Slice-to-Volume Registration of the fetal thorax to assess observed/expected total fetal lungs volume in fetuses with congenital diaphragmatic hernia who underwent Fetoscopic Endoluminal Tracheal Occlusion

Objective: To evaluate the feasibility of slice-to-volume registration (SVR) of the fetal thorax in congenital diaphragmatic hernia (CDH), and to compare observed-to-expected total fetal lung volume (O/E TFLV) obtained from original and SVR MRI volumes for prediction of postnatal survival. Methods: We retrospectively analyzed 35 fetuses with CDH undergoing MRI before and after fetoscopic endoluminal tracheal occlusion (FETO). High-resolution, motion-corrected SVR volumes were reconstructed using an automated pipeline. O/E TFLV was measured on both original and SVR volumes, and results were compared between survivors and non-survivors at 6 months of age. Agreement, correlation, and predictive accuracy were assessed using Pearson correlation and ROC analysis. Results: Four cases (11.4%) were excluded due to uncorrectable artifacts or technical issues. SVR reconstruction required only three acquisitions versus a median of six for conventional imaging. Post-processing took a median of 14 minutes, and lung delineation was longer for SVR than for original volumes. O/E TFLV measurements showed good to very good agreement between original and SVR volumes (r=0.75 before and r=0.96 after FETO), with comparable predictive accuracy for survival (AUCoriginal=0.81–0.83 vs AUCSVR=0.83, p>0.7). Conclusion: SVR of the fetal thorax is feasible in CDH and provides reliable O/E TFLV measurements with prognostic accuracy comparable to conventional MRI, while reducing the need for repeated acquisitions.