Purpose or Case Report: Computed tomography (CT) is widely used in pediatric thoracic imaging but involves ionizing radiation. Zero Echo Time (ZTE) MRI enables silent, free-breathing, non-contrast imaging of short-T2 tissues like lung and bone. We aimed to optimize a 4D ZTE MRI protocol for pediatric chest imaging and evaluate the impact of deep learning (DL) reconstruction on diagnostic image quality.
Methods & Materials: A 4D ZTE MRI sequence was implemented on 1.5T (Signa Artist) and 3T (Signa Premier XT, PET-MR) GE HealthCare scanners under IRB approval. Key parameters included 1.5 mm isotropic resolution, retrospective respiratory soft gating, and optional DL reconstruction. Scan time was under 5 minutes. The study included five healthy volunteers and four pediatric patients. Three pediatric radiologists independently scored image quality (lung parenchyma, airways, bony thorax, and overall) using a 5-point Likert scale. Non-DL and DL reconstructions were compared using the Wilcoxon signed-rank test with Holm correction for multiple comparisons. The primary endpoint was overall image quality in non-DL scans; secondary endpoints were quality improvements using DL.
Results: The protocol produced diagnostic-quality images without sedation. DL reconstruction significantly improved image scores in all categories. Median scores improved from 2.0 to 3.0 for overall image quality (p =0.031), airway (p = 0.047, Holm 0.094), lung parenchyma (p = 0.039, Holm 0.078), and bony thorax (p = 0.062,Holm 0.124). DL enhancement provided better visualization of key thoracic structures and improved consistency across respiratory phases.
Conclusions: This 4D ZTE MRI protocol enables radiation-free imaging of the pediatric chest during free breathing, offering a safer alternative for children requiring repeated scans. It shows strong potential for evaluating pectus excavatum, monitoring pulmonary nodules, and, through respiratory phase binning, assessing small airways and interstitial lung disease.