3D Deep Convolutional Neural Network-Based Ventilated Lung Segmentation Using Multi-nuclear Hyperpolarized Gas MRI

Abstract

Hyperpolarized gas MRI enables visualization of regional lung ventilation with high spatial resolution. Segmentation of the ventilated lung is required to calculate clinically relevant biomarkers. Recent research in deep learning (DL) has shown promising results for numerous segmentation problems. In this work, we evaluate a 3D V-Net to segment ventilated lung regions on hyperpolarized gas MRI scans. The dataset consists of 743 helium-3 (³He) or xenon-129 (¹²⁹Xe) volumetric scans and corresponding expert segmentations from 326 healthy subjects and patients with a wide range of pathologies. We evaluated segmentation performance for several DL experimental methods via overlap, distance and error metrics and compared them to conventional segmentation methods, namely, spatial fuzzy c-means (SFCM) and K-means clustering. We observed that training on combined ³He and ¹²⁹Xe MRI scans outperformed other DL methods, achieving a mean ± SD Dice of 0.958 ± 0.022, average boundary Hausdorff distance of 2.22 ± 2.16 mm, Hausdorff 95^th percentile of 8.53 ± 12.98 mm and relative error of 0.087 ± 0.049. Moreover, no difference in performance was observed between ¹²⁹Xe and ³He scans in the testing set. Combined training on ¹²⁹Xe and ³He yielded statistically significant improvements over the conventional methods (p < 0.0001). The DL approach evaluated provides accurate, robust and rapid segmentations of ventilated lung regions and successfully excludes non-lung regions such as the airways and noise artifacts and is expected to eliminate the need for, or significantly reduce, subsequent time-consuming manual editing.