Free Breathing Magnetic Resonance Elastography of the Lungs: An in-vivo Study.

Abstract

Purpose:

Lung stiffness alters with many diseases; therefore, several magnetic resonance elastography (MRE) studies were performed earlier to investigate the stiffness of the right lung during breathhold at residual volume and total lung capacity. The aims of this study were 1) to estimate shear stiffness of the lungs using MRE under free breathing and demonstrate the measurements’ repeatability and reproducibility; 2) to compare lung stiffness under free breathing to breathhold, and as a function of age, and gender.

Methods:

25 healthy volunteers were scanned on a 1.5T MRI scanner. Spin-echo dual-density spiral (SE-Spiral) and a spin-echo EPI (SE-EPI) MRE sequences were used to measure shear stiffness of the lungs during free breathing and breathhold at mid tidal volume, respectively. Concordance correlation coefficient (CCC) and Bland-Altman analyses were performed to determine the repeatability and reproducibility of the SE-Spiral derived shear stiffness. Repeated measures analyses of variances were used to investigate differences in shear stiffness between SE-Spiral and SE-EPI, right and left lungs, males and females, and different age groups.

Results:

Free breathing MRE sequence was highly repeatable and reproducible (CCC>0.86 for both lungs). Lung stiffness was significantly lower in breathhold than in free breathing (p<0.001); which can be attributed to potential stress relaxation of lung parenchyma or breathhold inconsistencies However, there was no significant difference between different age groups (p=0.08). The left lung showed slightly higher stiffness values than the right lung (p=0.14). There is no significant difference in lung stiffness between genders.

Conclusion:

This study demonstrated the feasibility of free breathing lung MRE with excellent repeatability and reproducibility. Stiffness changes with age and during the respiratory cycle. However, gender does not influence lungs stiffness.

Introduction:

The lungs are responsible for gas exchange between blood and atmosphere. Several lung diseases are associated with the exposure to atmospheric pollutants and contaminants, leading to conditions such as interstitial lung diseases (i.e. idiopathic pulmonary fibrosis (IPF)¹, hypersensitivity pneumonitis², sarcoidosis^3,4, asbestosis⁵), or emphysema^6,7. Some lung diseases are fatal with four million people losing their lives annually⁸. Furthermore, it was reported by the World Health Organization that chronic respiratory diseases cause 7% of global deaths annually. Chronic respiratory diseases are responsible for 4% of global burden, and from the period of 2005 to 2015, deaths were projected to increase by 17%⁸. These lung diseases can alter the lungs’ anatomical and mechanical properties; early detection of changes in these properties can be of key importance for improving the prognosis^9,10.

Conventional x-ray imaging and computed tomography (CT), which use ionization radiation, are clinically used to diagnose and visualize some lung diseases. They provide anatomical and morphological information about the extent of the disease, but no quantitative measure of mechanical properties of the lung. Spirometry is used clinically to determine lung function referred to as pulmonary function test (PFT). PFT can be used to diagnose lung diseases based on mechanical properties of the lung. However, it does not provide anatomical information regarding the extent or location of the disease. Respiratory system mechanical properties such as compliance, have been extensively investigated in-vivo^11–15. The chest wall, ribcage, diaphragm, and lungs compliance were studied using different non imaging techniques such as spirometry^12,13. It was found that the chest wall compliance decreased with age, but the lungs’ compliance remained the same at functional residual volume. However, it was reported that decrease in lung compliance in younger subjects was greater than in older subjects as the lung get inflated with air¹². Furthermore, it was reported by Estenne et al.¹³ that the ribcage and diaphragm compliance significantly decreased with age.

Given that compliance is the reciprocal of stiffness, magnetic resonance elastography (MRE) technique can be used to study the mechanical properties (i.e. shear stiffness) of the lungs. MRE is a phase contrast MRI technique that has been proven to detect spatial changes in mechanical properties (i.e. shear stiffness) of different body organs^16–20. Several studies have shown the feasibility of quantifying lung’s shear stiffness by using MRE^10,21,22. Those studies were performed on the right lung only during breathhold at residual volume (RV), mid-point of tidal volume (Mid), and total lung capacity (TLC)^10,23. It was shown that MRE was capable of detecting significant differences in lung’s shear stiffness across the respiratory cycle^10,23. Furthermore, two recent studies conducted on patients with interstitial lung disease and cystic fibrosis showed that there is an increase in lung’s shear stiffness in patients compared with healthy controls^21,24. Since lung patients may experience difficulties in breathing normally at rest, breathholds during a scan can be more challenging. Therefore, the first aim of this study was to investigate the feasibility, repeatability, and reproducibility of shear stiffness measurements of both lungs during free breathing by using MRE. The second aim was to investigate the change in lungs’ shear stiffness across age, gender, anatomical location (i.e. right vs. left lung), and different respiratory dynamics (i.e. free breathing vs. breathhold).

