Search

CN-121982376-A - Tumor image biomarker quantitative analysis system based on multi-mode MRI voxel space coupling and application

CN121982376ACN 121982376 ACN121982376 ACN 121982376ACN-121982376-A

Abstract

The application relates to the technical field of biology, and particularly discloses a tumor image biomarker quantitative analysis system based on multi-mode MRI voxel space coupling and application thereof. According to the application, the spatial coupling relation between the blood brain barrier damage and perfusion in the tumor can be quantitatively reflected through the voxel overlapping region proportion of the T1CE and CBF signals, the 1p/19q state can be efficiently predicted, the accurate prediction of IDH mutation and WHO classification (2-3-level vs. 4) can be realized in cooperation with clinical factors, and a high-efficiency and potential noninvasive imaging biomarker is provided for ADG preoperative molecular typing and classification prediction.

Inventors

  • YANG RUIMENG
  • LIN JIAXIN
  • LAI SHENGSHENG
  • LIANG FANGRONG
  • WEI RUILI
  • ZHANG WANLI
  • ZHAN JIE
  • ZHANG ZHENYANG

Assignees

  • 广州市第一人民医院（广州消化疾病中心、广州医科大学附属市一人民医院、华南理工大学附属第二医院）

Dates

Publication Date
20260505
Application Date
20251230

Claims (10)

