Quantifying Cardiothoracic Variation with Posture and Respiration to Inform Cardiac Device Design

Abstract

Purpose:

With extravascular implantable cardioverter defibrillator leads placed beneath the sternum, it is important to quantify heart motion relative to the rib cage with postural changes and respiration.

Methods:

MRI scans from five males and five females were collected in upright and supine postures at end inspiration [n=10 each]. Left and right decubitus [n=8 each] and prone [n=5] MRIs at end inspiration and supine MRIs at end expiration [n=5] were collected on a subset. Four cardiothoracic measurements, six cardiac measurements, and six cardiac landmarks were collected to measure changes across different postures and stages of respiration.

Results:

The relative location of the LV apex to the nearest intercostal space was significantly different between the supine and decubitus postures (average ± SD difference: −15.7 ± 11.4 mm; p < 0.05). The heart centroid to xipho-sternal junction distance was 9.7 ± 7.9 mm greater in the supine posture when compared to the upright posture (p < 0.05). Cardiac landmark motion in the lateral direction was largest due to postural movement (range: 23 – 50 mm) from the left decubitus to the right decubitus posture, and less influenced by respiration (5 – 17 mm). Caudal-cranial displacement was generally larger due to upright posture (13 – 23 mm caudal) and inspiration (7 – 20 mm cranial).

Conclusions:

This study demonstrates that the location of the heart with respect to the rib cage varies with posture and respiration. The gravitational effects of postural shifts on the heart position are roughly 2-3 times larger than the effects of normal respiration.

Introduction

Sudden cardiac death (SCD) is one of the most common causes of death, accounting for an estimated 30% mortality rate in patients with nonischemic cardiomyopathy (NICM) [1]. The advent of implantable cardioverter-defibrillators (ICDs) has proven to be a significant advancement in prevention of sudden cardiac death and extended ailments in patients with nonischemic cardiomyopathy [2]. ICD implantation is a class I indication in the context of primary prevention of sudden cardiac death in patients with heart failure and reduced left ventricular ejection fraction due to NICM [1]. Recent meta-analyses have found that ICDs significantly reduced all-cause mortality in patients with NICM [1]. Additionally, ICDs significantly reduced risk of sudden cardiac death [1]. However, long-term nonfatal outcomes after ICD implantation are not well-defined [3]. There seem to be some risks associated with ICDs, as even after the risk of death is accounted for, there remains a number of ICD-related complications that require reoperation or re-hospitalization [3]. These sequela are related to ICD lead failure, vascular obstruction, and device related infections [4]. Transvenous lead revisions and extractions to treat these complications are associated with morbidity and mortality [5]. Further, patients with venous obstructions and specific cardiac anatomies may prevent transvenous lead implantations.

The extravascular implantable cardioverter-defibrillator (EV ICD) developed by Medtronic is designed to provide the same benefits of traditional transvenous ICDs without invasive leads in the veins or the heart [6,7]. The first-in-human chronic experience of the EV ICD system reported a 90-day freedom from system/procedure related major complication rate of 94.1% and a defibrillation efficacy of 90% when testing for a ≥10 Joule (J) safety margin [6]. The generator of the EV ICD system is implanted subcutaneously at the left mid-axillary line and the lead is placed beneath the sternum in the anterior mediastinal space. Although there is room available in the anterior mediastinal space to accommodate a lead, relative stability of the lead in this space is unknown [8]. Unlike traditional transvenous ICDs, the EV ICD lead is not fixated on the heart, and the electrodes on the lead are not in the blood pool, nor contacting the myocardium. The system performance related to electrogram sensing, pacing, and defibrillation is sensitive to the relative position of the electrodes on the lead to the heart, and is crucial for safe and efficacious therapy. With this implementation, the effects of posture and respiration on relative movement of thoracic structures and the lead electrodes become increasingly important. Indeed, in the first-in-human chronic study, R-wave sensing amplitudes although relatively stable, were observed to have transient changes over time and with postures. For example, R-wave amplitude increased to 3mV in a bending posture from 1.5mV in the supine posture [9]. Additionally, in the study, pacing thresholds tended to be lowest with the patient lying on the right side and generally increased relative to supine posture when the patient was lying on the left-side, prone or bending over [10].