Methods:

Sequence Design:

For free breathing MRE, a prototype spin-echo dual-density spiral (SE-Spiral)²⁵ sequence was used. The spiral trajectory was designed based on the Archimedean trajectory k = (bθ)^αe^iθ (Eq. 1) to traverse k-space²⁶; where k is the complex k-space trajectory (i.e. k = kx + iky), (bθ)^α is the rate of increase in the radial direction with respect to the rate of spiral turning i.e. (bθ)^α determines the distance radially from one leaflet arm to the other and the density of a spiral trajectory throughout k-space. Implementing a spiral readout on the scanner is slew rate limited at the beginning of the acquisition, afterwards, it is amplitude limited due to hardware limitations^26–28. As a result, sample points close to the center of k-space are inherently oversampled beyond Nyquist sampling requirements even in a uniformly sampled k-space trajectory. This inherently oversampled center gives the spiral trajectory robustness to motion, at the expense of time efficiency, in which motion artifacts appears as blurring in the image²⁸. Therefore, the spiral sequence must be effectively designed to cover the entire k-space while minimizing time, and minimizing artifacts caused by long readout duration. Long readout duration will lead to background phase accumulation that degrades the image quality and generates lower signal. The spiral design efficiency is based on selecting an appropriate θ that uses the maximum slew rate at the beginning of the trajectory and then uses the maximum gradient amplitude as the readout progresses towards the edge of k-space^26,28.

To further improve robustness to motion and minimize long readout artifacts, a variable density spiral (VDS) might be used. VDS has several variants in which the sampling density varies throughout the trajectory. For example, the center of k-space is oversampled or fully sampled while the edge of k-space is under sampled. The opposite could be performed by under sampling or fully sampling the center of k-space and oversampling the edge of k-space²⁸. In this study, a dual density spiral (DDS) approach was used²⁵. A DDS approach oversamples the center of k-space then fully samples the edge of k-space²⁵. Unlike conventional VDS, which gradually transitions in sampling density from the center to the edge of k-space, DDS transitions rapidly from the oversampled center of k-space to the fully sampled edge of k-space. In this work, DDS was used instead of uniform density spiral to enable robustness to motion generated by heart and chest movement. Additionally, DDS was used instead of VDS to reduce readout time in order to minimize accumulated phase errors and signal decay associated with long readouts. However, the rate of oversampling is inversely proportional to SNR²⁹. Therefore, in this study, 4 times oversampling the center of k-space was chosen based on a tradeoff between the highest achievable SNR and the best qualitative appearance of the image during free breathing (i.e. minimal blurring artifact).

25 healthy volunteers (16 males and 9 females; 36.4±15.1 years old) were scanned three times after obtaining written informed consent approved by our Institutional Review Board (IRB). The first two scans (i.e. repeatability) were performed without moving or resetting the volunteer. Next, the volunteer was asked to step out of the scanner and then was repositioned for a third scan (i.e. reproducibility). All imaging was performed on a 1.5 T MRI scanner (MAGNETOM Aera, Siemens Healthcare, Erlangen, Germany), equipped with the MRE option. All subjects were positioned supine and headfirst in the scanner. Since both right and left lungs were scanned, two active drivers (in-house custom made and commercial driver system (Resoundant Inc., Rochester, MN)) with two to four passive drivers were used to induce sufficient vibrations at 50Hz into both lungs. The passive drivers were placed anterior on the apex and base of each lung as shown in figure 1. The size and number of the passive drivers were carefully chosen based on the body habitus of the volunteer.

Figure 1:

Schematic of passive drivers’ setup. To induce adequate vibrations into both lungs, one of the three combinations of passive drivers were chosen based on the body habitus (i.e. two large drivers, one large driver and two small drivers, or four small drivers). One large driver and two small drivers’ setup is shown.

