Deep learning approach for lung cavity estimation manuscript published

We are pleased to announce that ‘A dual-channel deep learning approach for lung cavity estimation from hyperpolarized gas and proton MRI’ has been published in the Journal of Magnetic Resonance Imaging.

The article assesses a deep learning-based dual-channel approach for lung cavity estimation (LCE). Lung ventilation can be quantified using the ventilation defect percentage (VDP), which is computed from segmentations from hyperpolarised gas MRI and proton (¹H) MRI; however, these scans are frequently misaligned, requiring LCE. A dual-channel approach, using Xenon-129-MRI and ¹H-MRI was assessed to determine if more accurate LCE can be generated versus single-channel deep learning-based methods, across a range of lung pathologies.

Results demonstrated that the dual-channel approach significantly outperformed single-channel approaches and may yield improved LCEs. The accuracy of LCEs did not diminish with changing volumes. In addition, the automatically generated VDPs showed strong agreement with those generated manually.

The full article is available here.

Read a summary of this article, and other NOVELTY publications and presentations here.