It has long been acknowledged that some of the difficulty in obtaining quality medical images can be attributed to the movement of abdominal and thoracic organs in respiration [11,12]. Researchers have also found that heart-induced motion during a normal free-breathing cycle changed up to 9 – 11 mm when displacement of key structures such as the Left Anterior Descending artery were measured [13]. Although displacement from resting positions of thoracic organs has been tracked in the past, there still exists a gap in knowledge about changes in distance between thoracic structures, such as the sternum and the heart.

Furthermore, the movement of thoracic structures during postural changes of the subject is important in the design of the EV ICD system and procedure. The displacement and morphology changes of abdominal organs and rib coverage area has been measured in the past using magnetic resonance imaging (MRI) [14,15]. When center of gravity movement of the abdominal organs—liver, spleen, and kidneys—was measured, it was found to vary significantly between supine and seated postures [14]. The greatest displacement of the liver when transitioning from the supine to seated posture was in the head-toe direction, and it showed a consistent inferior (mean ± SD: 19.6 ± 3.1 mm) and medial (left) (9.5 ± 3.4 mm) trajectory [14]. Rib coverage, or the projected area of the ribs onto abdominal organs, when measured in frontal, lateral, and posterior projections, also significantly varied between both postures [15]. The area of rib coverage for the liver decreased 1% in the anterior view and 2.9% in the lateral view in the seated posture [14]. The rib coverage area of the spleen increased 7.5%, and the right and left kidneys had a decrease in coverage of 9.1% and 4% when transitioning to the seated position [14]. These findings indicate that posture affects location and morphology of organs—which may have significant implications on heart morphology across postures.

The relative distances between thoracic structures can be readily measured using MRI. Unlike x-ray and computed tomography (CT) imaging, MRI of the thorax spares the subject exposure to ionizing radiation. MRI is also unique in its ability to acquire images in transverse, sagittal, and coronal planes without cumbersome patient positioning [16]. Although MRI scan times are long compared to x-ray and CT, techniques have been developed to avoid artifacts attributable to respiratory and cardiac motion—the latter of which is addressed by electrocardiogram (ECG)-gated MRI [16]. Additionally, the motion and difference in morphology of structures between posture changes can also be measured with data acquired from MRI, so long as the MRI device permits scanning in the desired postures. For example, the FONAR Upright™ MRI scanners have been used to measure changes in morphology of the lungs [14,15,17].

Previous work done with heart failure patients has elucidated that cardiothoracic morphology variation is critical to consider when designing cardiac devices [18]. For instance, the minimal distance between the sternum and heart and the angle between the left ventricle (LV) long-axis and the coronal plane are among several measures found to be greater in males when compared to females with heart failure [18]. The objective of the current study was to better understand how the heart moves relative to the rib-cage in 10 healthy male and female subjects due to postural changes (supine, standing, decubitus, prone) and respiration stage (end inspiration, end expiration). This study investigated the following hypotheses: (1) Cardiothoracic measures relating the heart location to the neighboring thoracic anatomy will differ between the supine and decubitus and upright postures and with respiration stage, (2) Movement of the heart will be largest in the lateral direction with decubitus posture changes, largest in the caudal-cranial direction with supine to upright posture change and respiration stage, and less influenced by posture or respiration in the anterior-posterior direction.

Materials and Methods

High quality ECG-gated MRI scans from five males and five females were collected as part of a prospective, non-randomized, unblinded clinical research study at Wake Forest School of Medicine. The study complied with Institutional Review Board policies approved by Wake Forest School of Medicine and all subjects provided informed consent. The 10 subjects consisted of a diverse group of individuals (Table 1). Five male and five female subjects were selected with three younger subjects (ages 22 – 29) and two older subjects (ages 57 – 66) per sex. Their BMIs ranged from 20.0 to 29.9 kg/m², and they encompassed a broad percentile range for chest circumference (11^th – 94^th) and waist circumference (45^th – 98^th). The subjects had a mean body surface area (BSA) of 2.0 ± 0.2 m², a measure used commonly in imaging studies to eliminate confounding effects on cardiac dimensions [18].

