The goal of this multi-center clinical trial is to evaluate the effectiveness of MRI-based computer-aided diagnosis software (V1) for glioma segmentation, gene prediction, and tumor grading. Machine learning methods such as high-precision tumor segmentation and classification and discrimination modeling can further optimize the non-invasive molecular diagnosis and prognosis prediction. The main question it aims to answer is whether the software can predict the molecular type and the prognosis quickly and correctly. The results will be compared with the real-world clinical data double-blindly. Finally, form a set of user-friendly automatic glioma diagnosis and treatment systems for clinics.

BACKGROUND:

The molecular type is crucial for surgical planning and post-operative treatment of glioma. MRI-based radiomics is an emerging technique that extracts unrevealed information including pathology, biomarkers, and genomics by using automated high-throughput extraction of a large number of quantitative features. With the help of artificial intelligence, MRI-based radiomics could be a promising noninvasive method to reveal molecular type by using a quantitative radiomics approach for glioma.

AIM:

MRI-based computer-aided diagnosis software (V1) is an MRI-based radiomics tool with machine learning methods such as high-precision tumor segmentation and classification and discrimination modeling that can further optimize the non-invasive molecular diagnosis and prognosis prediction. The main question it aims to answer is whether the software can predict the molecular type and the prognosis quickly and correctly.

PROCESS:

Participants will read an informed consent agreement before surgery and voluntarily decide whether or not to join the experimental group. They will undergo preoperative multimodal magnetic resonance imaging, which is the routine neuro-images of preoperative evaluation. After surgery, the patient's tumor tissue samples will undergo specialist genetic testing to obtain multiple molecular diagnostic results, such as isocitrate dehydrogenase (IDH), telomerase reverse transcriptase promoter (TERTp), the short arm chromosome 1 and the long arm of chromosome 19 (1p/19q), et al. The participants need to be followed up for 1-year after surgery. Also, their imaging data, genotype data, clinical history data, pathology data, and clinical follow-up data will be analyzed for the study.

The preoperative Multimodality imaging will be input to the software (V1), and glioma segmentation, gene prediction, tumor grading, and lifetime will be analyzed by the software. The results will be compared with the real-world clinical data double-blindly. In order to evaluate the estimation performance of the software, several indexes will be calculated including accuracy (ACC), sensitivity (SENS), and specificity (SPEC). Finally, form a promising set of user-friendly automatic glioma diagnosis and treatment systems for clinics.

Inclusion Criteria:

1. Age front 18 to 70 years old (not including threshold), gender is not limited;
2. Preliminary diagnosis of glioma patients and patients who plan to undergo surgical treatment;
3. Preoperative cranial MRI (T1, T2, T2 Flair, T1 enhanced GE company magnetic resonance package), tumor pathological examination (H&E section, Kuoran Gene Company package), acceptable follow-up and brain MRI scan;
4. The patient himself voluntarily participated and signed the informed consent in writing.

Exclusion Criteria:

1. Patients who only underwent biopsy rather than surgical tumor resection;
2. Postoperative pathologically confirmed non-glioma patients;
3. Patients with multiple glioma metastases or multiple gliomas;
4. Patients who died of complications in the early postoperative period;
5. The researcher believes that this researcher should not be included.

Preliminary diagnosis of glioma patients and patients who plan to undergo surgical treatment