Image Acquisition:

MRE:

In this study, the SE-Spiral sequence was used for free breathing MRE (figure 2) and a prototype spin-echo EPI (SE-EPI) sequence for breathhold MRE at mid-point of tidal volume (figure 3). Spin-echo refocusing was chosen to eliminate the effects of extremely short T₂* of the lungs (i.e. 1–3ms at 1.5T)¹⁰. To obtain adequate MRE signal in the lungs, the shortest possible TE was chosen. Parameters for both sequences were: FOV: 450×450 mm²; TR: 1020 ms; slice thickness: 10 mm; Number of slices: 10–17 axial slices; voxel size (interpolated) 1.76×1.76×10 mm³; and 4 MRE phase offsets. For the SE-Spiral the minimum TE achieved was 6.8 ms, acquisition matrix was 128×128 interpolated to 256×256, and number of interleaves (shots) was 10. For the SE-EPI the minimum TE achieved was 15 ms, acquisition matrix was 64×64 interpolated to 256×256, and the scan was performed in a single shot (i.e. EPI factor of 32 with a PAT acceleration rate of 2).

Figure 2:

Schematic of SE-Spiral MRE sequence. The unipolar MEGs (shaded in orange) with a duration of 2.27 ms, were placed around the 180° refocusing pulse to encode the motion. They were also used as crushers to ensure minimum possible TE (i.e. 6.8 ms) and to avoid stimulated echoes. The MEGs alternate in polarity every TR, in which positive MEGs are shaded in orange and negative MEGs in dashed orange lines.

Figure 3:

Schematic of SE-EPI MRE sequence. The unipolar MEGs (shaded in blue) with a duration of 2.27 ms, were placed around the 180° refocusing pulse to encode the motion. They were also used as crushers to ensure minimum possible TE (i.e. 15 ms) and to avoid stimulated echoes. The MEGs alternate in polarity every TR, in which positive MEGs are shaded in blue and negative MEGs in dashed blue lines.

Both sequences had two unipolar motion encoding gradient (MEG) lobes around the 180° refocusing pulse to further reduce TE as shown in figures 2 and 3. Each MEG lobe has a period of 2.27 ms, and both lobes were applied to read, phase, and slice directions to encode in-plane and through plane motion in separate breathholds for SE-EPI acquisition. MEG’s polarity alternates each TR to increase phase contrast.

To avoid heart and chest motion artifacts, a dual-density spiral was used with 4 times oversampling at the center of k-space²⁵ in combination with iterative self-consistent parallel imaging reconstruction (SPIRiT)³⁰. The scan time for each MRE encoding direction was 90 seconds.

To avoid motion artifacts in the SE-EPI sequence, a single shot EPI with a breathhold of 17 seconds per motion encoding direction was used. The breathhold was performed at the mid-point of the tidal volume by adequately instructing and training the subjects before their respective scans.

Lung Density:

It is known that lung density changes during the respiratory cycle^{10,22,23,31,32}. Lung density is crucial for accurately estimating the lung’s stiffness. In this study, lung density scans were performed using a 2D multi echo GRE sequence^31,32. All the scans for whole lung were acquired at mid-point of tidal volume with a breathhold of 16 seconds. The sequence parameters were: TR: 10 ms; TEs: 0.98, 1.62, 2.26, and 2.9 ms; FOV: 450×450 mm²; number of slices: 6 axial slices; slice thickness: 10 mm; acquisition matrix: 64×64 interpolated to 256×256; flip angle: 10°; and the number of averages: 4.

Image Analysis:

Shear Stiffness:

Shear stiffness can be calculated by using the following equation $\mu =\rho V_s^2$ (Eq. 2) where μ is the shear stiffness, ρ is the lung density, V_s is the wave speed. The wave speed can be obtained from the mechanical wave frequency (f) multiplied by the wavelength of the propagating wave (λ).

