Study summary
The goal of this observational study is to develop and validate a high-precision AI diagnostic model for cranial diseases by integrating clinical knowledge systems (pathophysiological classification, age stratification, and anatomical localization) to simulate radiologists' diagnostic thinking. The main question it aims to answer is: Does the AI model improve diagnostic accuracy and consistency across different hospital levels, physician qualifications, and clinical scenarios compared to traditional diagnosis? Participants' cranial MRI data (including T1, T2, FLAIR, DWI sequences) and clinical information will be collected retrospectively (2015-2025) and prospectively (2026) to train and validate the model, which will be evaluated through performance metrics (accuracy, sensitivity, specificity) and clinical efficacy assessments (doctor vs. model, with/without model assistance). This study will establish a new paradigm for clinical AI implementation, providing methodological support for precision diagnosis of neurological diseases.
Eligibility
Inclusion Criteria:
Patients clinically diagnosed with one of the following 48 cranial diseases: glioma, lymphoma, meningioma, pituitary macroadenoma, ependymoma, choroid plexus papilloma, schwannoma, medulloblastoma, metastatic tumor, chordoma, craniopharyngioma, germinoma, hemangioblastoma, cholesteatoma, teratoma, viral encephalitis, brain abscess, cerebral tuberculosis infection, cryptococcal encephalitis, cerebral cysticercosis, general soft meningitis (including bacterial, fungal infections, and autoimmune meningitis), tuberculous meningitis, ischemic stroke, cerebral venous sinus thrombosis, arteriovenous malformation, cavernous hemangioma, venous developmental malformation, aneurysm, cerebral small vessel disease, epidural hemorrhage, subdural hemorrhage, intracerebral hemorrhage, cerebral contusion, subarachnoid hemorrhage, diffuse axonal injury, multiple sclerosis, hippocampal sclerosis, focal cortical dysplasia, cortical or cerebral fissure malformations, basilar invagination, Chiari malformation, Dandy-Walker malformation, Rathke's cleft cyst, pituitary hypoplasia, adrenoleukodystrophy, Alzheimer's disease, Parkinson's disease, arachnoid cyst.
Exclusion Criteria:
Patients with other severe neurological diseases not included in the 48 specified cranial diseases; Patients unable to provide complete cranial imaging data (e.g., missing images or images of poor quality); Patients with severe cognitive impairment unable to cooperate with the study (e.g., unable to understand study procedures or communicate with site personnel); Pregnant or lactating women; Patients with severe dysfunction of major organs (e.g., heart, liver, kidney) that cannot tolerate study-related examinations; Patients who refuse to sign the informed consent form.