Impact of contrast enhancement boost and super-resolution deep learning reconstruction on pediatric congenital heart disease CTA scans: ultra-low contrast dose.

OBJECTIVE

To evaluate the feasibility of using contrast enhancement boost (CE-Boost) combined with super-resolution deep learning reconstruction (SR-DLR) to reduce contrast agent dosage in pediatric patients with congenital heart disease (CHD).

METHODS

A total of 72 pediatric CHD patients were divided into the low-contrast-dose group (CE-Boost group, injection of 0.5 mL per kilogram of body weight, n = 36) or the standard-scan group (CE-CT group, 1.5 mL per kilogram of body weight, n = 36), both groups undergoing imaging with SR-DLR. The two imaging protocols were compared based on radiation dose, objective image quality, subjective evaluation, and diagnostic accuracy. To quantitatively evaluate image quality, CT attenuation (HU) and standard deviation (SD) values were measured within ROIs at the four-chamber plane for cardiac chambers; an anonymized dataset was assessed using a double-blind methodology by two independent readers blinded to the clinical information and prior diagnoses of the pediatric patients.

RESULTS

CE-Boost combined with SR-DLR significantly reduced contrast agent usage (62.3% reduction compared to CE-CT, P < 0.001) while maintaining image quality comparable to the conventional contrast protocol (P > 0.05). There was no significant difference in radiation dose parameters, including dose-length product (DLP, mGy·cm) and volume-weighted CT dose index (CTDIvol, mGy) (all P > 0.1), while the effective dose (ED) in the CE-Boost group was slightly lower but not significant (0.36 vs. 0.43 mSv, P = 0.078). Additionally, the CE-Boost group's image quality metrics (CT values, SNR, CNR) remained stable, with no significant difference in subjective scores (P = 0.660).

CONCLUSION

CE-Boost combined with SR-DLR enables a significant reduction in contrast agent usage in pediatric CHD imaging while maintaining comparable image quality to conventional contrast protocols and optimizing SNR and CNR. This approach ensures diagnostic readability while minimizing contrast exposure, highlighting its feasibility and clinical value in pediatric CHD imaging.