Left and right lungs were delineated by an experienced user by drawing a region of interest (ROI) on each magnitude image throughout the whole lung excluding major pulmonary arteries and veins. Second, to eliminate longitudinal and reflected waves, a 4^th order Butterworth bandpass directional filter was applied to the phase difference images in 8 in-plane directions with cutoff values of 4–20 waves/FOV³³. Third, lung shear stiffness was calculated for each slice individually by using 2D direct inversion³⁴ (MRElab, Mayo Clinic, Rochester, Minnesota, USA) for each encoding direction. Fourth, the stiffness maps from all encoding directions are combined to generate a weighted stiffness map and the weighting was based on the first harmonic amplitudes. Fifth, the resultant stiffness maps were median filtered (3×3 kernel)^35,36 and 95^th percentile of the stiffness values was reported to eliminate the outliers arising from the noise in the wave data as described earlier³⁷. Finally, a mean stiffness value from each slice was obtained and then the overall mean of the whole right and left lung (i.e. the mean of all slices for each lung) was individually reported.

Lung Density:

Lung density was estimated relative to a Gadolinium-doped water phantom that was placed on the volunteers’ chest during lung density scans^31,32. The following equation S = I₀e^−TE/T2* (Eq.3) was used to obtain the initial signal (I₀) of the lung for each pixel in the ROI, where S is the mean signal of the lungs at a given TE (i.e. 0.98, 1.62, 2.26, 2.9 ms). Then the resultant initial signal of the lung (I₀) was used in the following equation $\rho =\frac{I_0}{CF\times I_{ph}}$ (Eq.4) to obtain the lung’s density, where ρ is the lungs density, I_ph is the mean signal of the Gadolinium-doped water phantom, and CF is a correction factor^10,23,31,32. The correction factor differs from a scanner to a scanner and depends on the phantom characteristics. In this study, the same phantom was used on two scanners of the same model and specifications. The first scanner correction factor was 1.736 and the correction factor for the second scanner was 1.782. After obtaining the lung’s density (ρ), shear stiffness calculation was corrected by using the estimated lung density in (Eq. 3).

Registration of lung density maps and Stiffness maps:

To report the true stiffness value of lungs, the lung density maps and stiffness maps were multiplied on a pixel-by-pixel basis. In order to minimize discrepancies in lung volume between lung stiffness and lung density scans, anatomical intensity based image registration was performed before the multiplication. In the image registration step, lung density images were geometrically transformed (i.e. translational, rotational, and scale transformations) to match pixel locations in lung stiffness images. A 50 histogram bins Mattes Mutual Information metric was used involving all pixels in the overlap region. Image registration was performed by using MATLAB, Math Works, Natick, MA) in which an ROI that encapsulates both lung and the heart was drawn on the lung density image. Then, similar ROI (i.e. encapsulating both lungs and the heart) was drawn on the MRE image. Given the high contrast between the lungs and chest wall, the intensity based registration used chest wall and heart to accurately register lung density images to MRE images.

Statistical Analysis:

To determine the repeatability and reproducibility of stiffness measurements using the free-breathing SE-Spiral sequence, concordance correlation³⁸ and Bland-Altman^39,40 analyses were performed. Two-way repeated measures analysis of variance (ANOVA) was conducted to examine the effect of lung side (right or left) and breath state (breathhold or free breathing) on shear stiffness with adjustments using Tukey’s post hoc method for the pairwise comparisons between lung side and breath state.

All subjects were divided in to age groups: younger (20–40 years) and older (>40 years) as in previous studies^15,41. Furthermore, two-way repeated measures ANOVA were implemented to explore the impact of lung side and age group (younger or older), gender, and lung region (top = apex of the lung, middle, and bottom = base of the lung region) on the shear stiffness during free breathing technique. Additional repeated measures ANOVA with adjustment for Tukey’s post hoc method between lung side and different age groups as well as between lung side and gender for lung density during breathhold technique. Tukey’s post hoc adjusts for the familywise error rate and reduces the likelihood of a Type I error. This analysis used SAS PROC MIXED and was conducted using SAS v9.4 (SAS Institute; Cary, NC; www.sas.com). The level of statistical significance was set at p<0.05.

Results:

Good discernible waves were observed using the SSE-Spiral and SE-EPI sequences as shown in Figure 4. Figure 4(a, g) shows magnitude images with no artifacts using SE-Spiral and SE-EPI sequences, respectively. Figure 4(b–e, h–k) shows 4 snapshots of wave images for one of the in-plane directions (i.e. x direction) with excellent wave propagation through right and left lungs. Figure 4(f, l) shows stiffness maps for right and left lungs during free breathing (SE-Spiral sequence) and breathhold (SE-EPI sequence), respectively.