Table 1:

Study Participants and Scans Collected.

Sex	Age (years)	Height (cm)	Weight (kg)	BMI (kg/m2)	BSA (m2)	Scans Collected
						S-EI	S-EE	U-EI	R-EI	L-EI	P-EI
M	25	181.1	91.6	28.2	2.1	•		•	•	•	•
M	25	180.3	75.3	23.1	2.0	•		•	•	•	•
M	29	185.2	84.8	24.6	2.1	•		•	•	•	•
M	57	186.4	103	29.9	2.3	•	•	•	•	•
M	60	186.7	94.3	26.7	2.2	•		•
F	22	169.9	66.6	23.0	1.8	•	•	•	•	•	•
F	25	169.2	76.7	25.5	1.9	•	•	•	•	•	•
F	25	169.2	58.1	20.0	1.7	•	•	•	•	•
F	60	164.6	68.9	25.3	1.8	•	•	•	•	•
F	66	158.0	59.4	24.0	1.6	•		•
Avg	39.4	175.1	77.9	25.0	2.0	10	5	10	8	8	5
SD	18.6	10.1	15.2	2.8	0.2	Sample Sizes

BMI: body mass index, BSA: body surface area, S-EI: Supine End Inspiration, S-EE: Supine End Expiration, U-EI: Upright End Inspiration, R-EI: Right Decubitus End Inspiration, L-EI: Left Decubitus End Inspiration, P-EI: Prone End Inspiration

Healthy subjects that enrolled each had, at minimum, one ECG-gated and one non-gated 0.6T MRI scan taken in an upright/standing posture at end inspiration [n=10], and one ECG-gated 3T MRI scan taken in a supine posture at end inspiration [n=10]. Additionally, a subset of the participants had the following additional ECG-gated 3T MRI scans: left lateral decubitus posture at end inspiration [n=8]; right lateral decubitus posture at end inspiration [n=8]; prone posture at end inspiration [n=5]; and supine posture at end expiration [n=5]. No contrast was used in any scans. The ECG-gated 0.6T scans were performed at FONAR Corporation in Melville, New York using a FONAR Corp. Upright™ MRI scanner, with a 34 cm field of view, pixel size 2.66 × 2.66 mm² and slice thickness 8.0 mm. A balanced steady-state free precession (bSSFP) 2DFT gradient echo sequence was used. A non-gated 0.6T scan with a breath-hold T1-weighted a 3D GRE sequence was also collected with a larger field of view (36 cm, voxel volume 2.5 mm³ isotropic, interpolated to 1.25 mm³ isotropic). All 3T scans were performed at Wake Forest School of Medicine in Winston-Salem, North Carolina using a Siemens Skyra model, with a 36 – 40 cm field of view, pixel size 1.92 × 1.92 mm² and slice thickness 2 mm. A 3D VIBE sequence was employed, with DIXON in/out-phase images to enhance segmentation.

