Simultaneous Single Breath-hold MR Imaging of Lung Perfusion and Structure using 3D Radial UTE

Laura C Bell1, Kevin M Johnson2,Sean B Fain1, Randi Drees3, Scott K Nagle1,2,4

Departments of 1Medical Physics , 2Radiology, 3Veterinary Medicine, and 4PediatricsUniversity of Wisconsin – Madison, WI, U.S.A

Outline Introduction

Background­ Challenges in Lung Imaging­ Current Structure and Perfusion Techniques

Methods­ Dog Experiments­ MR Protocol and Post-Processing

Results/Discussion

Future Work

Introduction

Many lung diseases cause structural changes in the lungs and also impair lung function such as pulmonary perfusion or ventilation. For example,

Often, structural imaging is desired to rule-out alternative diagnoses that may not manifest perfusion abnormalities

Recently, ultra short echo times (UTE) sequences have allowed for high resolution structural visualization of the lung

Lung Disease Structural Changes Perfusion Changes

Pulmonary Emboli Lumen of the pulmonary artery

Regional decrease in perfusion

Chronic Obstructive Pulmonary Disease (COPD)

Mucus plugging of the airways

Loss of ventilation results in compensatory decrease in perfusion

Fibrosis Architectural distortions Regional decrease in perfusion

Background: Lung Imaging Challenges

MR imaging of the lungs is difficult due to:

1) Cardiac and Respiratory Motion

2) Low Proton Density

(0.15 g/ml)

3) Lung alveoli create local field inhomogenities and susceptibility gradients very short T2* (~1ms)

†Fawcett/Gehr/Science Photo Library

Rapid dephasing of inherently low signal.

Structural lung imaging has recently benefited from the development of 3D Radial UTE sequences 1-3

Highly accelerated contrast-enhanced perfusion MR imaging has allowed for sufficiently high temporal resolution (~1 sec) with good spatial resolution (3 – 4 mm3) typically acquired with spoiled gradient echo sequences

Therefore, evaluation of both perfusion and vascular structure require separate scans which:– Increases scan time and number of

breath-holds in patients who are often short of breath

– Can make it difficult to correlate the perfusion and structural abnormalities due to misregistration

Background: Current Imaging Methods

Cystic Fibrosis patient[1] Togao O et al, MRM (2010), v64, 1491 – 1498

[2] Kuethe DO et al, MRM (2007), v57, 1058 - 1064

[3] Johnson KM et al, MRM (2012), Optimized 3D ultrashort echo time pulmonary MRI

[4] Wang K et al, J Magn Reson Imaging (2013), Pulmonary Perfusion MRI using IVD sampling and HYCR

To develop and demonstrate a high resolution breath-held 3D radial UTE acquisition to simultaneously visualize lung perfusion and structure.

Purpose

Methods

Nine healthy dogs were imaged twice (separated by 2 – 4 days) resulting in 18* datasets– 8 males & 1 female– Weight 10.7 7± 1.2 kg– Mean age 13 months

Dogs were ventilated and placed in the scanner in the supine position

3T clinical scanner

20 elements of a commercial 32-channel phased array chest coil

* The first two datasets had different scan parameters, and therefore were eliminated for analysis in this study.

Methods: Imaging Protocol Pseudo-random temporally interleaved 3D radial UTE imaging

during injection of 2.3 ml of gadobenate dimeglumine followed by a 17 ml saline flush at 2 ml/sec in the cubital vein

Dynamic 3D Radial UTE scan parameters:

Optimization of the 3D Radial UTE sequence† required no hardware modifications:– Slab-selective RF excitation with limited FOV– Variable read-out gradients– Radially oversampled projections

Scan Parameter Value

Total projections: 8,000

TR/TE: 3.3/0.08 ms

Flip Angle: 15°

Readout time: 1 ms

Field of View: 24 x 24 x 24 cm3

Spatial Resolution: 0.94 mm istropic

Acquired frame rate: 1 frame/sec

Total scan time: 33 sec breath-hold

†Johnson KM et al “Optimized 3D ultrashort echo time pulmonary MRI”

Methods: Data ReconstructionReconstruction of 3D Radial UTE data:

IIterative sensitivity encoding algorithm (iSENSE) was used For structural images: composite reconstruction using all 8,000

projections For time-resolved perfusion images: temporal view-sharing

method– Filter has a width of 1 sec at the center of k-space and

quadratically increases to 7 sec at the edge of k-space

One dataset of 8,000 projectio

ns

Iterative Sensitivity Encoding Algorithm

Composite image

of structure

33 time frames each w/

242 projs

Adaptive k-space filter:Width of 1 sec of

center of k-space and quadratically increase to 7 sec at the edge of

k-space

33 time frames each w/ ~2,000 projs

Iterative Sensitivity Encoding Algorithm

33 perfusion images

Methods: 3D Radial UTE Reconstruction Workflow Fin

al R

eco

nstru

cted

Imag

es

Raw

Data

Methods: Data Analysis

Relative Tissue Enhancement measurements– (Max signal – Baseline signal) / Baseline signal– Mean signal determined from circular ROI placed in the right

lung Temporal Waveforms

– Circular ROIs centered on pulmonary artery and aorta Right Ventricle (RV) to Aorta Transit Times

– Mean signal determined from circular ROIs centered in the RV and the descending aorta

– Transit time between max signal in RV and max signal in aorta Qualitative Relative Pulmonary Blood Flow (rPBF) maps

– Calculated by indicator dilution method on a pixel-by-pixel basis

,

†Meier P et al, J Appl Physiol (1954), 6:731-744

Results: Tissue Enhancement and Temporal

Waveforms Relative lung

enhancement: 7.7 ± 1.5 compared to baseline

RV to Aorta Transit Times: 7.4 ± 2.0 sec

0 5 10 15 20 25 30

Lung

Pulm Artery

Aorta

Time [sec]

Sig

na

l [A

.U.]

Coronal MIPS show first pass of contrast bolus at 1 frame/sec with 0.94 mm isotropic spatial resolution

Results: Tissue Enhancement and Temporal

Waveforms Coronal MIPS show first pass of contrast bolus at 1 frame/sec with 0.94 mm isotropic spatial resolution

Results: Qualitative rPBF maps co-registered to structure

Discussion: Same dataset = Intrinsically co-registrated

Structural images show good depiction of airway, vessels, and lung tissue

Perfusion images show normal physiologic gravitational gradient in the A/P direction

Minimal cardiac motion due to ultrashort TE and radial sampling

Top Row: rPBF maps co-registered to structureBottom Row: Corresponding structural images

Future Work Co-registration of lung structure and perfusion in

one breath-hold in healthy dogs is feasible:– Sufficient signal is available for structural visualization and

perfusion analysis– High isotropic resolution (0.94 mm)– Acquisition is robust to cardiac and respiratory motion– To authors’ knowledge this is the first example of time-

resolved 3D UTE sequence in large animals

Future work: Application of constrained reconstruction methods (e.g.

compressed sensing or HYPR) Pulse sequence development to allow for more quantitative

hemodynamic analysis Apply this method to different lung diseases

Christopher FrancoisSara PladziewiczRebecca JohnsonGrzegorz Bauman

Funding

Acknowledgements

Check out our other UTE lung abstracts at ISMRM this year:#2005 Johnson KM, “Lung Tissue Differentiation with Magnetization Transfer Prepared Multi-Echo UTE MRI”#4538 Kruger SJ, “3D Radial Oxygen Enhanced Imaging in Normal and Asthmatic Human Subjects”

UW MadisonThis project was supported by

Clinical and Translational Science Award (CTSA) program, previously through the National Center for Research Resources (NCRR) grant 1UL1RR025011, and now by the National Center for Advancing Translational Sciences (NCATS), grants 9U54TR000021.

The UW School of Medicine and Public Health from the Wisconsin Partnership Program.