Figure 4:

Top row (a-f) was acquired during free breathing by the SE-Spiral sequence. Bottom row (g-l) was acquired during a breathhold by the SE-EPI sequence. (a, g) shows magnitude image for the same volunteer with no artifacts in the right or left lungs. (b-e, h-k) shows snapshots of four phase offsets of the wave propagating through the right and left lungs. (f) shows the stiffness maps for both right and left lungs during free breathing with mean stiffness values of 1.03±0.37 kPa and 1.09±0.44 kPa, respectively. (l) shows the stiffness maps for both right and left lungs during a breathhold with mean stiffness values of 0.74±0.21 kPa and 0.73±0.21 kPa, respectively.

Free Breathing Repeatability and Reproducibility Measurements:

Stiffness measurements obtained using free breathing SE-Spiral sequence were highly repeatable with a concordance correlation coefficient (CCC) of 0.99 (95% confidence interval (CI): 0.97 – 0.99) for the right lung and 0.99 (95% CI: 0.98 – 0.99) for the left lung. In addition, figure 5 shows Bland-Altman plots with near-zero mean bias (i.e. 0.011 kPa for the right lung and 0.005 kPa for the left lung) with narrow limits of agreement (i.e. 2 standard deviation from mean) of −0.15 – 0.17 kPa and −0.11 – 0.12 kPa, respectively.

Figure 5:

Bland Altman’s plots of right (a) and left (b) lung’s shear stiffness repeatability measurements during free breathing. The data shows highly repeatable measurements with mean bias ~ zero (0.011 kPa for the right lung and 0.005 kPa for the left lung) and narrow limits of agreement (−0.15 – 0.17 kPa for the right lung and −0.11 – 0.12 kPa for the left lung).

SE-Spiral MRE sequence demonstrated excellent reproducibility of stiffness measurements across all the volunteers with a CCC of 0.88 (95% CI: 0.76 – 0.94) for the right lung and 0.86 (95% CI: 0.72 – 0.94) for the left lung. In addition, figure 6 shows Bland Altman’s plots with ~ zero mean bias (i.e. −0.07 kPa for the right lung and −0.06 kPa for the left lung) with narrow limits of agreement (i.e. 2 standard deviation from mean) of −0.56 – 0.42 kPa and −0.52 – 0.40 kPa, respectively.

Figure 6:

Bland Altman’s plots of right (a) and left (b) lung’s shear stiffness reproducibility measurements during free breathing. The data shows highly reproducible measurements with mean bias ~ zero (−0.07 kPa for the right lung and −0.06 kPa for the left lung) and narrow limits of agreement (−0.56 – 0.42 kPa for the right lung and −0.52 – 0.40 kPa for the left lung).

Free Breathing (SE-Spiral) vs Breathhold (SE-EPI) Shear Stiffness Measurements:

A significant difference (adjusted p<0.001 for both lungs) was present in mean shear stiffness measurements between free breathing (SE-Spiral) and breathhold (SE-EPI) sequences. As shown in table 1, the mean shear stiffness for the right lung during free breathing and breathhold were 1.55±0.50 kPa and 0.89±0.21 kPa, respectively. Similarly, the mean shear stiffness for the left lung during free breathing and breathhold were 1.69±0.45 kPa and 0.94±0.22 kPa, respectively. The difference between right lung’s mean shear stiffness during free breathing and breathhold was −0.65 kPa (95% CI: −0.81 kPa – −0.49 kPa). And the difference between left lung’s mean shear stiffness during free breathing and breathhold was −0.75 kPa (95% CI: −0.88 kPa – −0.62 kPa).

Table 1:

Mean shear stiffness values of the right and left lungs for all volunteers during free breathing across different age groups and gender.

Mean shear stiffness values in kPa
	Right Lung	Left Lung
Free Breathing	1.55±0.50	1.69±0.45
Breathhold	0.89±0.21	0.94±0.22
Age group 1 (20–40)	1.70±0.49	1.83±0.45
Age group 2 (40+)	1.23±0.37	1.37±0.23
Males	1.49±0.46	1.69±0.47
Females	1.64±0.59	1.67±0.44

Right vs Left Lung shear Stiffness Measurements:

As shown in table 1, the overall mean shear stiffness of the right and left lungs during free breathing was 1.55±0.50 kPa and 1.69±0.45 kPa, respectively.