Several heart structures were semi-automatically segmented from the supine end inspiration [n=10] and upright end inspiration [n=10] MRI scans using Mimics software (v.19.01, Materialise, Ann Arbor MI): left and right atria, left and right ventricles, pulmonary arteries and veins, coronary arteries and sinus, superior vena cava, and myocardium. For the supine end inspiration scans, soft tissue structures and bones were semi-automatically segmented from the ECG-gated MRI scan: sternum, costal cartilage, ribs, lungs, diaphragm, and the inferior vena cava. Additionally, several soft tissue structures and bones were semi-automatically segmented from non-gated MRI scans for the upright end inspiration scans: sternum, costal cartilage, ribs, lungs, diaphragm, and the inferior vena cava. Due to the difficulty visualizing and segmenting bones in the supine and upright MRIs, a script was developed to allow the user to identify bone landmarks in the MRI scans (using Mimics). Subsequently, the spine and ribs were morphed from atlas models to fit the patients’ supine anatomy, producing a segmentation of the spine and ribs [18]. Supine bones were then rigidly transformed to fit the patient’s upright MRI scan. Soft tissues in the non-gated MRI were segmented using a combination of manual contouring and thresholding. 3D stereolithography models of the anatomical structures were created using respective optimal settings in Mimics. The ECG-gated MRI had a smaller field of view which was unable to capture the entirety of the torso. Therefore, the segmented aorta, inferior vena cava, pulmonary artery, superior vena cava, blood pools, sternum, pulmonary veins, and the myocardium from the non-gated scans were registered to the gated scan MRI scan. This process produced 3D models of torso structures in the chest MRI coordinate system. The entirety of this process was only done for the supine and upright end inspiration scans.

Measurements Overview

For the supine end inspiration and end expiration scans, four cardiothoracic and six cardiac measurements were collected from MRI images using Aquarius iNtuition (v.4.4.11 TeraRecon, Foster City, CA, USA) from the 3D segmentation models using Geomagic Studio (v.2014, Geomagic Inc., Morrisville, NC) and a combination of both MRI images and 3D models on Mimics software (v.19.01, Materialise, Ann Arbor, MI). The cardiothoracic measurements were also collected from the MRI scans acquired in different postures (upright, decubitus, and prone postures at end inspiration). Finally, the movement of six cardiac landmarks with posture and respiration changes was measured in the supine, upright, and decubitus postures.

Cardiothoracic Measures

The following four cardiothoracic measures dealing with the location of the heart in relation to neighboring structures were collected (Figure 1): (1) the shortest path from the left ventricular (LV) apex to intercostal space, (2) the angle between the LV apex and intercostal space, (3) the minimal distance between sternum and heart, and (4) the heart centroid to xipho-sternal junction distance.

Figure 1:

Four cardiothoracic measurements.

The distance between the midpoint of the nearest superior and inferior intercostal cartilage to the LV apex was defined as the shortest path from the LV apex to the respective intercostal space. The angle between this resultant vector and the anterior-posterior axis then defined the angle between the LV apex and intercostal space. The minimal distance between the sternum and heart was measured by placing a landmark on the xipho-sternal junction on the coronal view and then finding the closest path connecting the sternum to the myocardium on the sagittal view. The heart centroid of the four chambers, including the myocardium, was calculated using Geomagic Studio in order to measure the distance to the xipho-sternal junction; this measure relied on segmentations of the anatomy and thus was collected only in the supine and upright end inspiration scans.

Cardiac Measures

Six cardiac measures were collected for the supine end inspiration and end expiration scans (Figure 2): (5) angle between the mitral valve plane and the LV long-axis, (6) the angle between the LV long-axis and coronal plane, (7) the LV long-axis length, (8) the LV short-axis length, (9) the left ventricular end diastolic diameter (LVEDD) at 50% of the LV long-axis length, and (10) the LV sphericity (defined as the LV long-axis length divided by the LV-short axis length).

Figure 2:

Six cardiac measurements.

The angle between the mitral valve plane and LV long-axis and the angle between the LV long-axis and coronal plane were measured on the four-chamber view. The LV long-axis length was measured on the four-chamber view, from the LV apex to the center of the mitral valve. LVEDD at 50% of the LV long-axis length was also measured on the four-chamber view, by measuring the length of the left ventricle perpendicular to the LV long-axis at 50% of its length. The LV short-axis length was measured on the two chamber view. LV sphericity was calculated by dividing the LV long-axis length by the LV short-axis length. Chamber volumes of the left and right atria and ventricles were also computed from the supine end inspiration segmentations.

