A research team from Children's Hospital Los Angeles has developed a novel computational model for brain blood flow that enhances the accuracy of magnetic resonance imaging (MRI) in assessing cerebral blood flow. This research, led by the Borzage Laboratory, was published in the journal Frontiers in Physiology, providing new technical support for clinical diagnosis.
Monitoring cerebral blood flow holds significant value for diagnosing occult brain injuries and diseases. During actual scanning, patient movement, variations in manual operation, or vascular structural anomalies can lead to data loss. Lead author of the study, Dr. Eamon Doyle, stated: "It is almost impossible to perfectly capture blood flow data from all four cerebral arteries, and technicians may fail to notice issues due to the degree of error."
The research team performed 258 phase-contrast MRI scans on 196 subjects, including data from 108 children and 88 adults. The pediatric group encompassed patients with seizures and tumors. By establishing a mathematical model, the researchers achieved precise estimation of cerebral blood flow across different age groups, even in cases where data from some vessels were missing.
Corresponding author Dr. Matthew Borzage noted: "These computational models demonstrate that even with only partial datasets, the data can still be effectively repaired and utilized." This technology can also be applied to pathological cerebral blood flow assessment, enabling clinicians to observe microvascular structures in the brain using standard 3T phase-contrast MRI equipment, which is typically used for cardiac imaging examinations.
The research team's next steps will involve advancing the automation of the analysis process to enable real-time imaging error correction. Dr. Borzage added: "The strength of the study lies in using a heterogeneous sample population that includes both children and adults, which helps deepen the understanding of population-wide patterns. By leveraging these cerebral blood flow data to identify anomalies and establish richer data correlations, diagnostic outcomes can be effectively improved."