There was a non-significant difference of −0.14 kPa (unadjusted p=0.032, adjusted p=0.14) between the right and left lung shear stiffness measurements during free breathing.

The overall mean shear stiffness of the right and left lungs during a breathhold was 0.89±0.21 kPa and 0.94±0.22 kPa, respectively. Also, there was a non-significant difference of −0.04 kPa (unadjusted p=0.48, adjusted p=0.90) between the right and left lung shear stiffness measurements during a breathhold.

Right and Left Lung shear Stiffness vs Age During Free Breathing (SE-Spiral) Sequence:

Figure 7 shows a decrease in lungs stiffness with age. The mean shear stiffness of the right lung in younger age group (i.e. 20–40 years old, n=17) and older age group (i.e. >40 years old, n=8) were 1.70±0.49 kPa and 1.23±0.37 kPa, respectively. The mean shear stiffness of the left lung for younger and older age groups were 1.83±0.45 kPa and 1.37±0.23 kPa during free breathing, respectively.

Figure 7:

Box and scatter plots of right (a, c) and left (b, d) lung’s shear stiffness measurements during free breathing across different age groups. The mean shear stiffness of the lung during free breathing at older age was not significantly different compared to younger age (adjusted p=0.08 for the right lung and adjusted p=0.08 for the left lung). The mean shear stiffness of the right lung in younger and older age groups were 1.70±0.49 kPa and 1.23±0.37 kPa during free breathing, respectively. The mean shear stiffness of the left lung for younger and older age groups were 1.83±0.45 kPa and 1.37±0.23 kPa during free breathing, respectively.

The difference in shear stiffness measurements between different age groups was significant prior to adjustment but became non-significant with Tukey’s post hoc adjustment method. The difference in mean shear stiffness for right lung between younger and older age groups was 0.47 kPa (unadjusted p=0.016, adjusted p=0.08). Similarly, the difference in mean shear stiffness for left lung between younger and older age groups was 0.46 kPa (unadjusted p=0.018, adjusted p=0.08).

Right and Left Lung shear Stiffness in Males vs Females During Free Breathing (SE-Spiral) sequence:

Overall, mean shear stiffness measurements of the right lung in males and females were 1.49±0.46 kPa and 1.64±0.59 kPa, respectively. There was no significant difference in stiffness measurements between males and females during free breathing for both lungs, which can be potentially attributed to limited sample size. The difference in mean shear stiffness of right lung between males and females was −0.15 kPa (adjusted p=0.88). Overall, mean shear stiffness of the left lung in males and females were 1.69±0.47 kPa and 1.67±0.44 kPa, respectively. The difference in mean shear stiffness of left lung between males and females was 0.02 kPa (adjusted p>0.99).

Right and Left Lung shear Stiffness vs Lung Region During Free Breathing (SE-Spiral) Sequence:

Each lung was divided into three regions as following; top, middle, and bottom with a significant overall interaction between the lung and different regions (p=0.015). As shown in figure 8, the overall mean shear stiffness for the top region was 1.55±0.62 kPa and 1.60±0.48 kPa for the right and left lung, respectively. The overall mean shear stiffness for the middle region was 1.56±0.58 kPa and 1.58±0.43 kPa for the right and left lung, respectively. The overall mean shear stiffness for the bottom region was 1.54±0.51 kPa and 1.87±0.58 kPa for the right and left lung, respectively. There was no significant difference between right and left lung top (mean shear stiffness difference=0.05 kPa, adjusted p=0.99) and middle (mean shear stiffness difference=0.02 kPa, adjusted p=1.00) regions. However, there was a significant difference between right and left lung bottom region (mean shear stiffness difference=0.33 kPa, adjusted p=0.003).

Figure 8:

Box plots of different regions of the right (a) and left (b) lung’s shear stiffness measurements during free breathing. The overall mean shear stiffness for the top region was 1.55±0.62 kPa and 1.60±0.48 kPa for the right and left lung, respectively. The overall mean shear stiffness for the middle region was 1.56±0.58 kPa and 1.58±0.43 kPa for the right and left lung, respectively. The overall mean shear stiffness for the bottom region was 1.54±0.51 kPa and 1.87±0.58 kPa for the right and left lung, respectively.