Cardiac Landmark Motion

The images were also used to digitize six anatomical landmarks: valve centroids (aortic, mitral, pulmonary, and tricuspid) and ventricular apex points (left and right endocardial apex). The 3D models were brought into 3-matic software (Materialise, Belgium) where a co-registration was conducted. All volunteer postures/respiration instances were registered to a baseline posture (supine end inspiration). The sternum and costal cartilage models were used for the co-registration and the superior portion of the lungs was used to verify correct co-registration (Figure 3). Matlab (MathWorks, Natick, MA) was used for further post-processing to calculate displacements and surface transformation matrices. Displacements are in the context of movement in the anterior-posterior axis, head-toe axis, and the left-right axis.

Figure 3:

Co-registration of anatomy from a supine end expiration (EE) scan to the baseline supine end inspiration (EI) scan.

Statistical Analysis

The means and standard deviations of the data were calculated. Data were not normally distributed per the Shapiro-Wilk’s test, and thus were analyzed with non-parametric Wilcoxon sign rank tests to compare differences between the posture and respiration states by means of R Studio (v.1.3.959, RStudio, Boston, MA). An α=0.05 was used to assess statistical significance.

Results

Cardiothoracic and Cardiac Measures

Summary statistics for the cardiothoracic and cardiac measures are provided in Table 2. Chamber volumes (average ± SD) were as follows: right atrium (54.9 ± 25.7 mL), right ventricle (103.9 ± 52.0 mL), left atrium (52.9 ± 25.3 mL), and left ventricle (111.1 ± 40.5 mL). Significant differences were observed in three cardiothoracic measures in regard to posture changes and significant differences were observed in one cardiothoracic measurement in regard to respiration state changes. The distance between the LV apex and the nearest intercostal space was significantly different between the supine and left decubitus postures (average ± SD diff. = −15.7 ± 11.4 mm; p < 0.05). The angle between the LV apex and the intercostal space (angle between the anterior-posterior axis and resultant vector ranging from the LV apex to intercostal space) was significantly different when comparing the supine posture to the left decubitus posture (+19.4 ± 14.5 degrees; p < 0.05) and the right decubitus posture (+20.3 ± 13.6 degrees; p < 0.05). Additionally, the angle between the LV apex and the intercostal space was also significantly different between the supine end inspiration and supine end expiration states (+19.8 ± 9.2 degrees; p < 0.05). The minimal distance between the sternum and the heart was not found to be significantly different between posture and respiration states, including the prone state. The heart centroid to xipho-sternal junction distance was significantly different between the supine and upright postures (+9.7 ± 7.9 mm; p < 0.05). The six cardiac measurements taken in the supine posture were not found to vary significantly between end inspiration and end expiration.

Table 2:

Average ± SD measurements and sample sizes collected.

Measurements		Supine EI (ref.)	Supine EE	Upright EI	Left Decub. EI	Right Decub. EI	Prone EI
		N=10	N=5	N=10	N=8	N=8	N=5
Cardiothoracic	1) LV Apex to Intercostal Space [mm]	22.16 ± 7.27	29.16 ± 6.69	19.28 ± 10.05	37.82 ± 14.02 *	19.68 ± 5.35	32.39 ± 10.38
	2) Angle: LV Apex/Intercostal Space [deg]	31.27 ± 13.19	12.29 ± 6.76*	33.66 ± 23.33	16.36 ± 8.12 *	15.42 ± 11.18 *	25.41 ± 15.59
	3) Minimal Distance between Sternum and Heart [mm]	17.65 ± 7.93	15.10 ± 6.33	12.63 ± 4.57	19.04 ± 5.88	14.67 ± 4.62	12.84 ± 9.74
	4) Heart centroid to xipho-sternal junction distance [mm]	61.38 ± 10.47		51.73 ± 6.68 *
Cardiac	5) Angle: Mitral Valve Plane/LV Long-Axis [deg]	92.18 ± 13.73	89.65 ± 3.39
	6) Angle: LV Long-Axis/Coronal Plane [deg]	50.80 ± 8.92	48.09 ± 4.00
	7) LV long-axis length [mm]	79.85 ± 8.30	75.27 ± 6.94
	8) LV short-axis length [mm]	44.48 ± 3.14	46.65 ± 5.60
	9) LVEDD at 50% LV long-axis length [mm]	39.96 ± 4.55	42.06 ± 4.05
	10) LV sphericity	1.80 ± 0.19	1.62 ± 0.14