  1. 1. A tumor image biomarker based on multi-mode MRI voxel space coupling, which is characterized in that the tumor image biomarker is the overlapping area proportion of T1CE and ASL_CBF voxels; defining a spatial overlap region at each percentile threshold: 1) The overlapping area T T1CE T CBF of the Top T1CE and Top CBF voxels reflects a co-localization area of vascular abnormality, blood brain barrier destruction and high perfusion, and represents a relatively active area of the tumor; 2) The Bottom T1CE and Top CBF voxel overlap region B T1CE T CBF , which reflects the region of increased perfusion but less reinforcement, represents a vascular subregion of abnormal blood vessels and relatively preserved or dysfunctional blood brain barrier; The ratio of the overlapping area of T1CE and ASL_CBF voxels is the ratio of the number of overlapping area voxels under each threshold value to the total number of the ROI voxels of the maximum layer of the whole tumor under the corresponding threshold value, and the calculation formula of the ratio is as follows: Or (b) ; Wherein: n represents the set percentile threshold, 10% -90% and the step size 10%; n represents a prime number; T T1CE n is a voxel set with the signal intensity of T1CE not less than the nth percentile; b T1CE n is the set of voxels with T1CE signal intensity < n-th percentile; T CBF n is a voxel set with CBF signal intensity not less than the nth percentile; ROIn denotes the n% set of ROI voxels.
  2. Application of T1CE and ASL_CBF voxel overlapping region proportion in noninvasive prediction in glioma molecular typing and grading, wherein the T1CE and ASL_CBF voxel overlapping region proportion is a ratio of the number of overlapping region voxels under each threshold value to the total number of total ROI voxels of the whole tumor maximum layer under the corresponding threshold value, and a calculation formula of the ratio is as follows: Or (b) ; Wherein: n represents the set percentile threshold, 10% -90% and the step size 10%; n represents a prime number; T T1CE n is a voxel set with the signal intensity of T1CE not less than the nth percentile; b T1CE n is the set of voxels with T1CE signal intensity < n-th percentile; T CBF n is a voxel set with CBF signal intensity not less than the nth percentile; ROIn denotes the n% set of ROI voxels.
  3. 3. The use of claim 2, wherein the glioma comprises an adult diffuse glioma.
  4. 4. The use of claim 2, wherein the glioma molecular typing comprises IDH mutation and 1p/19q co-deletion, and the fractionation comprises WHO 2 grade vs. 3-4 and WHO 2-3 grade vs. 4.
  5. 5. The use according to claim 2, wherein the T1CE and CBF voxels of the tumor maximum level ROI are extracted separately and sorted from high to low signal intensity, 10% -90%, step size 10% are set, and the T1CE and CBF map voxels are divided into Top voxels and Bottom voxels for a total of nine percentile thresholds.
  6. 6. A tumor image biomarker quantitative analysis system based on multi-mode MRI voxel space coupling, characterized by comprising the following steps: s1, selecting adult diffuse glioma patients, and collecting and processing image data; s2, detecting an IDH mutation state by adopting a polymerase chain reaction, and detecting a 1p/19q co-deletion state by adopting a fluorescence in situ hybridization method; s3, respectively extracting T1CE and CBF voxel values of a tumor maximum layer ROI, sequencing from high to low according to signal intensity, setting a percentile threshold, dividing the T1CE and CBF image voxels into Top voxels and Bottom voxels, defining two types of space overlapping areas under each percentile threshold, carrying out standardization processing on the two types of overlapping areas, calculating the ratio of the number of overlapping areas under each threshold to the total number of the full tumor maximum layer ROI voxels under the corresponding threshold, and constructing the ratio of the T1CE and ASL_CBF voxel overlapping areas; the calculation formula of the ratio is as follows: Or (b) ; Wherein: n represents the set percentile threshold, 10% -90% and the step size 10%; n represents a prime number; T T1CE n is a voxel set with the signal intensity of T1CE not less than the nth percentile; b T1CE n is the set of voxels with T1CE signal intensity < n-th percentile; T CBF n is a voxel set with CBF signal intensity not less than the nth percentile; ROIn represents an n% set of ROI voxels; And S4, carrying out statistical analysis on the data obtained in the step S3, taking the overlapping region proportion of different percentile thresholds as independent variables, constructing a single-factor logistic regression model, and taking the age and the gender into the regression model to construct a multi-factor logistic regression model.
  7. 7. The quantitative analysis system according to claim 6, wherein in the step S1, the collecting and processing of the image data includes the steps of: Image scanning is carried out on adult diffuse glioma patients, the scanning sequence comprises T1WI, T2-FLAIR, T1CE and ASL, a GE HDxt 3.0.0T MR post-processing workstation generates a CBF image, all image data are subjected to N4 bias field correction to eliminate intensity non-uniformity, then non-linear registration is carried out on the image data by Advanced Normalization Tools to the T1WI space, a person with non-ideal registration is subjected to manual correction in ITK-SNAP software, then resampling is carried out to 1X 1mm 3 isotropic voxels to reduce partial volume effect, and a maximum-layer tumor region of interest ROI including tumor cores, necrosis and edema is delineated on the T2WI or the T2-FLAIR.
  8. 8. The quantitative analysis system according to claim 6, wherein in step S3, the T1 CE-asl_cbf voxel overlap region ratio includes a intensified high perfusion overlap region ratio rT T1CE T CBF or a low intensified high perfusion overlap region ratio rB T1CE T CBF ; The enhanced high perfusion overlap region ratio rT T1CE T CBF is more highly trended in a high malignancy subgroup, which includes IDH wild type, 1p/19q non-co-deleted, WHO 3-4 grade, WHO 4 grade; The low-intensity high-perfusion overlap region proportion rB T1CE T CBF is higher in the low-malignancy subgroup, which includes IDH mutant, 1p/19q co-deleted, WHO 2 grade, and WHO 2-3 grade.
  9. 9. The quantitative analysis system according to claim 6, wherein in the step S4, statistical analysis is performed on all data using Zstats 1.0.0 software and Python 3.8, normal evaluation and variance alignment analysis are performed on continuous variables using Shapiro-Wilk test, and group comparison is performed using independent sample t test or Mann-Whitney U test, respectively, according to data distribution characteristics.
  10. 10. The use of the quantitative analysis system according to any one of claims 6-9 for noninvasively predicting molecular typing and classification of adult diffuse gliomas.