For the right lung there was no significant interaction between different regions (top vs middle, top vs bottom, and middle vs bottom, adjusted p=1.00). Similarly, for the left lung the interaction between top and middle region was not significant (adjusted p=1.00). However, both interactions with the bottom region within the left lung resulted in significant difference. The interaction between bottom and middle region had an adjusted p=0.005 and the interaction between the bottom and top region resulted in an adjusted p=0.020.

Lung Density During Breathhold (SE-EPI) sequence:

Given that the lung density changes during breathing, lung density was acquired during a breathhold at mid-point of tidal volume. The estimated mean lung density for right and left lungs for all the volunteers (n=25) was 0.19±0.03 g/cm³ and 0.21±0.03 g/cm³, respectively.

The mean lung density for younger and older age groups for right lung was 0.21±0.02 g/cm³ and 0.16±0.02 g/cm³, respectively. The mean lung density for younger and older age groups for the left lung was 0.23±0.03 g/cm³ and 0.18±0.02 g/cm³, respectively. There was a significant difference of 0.05 g/cm³ in right lung mean density between both age groups (adjusted p<0.001). Similarly, there was a significant difference of 0.05 g/cm³ in left lung mean density between both age groups (adjusted p=0.002).

The mean lung density for males and females for right lung was 0.193±0.03 g/cm³ and 0.186±0.03 g/cm³, respectively. The mean lung density for males and females for the left lung was 0.2110±0.04 g/cm³ and 0.2085±0.03 g/cm³, respectively. The non-significant difference in right lung mean density between males and females was 0.007 g/cm³ (adjusted p>0.96). Similarly, the difference in left lung mean density between males and females was 0.003 g/cm³ and non-significant (adjusted p>0.99).

Discussion:

This study has demonstrated the feasibility of performing MRE during free breathing in both lungs covering the entire lung volume. It was shown that the shear stiffness measurements obtained during free breathing using the SE-Spiral sequence were highly repeatable and reproducible across all the volunteers. Initially, it was assumed in this study that the resultant shear stiffness during free breathing was a representation of average stiffness across the tidal volume. Therefore, in this study, the resultant stiffness map during free breathing was considered as the stiffness of the lung at mid-point of tidal volume. In other words, in this study, it was assumed that the stiffness map values during free breathing will be similar to the stiffness map values during a breathhold at mid-point of tidal volume. However, it was shown that the shear stiffness during free breathing was significantly higher than during a breathhold; which can be attributed to potential stress relaxation of lung parenchyma or breathhold inconsistencies. Furthermore, this study has showed slightly higher stiffness in the left lung compared to the right lung during free breathing. However, the results were not statistically significant after adjustment. It was also shown that the shear stiffness of the lung decreased with age which might be attributed to decrease in chest wall compliance (i.e., reciprocal of stiffness) during free breathing, though these results were not statistically significant after adjustment. Finally, it was shown that there is no significant difference in lung stiffness between males and females.

This study has demonstrated higher stiffness of the lungs during free breathing than during a breathhold and we believe that this might be attributed to stress relaxation. In a study performed by Faffe et al.⁴², it was reported that there was a pressure decay in the lung during breathhold due to stress relaxation. Stress relaxation happens in viscoelastic tissues such as lungs where the stress progressively decreases during constant deformation. Due to non-linear nature of stress-strain curve in viscoelastic materials, decreased stress causes stiffness to decrease. Therefore, during breathhold MRE scan, the stress can be potentially decreased due to ceased airflow leading to a lower stiffness estimate compared to a free breathing MRE scan. The significant difference between lung stiffness during free breathing and breathhold might be attributed to discrepancies in subject breathholds. In this study, subjects were initially trained and instructed to hold the breath at mid-point of the tidal volume for SE-EPI sequence. However, there was no external apparatus to monitor/confirm the breathhold at mid-point of tidal volume; and therefore, there could potentially be a possibility that subjects might be holding the breath at RV because of our instruction procedure. In addition, recent studies have reported that the diaphragm and chest muscles have complex line of motion, in which their motion is not linear and the diaphragm contributes to the respiratory dynamics more than the chest wall muscles^43–45. Also, it was reported by Kondo et al.⁴³ that the diaphragm movement is not sinusoidal during shallow breathing and forced maximal breathing, which may result in discrepancies in coinciding a given breathhold state to the same state during free breathing leading to stiffness variation between breathhold measurement versus free breathing. Therefore, we believe that comparing lung’s shear stiffness during free breathing and a given breathhold state needs further investigation.