^*indicates p<0.05 for measurement difference compared to the supine end inspiration posture

Abbreviations: EI (end inspiration); EE (end expiration); Decub. (decubitus), Ref. (reference)

Cardiac Landmark Motion

Cardiac landmark motion analysis was completed in 9 subjects and 5 postures and at both end inspiration and end expiration for the supine posture (Table 3 and Appendix). One subject (57-year-old male) was excluded due to poorer image quality. Six landmarks (Aortic Valve, Left Ventricular Apex, Mitral Valve, Tricuspid Valve, Pulmonic Valve, and Right Ventricular Apex) were identified, and the average landmark movement with changes in posture and respiration was measured in relation to the supine end inspiration state.

Table 3:

Landmark motion for the Aortic Valve (AV), Left Ventricular Apex (LVA), Mitral Valve (MV), and Tricuspid Valve (TCV), Pulmonic Valve, and Right Ventricular Apex (RVA) reported as mean ± SD (min, max) relative to the supine end inspiration scan.

Landmark Motion Relative to Supine End Inspiration, mm		Upright EI	Left Decub. EI	Right Decub. EI	Supine EE
		N=9	N=4	N=4	N=4
A-P Motion (−) towards sternum (+) towards spine	AV	−2.3 ± 18.9 (−19.7, 17.8)	10.8 ± 15.9 (−4.1, 32)	2.5 ± 13.0 (−11.6, 19.9)	6.9 ± 8.0 (−4.5, 13.9)
	LVA	−14.6 ± 11.4 (−30.7, 9.6)	7.9 ± 9.6 (−0.4, 17.6)	−3.2 ± 14.8 (−17.6, 16.3)	19.7 ± 14.3 (8.4, 40.5)
	MV	−16.7 ± 14.1 (−33.7, −2)	1.3 ± 4.3 (−3.3, 5.3)	−3.5 ± 0.0 (−3.5, −3.5)	11.6 ± 4.2 (7, 15.3)
	TCV	−15.3 ± 0.0 (−15.3, −15.3)	−0.1 ± 3.0 (−4.1, 2.8)	−3.9 ± 0.0 (−3.9, −3.9)	12.7 ± 5.0 (7.3, 17.2)
	PV	−10.9 ± 12.2 (−19.6, −2.3)	−0.1 ± 0.8 (−0.7, 0.5)	−6.5 ± 3.2 (−10.2, −4.5)	−1.9 ± 25.2 (−30.9, 15)
	RVA	−11.5 ± 13.7 (−25.3, 16)	8.3 ± 7.9 (0.5, 19.3)	2.4 ± 10.5 (−10.7, 14.7)	22.7 ± 11.8 (12.9, 39.8)
Lateral Motion (−) towards right (+) towards left	AV	−6.9 ± 17.0 (−26.4, 4.5)	1.6 ± 16.2 (−18.6, 15.6)	−20.9 ± 14.9 (−43, −11.5)	5.0 ± 7.5 (−4.4, 13.5)
	LVA	−4.2 ± 14.1 (−28.4, 12.6)	20.3 ± 16 (0.5, 36.1)	−29.2 ± 10.8 (−42.9, −16.7)	6.9 ± 2.2 (4.6, 9.6)
	MV	−7.1 ± 6.7 (−13.2, 2.3)	16.4 ± 25.1 (−0.6, 45.2)	−15.7 ± 0.0 (−15.7, −15.7)	17.0 ± 17.6 (1.8, 36.3)
	TCV	−11.1 ± 0.0 (−11.1, −11.1)	12.4 ± 9.7 (1.3, 24.5)	−16.9 ± 0.0 (−16.9, −16.9)	6.0 ± 6.3 (0.1, 12.6)
	PV	8.6 ± 7.1 (3.6, 13.7)	6.8 ± 11.9 (−1.6, 15.2)	−11.3 ± 5.6 (−15.6, −4.9)	−10.1 ± 24.7 (−38.7, 7.8)
	RVA	−15.7 ± 13.2 (−33.2, −0.5)	19.3 ± 13.7 (2.3, 35.9)	−26.3 ± 13.9 (−41.8, −9.6)	5.1 ± 2.9 (1.3, 7.8)
Head-Toe Motion (−) towards toe (+) towards head	AV	−20.9 ± 12.7 (−34, −8.6)	−5.9 ± 7.8 (−17.3, −0.9)	−3.1 ± 5.5 (−9.7, 3)	−0.7 ± 5.7 (−8.4, 3.8)
	LVA	−1.1 ± 10.4 (−16.7, 9.7)	2.9 ± 1.2 (1.7, 4.3)	3.7 ± 5.5 (−1.5, 11.4)	−2.5 ± 7.9 (−10, 6.9)
	MV	−9.0 ± 11.9 (−25.9, 1.9)	10.7 ± 25.4 (−9.6, 39.2)	−3.1 ± 0 (−3.1, −3.1)	10.9 ± 23.7 (−7.6, 37.6)
	TCV	−9.2 ± 0.0 (−9.2, −9.2)	−5.6 ± 4.0 (−9.6, −1)	−7.8 ± 0.0 (−7.8, −7.8)	−6.2 ± 11.9 (−19.9, 2.2)
	PV	−6.7 ± 0.9 (−7.3, −6.1)	−2.9 ± 0.3 (−3.1, −2.7)	−5.4 ± 6.8 (−13.2, −0.6)	−3.8 ± 4.7 (−7.4, 1.6)
	RVA	−1.9 ± 10.7 (−17.2, 15.7)	−1.0 ± 6.8 (−8.2, 6.8)	5.6 ± 8.0 (−5.7, 11.6)	−8.2 ± 9.5 (−16.5, 4.4)

