Robust and Rapid 3D High-Resolution Cranial bone imaging for pediatric patients using MRI

Project Summary
Pediatric patients are more vulnerable to radiation exposure when compared to adults. Each year, 2.2 million
pediatric head computed tomography (CT) scans utilizing ionizing radiation are performed in the United States.
Head trauma and craniosynostosis are two of the most common pediatric conditions requiring head CT scans.
Multiple CT scans are often performed during clinical follow-up, exacerbating the cumulative risk of radiation
exposure. Head trauma is common in children, frequently resulting in a skull fracture. Craniosynostosis is a
congenital disability defined by a prematurely fused cranial suture. Standard clinical care for pediatric patients
with head trauma or craniosynostosis employs 3D high-resolution cranial CT images to identify cranial fractures
or cranial suture patency. The National Cancer Institute reported that radiation exposure from multiple head CT
scans in children has the potential to triple the risk of leukemia and brain cancer due to radiosensitivity of their
bone marrow and brain tissue. Magnetic resonance imaging (MRI) is a safe alternative without ionizing radiation.
Existing “black bone” MRI methods rely on a diminished bone signal in a standard gradient echo scan to image
the skull. Though these methods have shown encouraging results, they have not translated into clinical practice
due to several challenges: motion artifacts, long acquisition time, and subjective manual image processing. Since
pediatric patient movement is very common, sedation has been routinely used to minimize motion artifacts in an
MR scan. Unfortunately, sedation is associated with risks including developmental delay and cardiopulmonary
complications. It takes several minutes to acquire high-resolution MR images, which can be challenging for
pediatric subject compliance and limits clinical adoption. Due to poor signal contrast between bone and its
surrounding tissues in MR images, existing manual signal intensity-based approaches are challenging and not
suitable for clinical translation. Our primary goal is to develop novel MR techniques to provide CT-equivalent 3D
high-resolution cranial bone imaging. Four specific aims are proposed: 1) develop motion correction to address
head motion in unsedated pediatric patients; 2) develop an MR image reconstruction method regularized by a
deep-learning prior to reduce MR acquisition time to 1 minute or below; 3) develop a 3D Bayesian neural network
to estimate pseudo-CT (pCT) and uncertainty maps from MRI for robust and automated image post-processing;
and 4) determine the clinical utility of pCT in identifying cranial fractures and cranial suture patency. This study
will have a profound impact on pediatric health by removing the risks associated with radiation and sedation.

Grant Number: 5R01EB032713-04
NIH Institute/Center: NIH
Principal Investigator: Hongyu An