Neonatal Craniocerebral Diseases
Conditions
Interventions
Gold Standard: Two senior physicians with at least 5 years of experience in neonatal cranial ultrasound diagnosis, who are blinded to the study groups and do not communicate with each other, will inde
Index test:artificial intelligence (AI) software-assisted cranial ultrasound
Sponsors
The Third Affiliated Hospital of Guangzhou Medical University
Eligibility
Sex/Gender
All
Age
No minimum to 1 Years
Inclusion criteria
Inclusion criteria: 1. Retrospective Study Phase (1) Neonates who underwent cranial ultrasound examination during hospitalization or outpatient visits in the Department of Neonatology or the Department of Obstetrics of our hospital from January 1, 2017 to July 1, 2026; (2) Complete and clear cranial ultrasound imaging data stored in the hospital PACS system, including standard section images such as coronal, sagittal, and axial planes; (3) Complete basic demographic information, perinatal data, clinical diagnoses, laboratory test results, ultrasound diagnoses, and follow-up outcome data recorded in the electronic medical record system; (4) Complete clinical data with no critical information missing, sufficient for AI model training and data analysis. 2. Prospective Study Phase (1) Approved by the hospital ethics committee, neonates who undergo routine neonatal cranial ultrasound examination in our hospital from August 1, 2026 to July 31, 2031; (2) No restrictions on gestational age, birth weight, or sex, covering full-term, preterm, high-risk, and other neonatal populations; (3) Guardians are informed of the study contents and provide informed consent in accordance with the ethics committee’s requirements; (4) Complete standard cranial ultrasound examination with clear imaging data, and able to cooperate with clinical follow-up.
Exclusion criteria
Exclusion criteria: (1) Cranial ultrasound images are blurred, defective, or missing standard planes, rendering them unsuitable for image analysis and annotation; (2) Critical clinical data—including perinatal information, clinical diagnosis, laboratory tests, and follow-up data—are missing, preventing complete data integration; (3) Presence of severe chromosomal abnormalities, congenital brain malformations, inherited metabolic diseases, or severe malformations of vital organs (e.g., heart, liver, kidneys) that may interfere with the independent diagnosis of intracranial lesions and interpretation of results; (4) The case records contain logical inconsistencies, duplicate entries, or other issues that compromise data authenticity; (5) In the prospective study, guardians refuse to cooperate, are lost to follow-up, or do not meet the relevant ethical requirements.
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| diagnostic accuracy;sensitivity (true positive rate);specificity (true negative rate);area under the ROC curve (AUC) of the AI model;misdiagnosis rate;inter?physician diagnostic agreement (Kappa value); | — |
Secondary
| Measure | Time frame |
|---|---|
| time per single diagnosis;missed diagnosis rate; | — |
Countries
China
Contacts
Public ContactYi Wang
The Third Affiliated Hospital of Guangzhou Medical University
Outcome results
None listed