 | MeVis GmbH | A method for coloring of voxels and image data processing and visualization system |
| CN109862911A (en) * | 2016-07-13 | 2019-06-07 | 比奥根Ma公司 | The dosage of LINGO-1 antagonist and purposes for treating demyelinating disorder |
| CN106600579A (en) * | 2016-11-30 | 2017-04-26 | 上海联影医疗科技有限公司 | Magnetic resonance image wave spectrum analysis method and device |
| KR20200041616A (en) * | 2018-10-12 | 2020-04-22 | 연세대학교 원주산학협력단 | Apparatus and method for brain network analysis using magnetic resonance spectroscopy |
| CN110827242A (en) * | 2019-10-23 | 2020-02-21 | 华中科技大学 | Intracranial aneurysm detection method and system based on convolutional neural network |
| WO2022099303A1 (en) * | 2020-11-06 | 2022-05-12 | The Regents Of The University Of California | Machine learning techniques for tumor identification, classification, and grading |
| CN113987905A (en) * | 2020-12-18 | 2022-01-28 | 天津市特种设备监督检验技术研究院(天津市特种设备事故应急调查处理中心) | Escalator braking force intelligent diagnosis system based on deep belief network |
| CN115774227A (en) * | 2021-09-07 | 2023-03-10 | 上海联影医疗科技股份有限公司 | Spectral imaging method and magnetic resonance imaging system |
| CN113808090A (en) * | 2021-09-08 | 2021-12-17 | 中国科学院深圳先进技术研究院 | Magnetic resonance image processing method, terminal device and computer storage medium |
| CN114496280A (en) * | 2022-01-14 | 2022-05-13 | 西北工业大学 | Multi-element time sequence data enhancement method for wound surface microenvironment data |
| CN116385438A (en) * | 2023-06-05 | 2023-07-04 | 济南科汛智能科技有限公司 | Nuclear magnetic resonance tumor region extraction method |
Non-Patent Citations (3)
| Title |
|---|
| Modified dynamic time warping (MDTW) for estimating temporal dietary patterns;Nitin Khanna 等;《2017 IEEE Global Conference on Signal and Information Processing (GlobalSIP)》;全文 * |
| MRS波谱分析在胶质瘤术后的临床应用;佟鑫;徐广军;徐秋实;房晓萱;;中国实验诊断学(第02期);全文 * |
| 高b值DWI联合DCE-MRI对前列腺癌的诊断价值;曾俊杰;邓巧玲;李胜;陈松;;中国医刊(第08期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116740064A (en) | 2023-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107230197B (en) | Tropical cyclone objective strength determination method based on satellite cloud image and RVM | |
| CN113017596A (en) | Magnetic resonance multi-parameter quantification method and application thereof | |
| CN117764864A (en) | nuclear magnetic resonance tumor visual detection method based on image denoising | |
| Simicic et al. | In vivo macromolecule signals in rat brain 1H‐MR spectra at 9.4 T: parametrization, spline baseline estimation, and T2 relaxation times | |
| CN102866373B (en) | Temperature measurement correction method and system in magnetic resonance temperature imaging | |
| CN110501664A (en) | The magnetic resonance imaging system of Magnetic Resonance Spectrum method for correcting phase and application this method | |
| CN116740064B (en) | Nuclear magnetic resonance tumor region extraction method | |
| CN110553631B (en) | Water level measurement series error analysis method about water level flow relation | |
| CN102831599B (en) | Registering method of medical image with uneven light and shade | |
| CN113049118B (en) | Flash element testing device and method for infrared focal plane detector | |
| CN110647870B (en) | Method for calculating approximate entropy of resting state fMRI data based on sliding window | |
| CN109655415B (en) | Wavelength offset correction method and device and computer equipment | |
| CN107782447B (en) | Space dimension automatic identifying method and system in imaging spectrometer spectral calibration | |
| CN110807768A (en) | Remote sensing image quality evaluation method based on MTF | |
| US9916648B1 (en) | Method for correcting a diffusion image having an artifact | |
| CN104832161A (en) | Automatic depth correction method based on dual-scale correlation contrast | |
| CN114037023A (en) | Method for correcting abnormal water vapor data of polar region of microwave radiometer by using marine satellite | |
| CN112129407B (en) | Method for obtaining full-moon phase period moon disk irradiance of slit type spectrometer | |
| Jog et al. | Pulse sequence based multi-acquisition MR intensity normalization | |
| Belsley et al. | Optimal flip angles for in vivo liver 3D T1 mapping and B1+ mapping at 3T | |
| TH2101003565A (en) | How to Identify an Incorrectly Calibrated or Unscaled Infrared Spectrometer? | |
| WO2022188026A1 (en) | Magnetic resonance multi-parameter quantification method and application thereof | |
| CN117664909B (en) | Thermal cycle plate nondestructive testing method based on near infrared spectrum | |
| CN118310635B (en) | Focal plane detection method and system for infrared detector | |
| CN117877078A (en) | Vein equipment infrared lamp bead abnormality detection method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |