1. To retrospectively explore the feasibility of multi-dimensional heterogeneity imaging features of MRI in predicting the status of key gene mutations in high-grade gliomas;
2. To prospectively explore the correlation between multi-dimensional heterogeneous MRI image features and prognosis of high-grade glioma patients.

Glioblastoma, the most prevalent primary intracranial tumor, is characterized by its formidable therapeutic resistance, primarily attributed to its intrinsic heterogeneity. This heightened heterogeneity is not solely confined to inter-tumoral variations across different individuals but also encompasses considerable intratumoral diversity. The pervasive notion among the scientific community posits that this intratumoral heterogeneity substantiates an endogenous mechanism for drug resistance, thereby exerting substantial influence upon the design of clinical trials, prognostic prediction, and patient outcomes. Preceding methodologies for assessment are beleaguered by a constellation of challenges, impeding precise evaluation of global tumor heterogeneity and necessitating innovative modalities to surmount this impasse. MRI imaging, endowed with non-invasiveness and user-friendliness, surmounts the biases of single-point sampling, enabling comprehensive and dynamic appraisal of glioblastomas. Notably, high-grade gliomas exhibit pronounced microenvironmental pressure selectivity and adaptability, akin to species occupation within distinct ecological niches. This phenomenon, termed "habitat," manifests as a visual representation of the tumor's spatial distribution and temporal evolution, thus facilitating real-time, longitudinal monitoring. Given the substantial imaging heterogeneity inherent to glioblastomas, they stand as an opportune subject for habitat imaging techniques compared to their neoplastic counterparts.

The present investigation endeavors to leverage multi-center, multi-dimensional MRI spatial heterogeneity analysis to predict pivotal genes germane to prognosis and therapy in high-grade gliomas, ultimately constructing a stratified prognostic model for afflicted patients.

Inclusion Criteria:

Retrospective Study:

1. Participants aged 18 to 70 years, of any gender.
2. Confirmed postoperative pathology of adult diffuse glioma (WHO Grade III-IV).
3. Standard MR contrast-enhanced imaging performed within 10 days before surgery.
4. No history of prior radiotherapy or chemotherapy before surgery.
5. Absence of concurrent significant comorbidities or other tumors.
6. Presence of molecular testing results (including IDH, MGMT, 1p19q, TERT, CDKN2A/B, BRAF).
7. Availability of comprehensive clinical and follow-up data.

Prospective Study:

1. Participants aged 18 to 70 years, of any gender.
2. Clinically suspected to have high-grade gliomas preoperatively, with final pathology confirming high-grade gliomas.
3. Stable vital signs and capable of cooperating for a 40-minute MR scan.
4. Absence of significant underlying medical conditions or history of other tumors.
5. Documentation of informed consent through a signed consent form.

Exclusion Criteria:

Retrospective Study:

1. MRI images with artifacts or presence of intratumoral hemorrhage.
2. Incomplete clinical data available.

Prospective Study:

1. Individuals with claustrophobia or other reasons unable to undergo MRI scans.
2. History of allergic reactions to MRI contrast agents.
3. Inappropriate for prolonged MRI scans due to other reasons.

High-grade glioma patients were recruited from multiple research centers, including Renji Hospital, Shanghai Jiao Tong University School of Medicine, Huashan Hospital, Fudan University; Shanghai Jing'an District Central Hospital, and Nantong First People's Hospital.