Landmark motion results were initially noted as displacement from the supine end inspiration state. Figure 4 illustrates the average motion due to postural changes (solid diamonds) in the right and left decubitus states when compared to the baseline supine end inspiration posture (non-filled diamond). Landmark motion in the left-right lateral direction was largest due to postural movement (range: 23 – 50 mm) from the left decubitus to the right decubitus posture (Figure 4A). Respiration induced lateral motion was smaller (range: 5 – 17 mm; Figure 4B). Landmark displacement in the head-toe axis for all landmarks was generally larger due to upright posture (range: 13 – 23 mm caudal; Figure 4C) and inspiration (range: 7 – 20 mm cranial; Figure 4D). Besides the AP axis, the LV apex tends to move in all axes across all postures more than other landmarks. There was much less respiration induced movement in the antero-posterior dimension (Table 3) than head-toe and left-right.

Figure 4:

Posture and respiration effects on aortic valve, mitral valve, tricuspid value, and left ventricular (LV) apex landmark movement relative to the baseline supine end inspiration landmark positions (non-filled diamonds). (A) Average lateral motion due to posture. (B) Average lateral motion due to respiration, with posture landmark motion grayed out for comparison. (C) Average head to toe motion due to posture. (D) Average head to toe motion of due to respiration, with posture landmark motion grayed out for comparison.

Discussion

The distance between the heart centroid to the xipho-sternal junction was found to be 9.7 mm greater in the supine posture when compared to the upright posture. Research regarding posture-induced movement of the abdominal organs aligns with the findings of this study. Supine to seated postural adjustments result in 6 – 26 mm movement of various abdominal organs [14], which is on the same order of magnitude observed here for supine to upright postural movement of the heart. Additionally, the previous findings of rib coverage of abdominal organs changing significantly between postures [15] also aligns with the findings in this study that the relative location of the LV apex to the nearest intercostal space varied between the decubitus and supine postures. The relative location of the LV apex to the nearest intercostal space also varied based on respiratory stage, aligning with findings by Qi et al. 2012 of left anterior descending artery movement in the normal breathing cycle [13]. Overall, the data in this study contribute to findings that the relative positions of organs with respect to the rib cage vary due to posture and respiration.

Motion of the six cardiac landmarks tracked in our study were on the order of ± 10 mm in the head-to-toe direction, aligning closely with the 9 – 11 mm of movement of the left anterior descending artery observed by Qi et al., 2012 during a normal free-breathing cycle [13]. Overall, posture has a larger effect on heart movement than does respiration. As one might expect, decubitus postures produce the largest lateral movement and a standing posture produces the largest caudal movement of the heart, due to the influence of gravity. While postural adjustment can cause two to three times the amount of movement of the heart compared to the movement associated with normal respiration, these motion trajectories are overall less than 50 mm.

Limitations

It is important to consider the limitations of the 0.6T upright MRI scans due to their lower resolution compared to the 3T supine, decubitus, and prone scans. One can measure differences only as well as the scan resolution permits. Additionally, there is some amount of subjectivity involved in the segmentation of the structures and points of interest. However, the measurement differences and landmark motions observed with posture and respiration changes were generally above the in-plane scan resolution of both the 0.6T and 3T ECG-gated scans (2.66 and 1.92 mm, respectively). Further, the segmented models from chest scans were confirmed by two reviewers and the heart segmentations were reviewed by a medical imaging expert to reduce intra- and inter-observer variability.

Segmentations were only performed for supine and upright end inspiration scans; thus, measurement of the heart centroid to xipho-sternal junction distance was not performed for the other postures or end expiration. Cardiac measures of heart anatomy were made using images acquired in the supine posture only, as they are not expected to change with posture.

The small sample size of this study may limit generalizability of the results. Nevertheless, while likely underpowered, this study identified several significant differences in cardiac and cardiothoracic measurements across both posture and respiration states when compared to the supine end inspiration state. By doing so, this study composes the foundation of data that could be used to inform design of biomedical devices that may be implanted into the thoracic cavity and provides tangible data to fill the void in literature describing specific motion changes.

Applications

Specific motion changes in the cardiothoracic regions across a multitude of postures can inform the design of medical devices, particularly those that need to be implanted in the region. For example, the EV ICD lead is implanted in the anterior mediastinal space, and relative motion of the heart with posture and respiration may affect the therapy of the device. The results of this study can help put into context any effect posture and respiration may have on the therapy provided by the EV ICD system. Despite the small subject sample size, the detailed analysis conducted provides morphological measures derived from MRI data that delineate posture and respiration changes to the heart and the cardiothoracic region, which may be of value to clinicians and engineers who interface with medical devices.

Data on postural-based movement of cardiothoracic structures may impact clinical care more generally, as patients are typically evaluated on the basis of MRI scans acquired in a single supine posture. Insight on stereotypical movement of cardiothoracic structures may provide benefit to clinicians who note incidental findings on patient radiology. Finally, the pilot data reported in this study can be used to inform the development of similar studies in larger populations and in the development of devices that provide more robust therapy to patients.

Conclusion

This study demonstrates that the location of the heart with respect to the rib cage varies with posture and respiration. The gravitational effects of postural shifts on the position of the heart are roughly two to three times larger than the effects of normal respiration. The cardiac landmark motion measures, cardiothoracic measurements and models, and the cardiac measurements from this study could assist with surgical planning and the development of highly sophisticated implantable cardioverter-defibrillators and pacemakers to manage heart failure. To our knowledge, MRI-based measures of cardiac morphology across as many posture states as are reported here have not previously been published. These measurements may also be of value in surgical planning and in the contextual interpretation of radiological images of the heart. This particular study answers the call of previous design limitations of medical devices, which did not have data controlling for posture and respiration states when examining movement of thoracic structures.

APPENDIX

Figure A.1:

Landmark motion in X-direction (lateral motion).

Figure A.2:

Landmark motion in Y-direction (anterior-posterior direction).

Figure A.3:

Landmark motion in Z-direction (head to toe direction).