MRI of the airways and lungs Including hyperpolarized techniques Pulmonary Medicine Jason C. Woods Radiology Neonatology
Disclosure statements: Financial relationships: Dr. Woods is a consultant to and has received research support from Vertex Pharmaceuticals, Inc. and Grifols, Inc. Neither financial relationship relates directly to the work discussed.
Lung is most challenging solid organ to image 1. Large and moves with respiration (motion artifacts) 2. Low density (r = 0.2 g/cm 3 at TLC) 3. Multiple air-tissue interfaces (alveoli) cause fast MRI decay of signal Techniques we use X-ray (not tomographic) X-ray CT (fairly high ionizing radiation) A few other techniques like ultrasound (streaky results) (EE Vassalou, et al, J Ultrasound Med 2017, doi Sept 6) MRI (no radiation, but historically bad for parenchyma) (now pretty good!)
Challenges led to new innovation for in-vivo imaging 1. UTE and FT MRI sequences (short echo time) Mostly structural imaging (& self gating) Some functional techniques Oxygen enhanced & multi-volume matching 6 cm 2. Neonatal MRI 1.5T ONI / GE hybrid (Cincinnati Children s for 5 yrs, now at Sheffield) 3. Hyperpolarized-gas MRI ( 3 He or 129 Xe) Realtime ventilation (breath hold for 10-15 s) Measure of alveolar-airspace size Measures of gas exchange Healthy Control patient 1 yr Cystic Fibrosis CF patient Neonate 1 yr 2 wks! FEV 1 = 98% FEV 1 = 102%
MRI signal Complementary Techniques: UTE and HP-gas MRI CF is a model lung disease, since structure+function abnormalities 2 1.6 1.2 Structure via UTE MRI Lung parenchymal signal decays quickly with time Radial k-space techniques allow short echo time Lung parenchymal signal Lung T 2 * = 0.8 ms (3T) 1.5 Tesla 3 Tesla Function (ventilation): 129 Xe MRI Gas with strong magnetic signal 129 Xe Subject in MRI Scanner 0.8 0.4 TE = 0.2 ms TE = 1.3 ms Axial Slices Acquired 129 Xe in lungs 0 0 1 2 3 4 5 time after rf pulse (ms) TE = 2.8 ms Defects identified
UTE MRI Total Mean Score CF: UTE MRI comparison to CT: 1-3 yr olds CT FRC UTE MRI FRC Score both MRI and CT via Brody Score Lung Abnormalities 14 12 10 8 6 4 2 Bronchiectasis (BR) Ground glass opacity (GGO) Bronchial wall thickening (BWT) Mucus Plugging (MP) Consolidation (Con) Air trapping (AT) Overall Brody Score y = 0.47x + 0.87 R² = 0.81 0 0 5 10 15 20 25 Roach, et al., Annals ATS CT Total 2016; Mean Score BR GG O BWT MP Con AT AT BR GG O BWT MP Con AT AT
Hyperpolarized 129 Xe MRI Subject in MRI Scanner Control, 6 y.o. female FEV 1 = 95%, VDP = 1.8% Cystic Fibrosis, 15 y.o. female FEV 1 = 72%, VDP = 32.2% Bag of HP 129 Xe Axial Slices Acquired Defects identified Inhaled 129 Xe Cystic Fibrosis, 11 y.o. male FEV 1 = 102%, VDP = 27.5% 1. Thomen et al. J Cyst Fibros. 2016 2. Walkup et al. Pediatr Radiol. 2016
129 Xe MRI in a control subject, and in a patient with CF 14 y.o. male control subject, FEV 1 = 103% (normal lung function) All control subjects: uniform 129 Xe ventilation and low 129 Xe ventilation defect percentage (VDP) 15 y.o. female CF subject, FEV 1 = 73% LU LL RU RM RL
CF FEV 1 = 102% CF FEV 1 = 81% Control FEV 1 = 115% 129 Xe Ventilation Defect Percentage (VDP) in CF 129 Xe ventilation MRI is a very sensitive technique for measuring airway obstruction N = 11 N = 10 Ages 6-16 RL Thomen et al, J Cyst Fibrosis 2016 Jul 28. pii: S1569-1993(16)30565-3. doi: 10.1016/j.jcf.2016.07.008
Together get function, structure, and microstructure Hyperpolarized Xenon MRI UTE MRI function structure LU LL RU RM RL Quantify by lobe, segment, or region
Self-gating proton MRI: motion correction and respiration gating Tidal expiration Tidal inspiration Higano NS, et al., Magn Reson Med 2017; 77: 1284-1295
MRI signal High-resolution pulmonary MRI in neonates is feasible (no sedation, free breathing or normal resp support) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 After new sequences and 2-yr optimization: High-resolution images, high signal Lung parenchymal signal 1.5 Tesla TE = 0.2ms 0 2 4 TE = 2.5ms time after rf pulse (ms) NICU control patient no lung abnormalities Hahn AD, et al, J Magn Reson Imag 2016 doi: 10.1002/jmri.25394
NS Higano, et al., J Magn Reson Imag 2017 (in press) doi 10.1002/jmri.25643 JM Stein, et al., Pediatric Radiology 2016; 46: 1804 Use to quantify density, like CT? 5 Neonates (~ 40 wks PMA), clinical CT + research MRI CT UTE MRI BPD (Bronchopulmonary Dysplasia) BPD, R 2 = 0.81 PIG, R 2 = 0.79 PoS, R 2 = 0.23 PIG, R 2 = 0.82 TEF*, R 2 = 0.73 Pulmonary Intersititial Glycogenosis Poland Syndrome
Now we can, with MRI gating and motion correction: 1. Calculation of parenchymal densities and opacities, left & right-lung tidal volumes 2. Visualization of ventilation via ciné loops (from binned reconstructions) 3. Visualization and measurement of airway-walls (structure, malacia) 4. Measure ventilation and gas exchange via 129 Xe MRI BPD CDH post repair CDH post repair Malacia/collapse
Does pulmonary imaging matter? 1. Does imaging relate to a. disease severity? b. interventional efficacy? 2. Can we meaningfully phenotype? 3. Does imaging or image-phenotyping help predict outcomes? 4. Do we get unexpected new knowledge?
Parenchymal score stratifies clinical BPD severity Clinical severity? BPD severity (NICHD/NHLBI) Cohort size N=41 total Term control N = 5 Preterm control N = 4 Mild BPD N = 7 Moderate BPD N = 6 Severe BPD N = 20 (2 are deceased)
Conclusions Pulmonary MRI is feasible, practical (even in NICU) In BPD we can quantify prematurity-associated pulmonary abnormalities Predictive of severity and outcomes New UTE methods with self-gating Resolution approaching CT (0.7mm), signal ~ density Removal of bulk motion Respiratory gating, regional physiology Hyperpolarized gas MRI Very sensitive measure of early lung obstruction Quantification of airspace size possible even in neonates Healthy Cystic Fibrosis FEV 1 = 98% FEV 1 = 102%