The difference in stiffness between right and left lung as a whole and their bottom regions might be attributed to the difference in their anatomical structure (i.e. size and shape) and location. Left lung is smaller in total volume⁴⁶, narrower and longer than the right lung, and it contains the cardiac notch. Additionally, the stomach (i.e. hollow organ) is located under the diaphragm on the left side of the body while the liver (i.e. solid organ) is located under the diaphragm on the right side of the body. And in a study performed by Palthow et al. it was reported that during breathing, the region of the diaphragm under the left lung travels longer distance in comparison to the region under the right lung⁴⁷. As a result, the left lung experiences greater pressure forces exerted by the diaphragm during breathing. In summary, due to the difference in geometry and boundary conditions between right and left lungs, we anticipate that the stiffness of the left lung might be slightly higher than the right lung during breathing and breathhold.

The decrease in lung stiffness with age might be attributed to the decrease in the respiratory system compliance as reported by Sharma et al¹⁵. In older population, the chest wall, ribcage, and diaphram^12,13 become less compliant resulting in less air inhaled into the lungs, indirectly causing the lung to become softer. However, Mittman et al.¹² reported that lungs compliance doesn’t change with age. Nevertheless, the chest wall decreases in compliance with age due to costal ribs cartilage calcification¹³. Moreover, this study has showed that lungs density tends to decrease with age. This decrease in lungs density might be attributed to the degeneration of elastic fibers in the alveolar ducts¹⁵.

In previous studies conducted by Mariappan et al. and Fakhouri et al.^10,22,23 only the right lung was scanned to avoid artifacts caused by the heart motion in the left lung. However, in this study both right and left lungs were scanned by using SE-Spiral and SE-EPI sequences. Heart motion artifacts covering left lung were minimized in SE-Spiral sequence by four times over sampling the center of k-space relative to the outer k-space. Oversampling the center of k-space resulted in increased robustness to motion (i.e., heart and breathing motion); hence, the dependency on the number of respiratory cycles was not observed. After reconstructing the data, the data has a very minimal spiral artifacts in the background (figure 4). Also, significant artifacts caused by the heart motion were not observed. When using the SE-EPI sequence, artifacts in the left lung were not visible by performing the scan in a single shot and by selecting the phase encoding direction anterior to posterior direction. Another advantage of using the SE-EPI sequence with single shot is that it significantly reduces ghosting caused by shifts in center of k-space from one shot to the other and also any variation in motion from multiple shots.

Due to the presence of air/tissue interface in the lung, susceptibility artifacts play a major role in degrading MR image quality of the lungs. To minimize the effects of susceptibility artifacts, all scans were conducted in a 1.5 T scanner by using SE-based sequences. To increase the signal in the lungs, minimum possible TEs were chosen. This was executed by using fast excitation and refocusing RF pulses, which have a duration of 640μs. Additionally, fractional motion encoding was used in which the MEGs duration was approximately one fifth of the period of mechanical excitation. In addition, the 1:1 bipolar MEG was split into two unipolar MEGs placed around the refocusing pulse. Finally, the additional time taken by the crushers was compensated by using MEG’s and therefore, no crushers were included in the sequences. All the above strategies aided in minimizing TE.

In this study, different driver size and quantity (i.e. 2 to 4 drivers) were used based on the body habitus of the subject. These different setups were used to make sure that there are sufficient waves propagating through the lungs. This study has shown that changing the driver size and quantity has no effect on the shear stiffness measurements while there are sufficient waves propagating in the lungs. This can be clearly observed by the highly repeatable and reproducible shear stiffness measurements for the right and left lungs across different subjects.

Conclusion:

This study has shown the feasibility of quantifying the lungs shear stiffness under free breathing without inhalation of any contrast agents. SE-Spiral sequence has demonstrated an excellent repeatability and reproducibility of shear stiffness measurements for both right and left lungs during free breathing. In addition, this study has showed that the left lung is stiffer than the right lung. Additionally, it was shown that lung shear stiffness decreased with increase in age. However, lungs shear stiffness was not affected by gender.