Super‐Resolution Cine Image Enhancement for Fetal Cardiac Magnetic Resonance Imaging

• Published in: Journal of magnetic resonance imaging Vol. 56; no. 1; pp. 223 - 231
• Main Authors: Berggren, Klas, Ryd, Daniel, Heiberg, Einar, Aletras, Anthony H., Hedström, Erik
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 01.07.2022; Wiley Subscription Services, Inc
• Subjects: Artificial neural networks; Cardiac and Cardiovascular Systems; Cardiology and Cardiovascular Disease; Clinical Medicine; congenital heart defect; Doppler effect; Female; fetal cardiovascular magnetic resonance; Fetus - diagnostic imaging; Fetuses; Field strength; Heart; Humans; Image acquisition; Image enhancement; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Image quality; Image reconstruction; Image resolution; Interpolation; Kardiologi; Kardiologi och kardiovaskulära sjukdomar; Klinisk medicin; Magnetic Resonance Imaging; Magnetic Resonance Imaging, Cine - methods; Medical and Health Sciences; Medical imaging; Medicin och hälsovetenskap; Neural networks; Pediatrics; Pediatrik; Prospective Studies; Radiologi och bildbehandling; Radiology and Medical Imaging; Radiology, Nuclear Medicine and Medical Imaging; Reproducibility of Results; Resonance; Statistical analysis; Statistical tests; super resolution; fetal cardiovascular magnetic resonance; super resolution; congenital heart defect
• ISSN: 1053-1807; 1522-2586; 1522-2586
• DOI: 10.1002/jmri.27956

Background Fetal cardiac magnetic resonance imaging (MRI) improves the diagnosis of congenital heart defects, but is sensitive to fetal motion due to long image acquisition time. This may be overcome with faster image acquisition with low resolution, followed by image enhancement to provide clinically useful images. Purpose To combine phase‐encoding undersampling with super‐resolution neural networks to achieve high‐resolution fetal cine cardiac MR images with short acquisition time. Study Type Prospective. Subjects Twenty‐eight fetuses (gestational week 36 [interquartile range 33–38 weeks]). Field Strength/Sequence 1.5 T, balanced steady‐state free precession (bSSFP) cine sequence. Assessment Images were acquired using fully sampled Doppler ultrasound‐gated clinical bSSFP cine as reference, with equivalent cine sequences with decreased phase‐encoding resolution (25%, 33%, and 50% of clinical standard). Two super‐resolution methods based on convolutional neural networks were proposed and evaluated (phasrGAN and phasrresnet). Data were partitioned into training (36 cine slices), validation (3 cine slices), and test sets (67 cine slices) without overlap. Conventional reconstruction methods using bicubic interpolation and k‐space zeropadding were used for comparison. Three blinded observers scored image quality between 1 and 10. Statistical Tests Image scores are reported as median [interquartile range] and were compared using Mann–Whitney's nonparametric test with P < 0.05 showing statistically significant differences. Results Both proposed methods showed no significant difference in image quality compared to clinical images (8 [7–8.5]) down to 33% (phasrGAN 8 [6.5–8]; phasrresnet 8 [7–8], all P ≥ 0.19) phase‐encoding resolution, i.e., up to three times faster image acquisition, whereas bicubic interpolation and k‐space zeropadding showed significantly lower quality for 33% phase‐encoding resolution (both 7 [6–8]). Data Conclusion Super‐resolution enhancement can be used for fetal cine cardiac MRI to reduce image acquisition time while maintaining image quality. This may lead to an improved success rate for fetal cine MR imaging, as the impact of fetal motion is lessened by shortened acquisitions. Level of Evidence 1 Technical Efficacy Stage 2