Description

Tumor image biomarker quantitative analysis system based on multi-mode MRI voxel space coupling and application Technical Field The application relates to the technical field of biology, in particular to a tumor image biomarker quantitative analysis system based on multi-mode MRI voxel space coupling and application thereof. Background Adult Diffuse Glioma (ADG) is one of the most common and high-mortality primary malignant tumors of the central nervous system. The 2021 WHO CNS tumor classification incorporates molecular features into a diagnostic system, and classifies ADG into three classes, IDH mutation with 1p/19q co-deleted oligodendroglioma, IDH mutation with 1p/19q non-co-deleted astrocytoma and IDH wild-type glioblastoma, based on IDH mutation and 1p/19q co-deleted status. Molecular typing and grading directly affect treatment selection and prognosis evaluation, but tissue biopsy or gene detection is expensive, time consuming and invasive, and presents a sampling bias risk, and there is a critical clinical need for noninvasive, repeatable pre-operative prediction means. In recent years, multi-modal MRI combined with radiology and deep learning has demonstrated potential in pre-glioma prediction, but clinical transformations remain challenging due to complex image processing, multiple feature choices, and sample size limitations. Disclosure of Invention The application aims to overcome the defects of the prior art and provide a tumor image biomarker quantitative analysis system based on multi-mode MRI voxel space coupling and application thereof. In order to achieve the above purpose, the technical scheme adopted by the application is as follows: the application provides a tumor image biomarker based on multi-mode MRI voxel space coupling, wherein the tumor image biomarker is the overlapping region proportion of T1CE and ASL_CBF voxels; defining a spatial overlap region at each percentile threshold: 1) The overlapping area T T1CETCBF of the Top T1CE and Top CBF voxels reflects a co-localization area of vascular abnormality, blood brain barrier destruction and high perfusion, and represents a relatively active area of the tumor; 2) The Bottom T1CE and Top CBF voxel overlap region B T1CETCBF, which reflects the region of increased perfusion but less reinforcement, represents a vascular subregion of abnormal blood vessels and relatively preserved or dysfunctional blood brain barrier; The ratio of the overlapping area of T1CE and ASL_CBF voxels is the ratio of the number of overlapping area voxels under each threshold value to the total number of the ROI voxels of the maximum layer of the whole tumor under the corresponding threshold value, and the calculation formula of the ratio is as follows: Or (b) ; Wherein: n represents the set percentile threshold, 10% -90% and the step size 10%; n represents a prime number; T T1CE n is a voxel set with the signal intensity of T1CE not less than the nth percentile; b T1CE n is the set of voxels with T1CE signal intensity < n-th percentile; T CBF n is a voxel set with CBF signal intensity not less than the nth percentile; ROIn denotes the n% set of ROI voxels. Tumor angiogenesis and perfusion changes are important bases for ADG aggressiveness and therapeutic response. Highly invasive tumors are often accompanied by vascular abnormalities and blood-brain barrier (BBB) destruction, which can be reflected by T1-weighted contrast enhancement (T1-weighted contrast-enhanced, T1 CE). Arterial spin labeling (ARTERIAL SPIN labeling, ASL) MRI techniques can non-invasively measure cerebral blood flow (cerebral blood flow, CBF) and provide information on tumor perfusion and vascular function. The application discusses the application value of the method in the molecular typing and grading before ADG operation through the spatial overlapping proportion analysis of the voxel level T1CE and the CBF signal, and provides a foundation for the development of noninvasive image biomarkers. The application also provides application of the T1CE and ASL_CBF voxel overlapping region proportion in noninvasive prediction in glioma molecular typing and grading, wherein the T1CE and ASL_CBF voxel overlapping region proportion is a ratio of the number of overlapping region voxels under each threshold value to the total number of total ROI voxels of the full tumor maximum layer under the corresponding threshold value, and a calculation formula of the ratio is as follows: Or (b) ; Wherein: n represents the set percentile threshold, 10% -90% and the step size 10%; n represents a prime number; T T1CE n is a voxel set with the signal intensity of T1CE not less than the nth percentile; b T1CE n is the set of voxels with T1CE signal intensity < n-th percentile; T CBF n is a voxel set with CBF signal intensity not less than the nth percentile; ROIn denotes the n% set of ROI voxels. As a preferred embodiment of the use of the present application, the glioma comprises an adult diffuse glioma. As a prefer