Key points are not available for this paper at this time.
Purpose Integrating the SVRTK methods within the Gadgetron framework enables automated 3D fetal brain and body reconstruction in the low-field 0.55T MRI scanner within the duration of the scan. Methods A deep-learning based, integrated, robust, and deployable workflow from several motion-corrupted individual T2-weighted single-shot Turbo Spin Echo stacks to produce super-resolved 3D reconstructed fetal brain and body is enabled by combining automated deformable and rigid Slice-to-Volume (D/SVR) reconstruction adapted for low field MRI with a real-time scanner-based Gadgetron workflow. Qualitative evaluation of the pipeline in terms of image quality and efficiency is performed in 12 prospectively acquired fetal datasets from the 22-40 weeks gestational age range. Results The reconstructions were available on average 6:42±3:13 minutes after the acquisition of the final stack and could be assessed and archived on the scanner console during the ongoing fetal MRI scan. The output image data quality was rated as good to acceptable for interpretation. The additional retrospective testing of the pipeline on 83 0.55T datasets demonstrated stable reconstruction quality for low-field MRI. Conclusion The proposed pipeline allows scanner-based prospective motion correction for low-field fetal MRI. The main novel components of this work are the compilation of automated fetal and body D/SVR methods into one combined pipeline, the first application of 3D reconstruction methods to 0.55T T2-weighted data, and the online integration into the scanner environment.
Uus et al. (Tue,) studied this question.