Lung morphology assessment of cystic fibrosis using MRI with ultra-short echo time at submillimeter spatial resolution

Background Ultrashort echo time (UTE) proton MRI has gained popularity for assessing lung structure and function in pulmonary imaging; however, the development of rapid biomarker extraction and regional quantification has lagged behind due to labor‐intensive lung segmentation. Purpose To evaluate a deep learning (DL) approach for automated lung segmentation to extract image‐based biomarkers from functional lung imaging using 3D radial UTE oxygen‐enhanced (OE) MRI. Study Type Retrospective study aimed to evaluate a technical development. Population Forty‐five human subjects, including 16 healthy volunteers, 5 asthma, and 24 patients with cystic fibrosis. Field Strength/Sequence 1.5T MRI, 3D radial UTE (TE = 0.08 msec) sequence. Assessment Two 3D radial UTE volumes were acquired sequentially under normoxic (21% O2) and hyperoxic (100% O2) conditions. Automated segmentation of the lungs using 2D convolutional encoder‐decoder based DL method, and the subsequent functional quantification via adaptive K‐means were compared with the results obtained from the reference method, supervised region growing. Statistical Tests Relative to the reference method, the performance of DL on volumetric quantification was assessed using Dice coefficient with 95% confidence interval (CI) for accuracy, two‐sided Wilcoxon signed‐rank test for computation time, and Bland–Altman analysis on the functional measure derived from the OE images. Results The DL method produced strong agreement with supervised region growing for the right (Dice: 0.97; 95% CI = [0.96, 0.97]; P < 0.001) and left lungs (Dice: 0.96; 95% CI = [0.96, 0.97]; P < 0.001). The DL method averaged 46 seconds to generate the automatic segmentations in contrast to 1.93 hours using the reference method (P < 0.001). Bland–Altman analysis showed nonsignificant intermethod differences of volumetric (P ≥ 0.12) and functional measurements (P ≥ 0.34) in the left and right lungs. Data Conclusion DL provides rapid, automated, and robust lung segmentation for quantification of regional lung function using UTE proton MRI. Level of Evidence: 2 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2019;50:1169–1181.

Section: Discussionmentioning

confidence: 98%

mentioning

confidence: 99%

Deep convolutional neural networks with multiplane consensus labeling for lung function quantification using UTE proton MRI

Zha

Fain

Schiebler

et al. 2019

“…Recently, ultrashort echo time (UTE) MRI with 3D radial k ‐space trajectories has shown promise in overcoming these obstacles. However, experience with 3D radial UTE chest MR is limited to animal models and small human studies, mostly in adults . Moreover, 3D radial acquisitions suffer from inherent inefficiencies in fully populating k ‐space, leading to trade‐offs between longer scan times and data undersampling with lesser resolution …”

mentioning

confidence: 99%

Free‐breathing pediatric chest MRI: Performance of self‐navigated golden‐angle ordered conical ultrashort echo time acquisition

Zucker

Cheng

Haldipur

et al. 2017

“…The development of ultrashort echo time (UTE) acquisition has enabled obtaining sufficient signal in the parenchyma despite the short T 2 *, while recent technical developments have allowed 3D acquisitions with high spatial resolution . Multiple studies have demonstrated nowadays an interest in noncontrast‐enhanced UTE for the evaluation of various pulmonary and airway disorders, including nodules or masses, ground glass opacity, micronodules, nodules, patchy shadow or consolidation, emphysema or bullae, bronchiectasis, reticular opacity, honeycomb, and traction bronchiectasis . Besides pulmonary disease, UTE imaging has also been evaluated in other regions of the body, with a potential in qualitative and quantitative assessment in the musculoskeletal system, but without current use in routine practice.…”

mentioning

confidence: 99%

Ultrashort echo time imaging of the lungs under high‐frequency noninvasive ventilation: A new approach to lung imaging

Delacoste

Dournes

Dunet

et al. 2019

BackgroundAlthough ultrashort echo time (UTE) sequences allow excellent assessment of lung parenchyma, image quality remains lower than that of computed tomography (CT).PurposeTo investigate a high‐frequency noninvasive ventilation (HF‐NIV) technique allowing a stabilized inspiration and to compare image quality with current dedicated MR sequences.Study TypeProspective.PopulationTen healthy volunteers.Field Strength/Sequence3D radial UTE sequence at 1.5T.AssessmentUTE‐HF‐NIV sequence was compared with UTE‐free‐breathing (UTE‐FB), reconstructed at end expiration (UTE‐Exp) and average (UTE‐Avg), and breath‐hold VIBE sequences. The distance from lung apex to the dome of the right hemidiaphragm was measured. Visual assessment of the visibility and sharpness of normal anatomical structures was carried out. Dedicated software also quantitatively evaluated vessel–lung and right lung–liver interface sharpness. Apparent signal ratio (Sr) and contrast ratios (Cr) were quantitatively evaluated.Statistical TestsWilcoxon signed rank test for visual scores, paired t‐test for continuous variables, significance at P < 0.05.ResultsThe distance between apex and the right hemidiaphragmatic dome was significantly larger (P < 0.001) with UTE‐HF‐NIV compared with UTE‐FB and VIBE acquisitions. Vessel and airway visibility had identical median visual scores with all UTE methods. Median visual scores for sharpness of vessels and airways were significantly higher (P < 0.001) with HF‐NIV (vessels = 3; airways = 2) than in UTE‐FB (vessels = 2; airways = 1) and VIBE (vessels = 1; airways = 1). Software‐based vessel sharpness evaluation resulted in larger values in 8/10 volunteers with UTE‐HF‐NIV (67.3 ± 9.8) compared with UTE‐Avg (62.3 ± 12.6) but the average difference was not significant (P = 0.28). The sharpness of the lung–liver interface was significantly higher (P < 0.001) with HF‐NIV (17.3 ± 5.3) compared with UTE‐Avg (14.1 ± 3.9). Significantly higher values (P < 0.01) of Sr and Cr were observed with UTE‐HF‐NIV compared with UTE‐FB and VIBE.Data ConclusionHF‐NIV allowing acquisition at full inspiration significantly improves image quality for lung imaging. This could offer the option to alternate some follow‐up CT studies by using this technique. Level of Evidence: 2 Technical Efficacy: Stage 1J. Magn. Reson. Imaging 2019;50:1789–1797.