pii: jc- 00415 -15http://dx.doi.org/10.5664/jcsm.6444
S CI E NT IF IC IN VES TIGATIONS
Dynamic Volume Computed Tomography Imaging of the Upper Airway in Obstructive Sleep Apnea Robert J. Fleck, MD1; Stacey L. Ishman, MD, MPH2,3,4; Sally R. Shott, MD2,4; Ephraim J. Gutmark, PhD, DSc 4,5; Keith B. McConnell, MS3; Mohamed Mahmoud, MD1,6; Goutham Mylavarapu, PhD3; Dhananjay R. Subramaniam, MS5; Rhonda Szczesniak, PhD3,7; Raouf S. Amin, MD3,8 Division of Radiology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 2Division of Pediatric Otolaryngology - Head and Neck Surgery, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 3Division of Pulmonary Medicine, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 4Department of Otolaryngology - Head and Neck Surgery, University of Cincinnati School of Medicine, Cincinnati, OH; 5Department of Aerospace Engineering and Engineering Mechanics, CEAS, University of Cincinnati, Cincinnati, OH; 6Department of Anesthesia, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 7Division of Biostatistics and Epidemiology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 8Department of Pediatrics, University of Cincinnati School of Medicine, Cincinnati, OH
1
Study Objectives: To describe a dynamic three-dimensional (3D) computed tomography (CT) technique for the upper airway and compare the required radiation dose to that used for common clinical studies of a similar anatomical area, such as for subjects undergoing routine clinical facial CT. Methods: Dynamic upper-airway CT was performed on eight subjects with persistent obstructive sleep apnea, four of whom were undergoing magnetic resonance imaging and an additional four subjects who had a contraindication to magnetic resonance imaging. This Health Insurance Portability and Accountability Act-compliant study was approved by our institutional review board, and informed consent was obtained. The control subjects (n = 41) for comparison of radiation dose were obtained from a retrospective review of the clinical picture-archiving computer system to identify 10 age-matched patients per age-based control group undergoing facial CT. Results: Dynamic 3D CT can be performed with an effective radiation dose of less than 0.38 mSv, a dose that is less than or comparable to that used for clinical facial CT. The resulting data- set is a uniquely complete, dynamic 3D volume of the upper airway through a full respiratory cycle that can be processed for clinical and modeling analyses. Conclusions: A dynamic 3D CT technique of the upper airway is described that can be performed with a clinically reasonable radiation dose and sets a benchmark for future use. Keywords: Down syndrome, four-dimensional computed tomography, obstructive volumetric computed tomography, radiation dosage, sleep apnea, sleepdisordered breathing Citation: Fleck RJ, Ishman SL, Shott SR, Gutmark EJ, McConnell KB, Mahmoud M, Mylavarapu G, Subramaniam DR, Szczesniak R, Amin RS. Dynamic volume computed tomography imaging of the upper airway in obstructive sleep apnea. J Clin Sleep Med. 2017;13(2):189–196.
I N T RO D U C T I O N
BRIEF SUMMARY
Current Knowledge/Study Rationale: Dynamic 3D CT produces a unique 3D dynamic dataset for evaluation of the upper airway for clinical or research use. Dynamic 3D CT can be performed with a radiation dose less than that of a static examination of the exact same anatomy imaged for a facial CT for common clinical indications. Study Impact: Dynamic 3D CT is an alternative method capable of obtaining a uniquely dynamic 3D dataset for the evaluation of obstructive sleep apnea in patients with persistent, problematic sleep apnea and could be used to guide surgical treatment to result in better outcomes. The effective radiation dose for dynamic 3D CT is equal to or less than that of a facial CT, setting a benchmark for use of the technique.
Obstructive sleep apnea (OSA) is a common disorder that manifests as repeated upper airway (UA) obstruction during sleep. Increasing prevalence of OSA in both pediatric and adult populations has been associated with increasing obesity and metabolic syndrome in the overall population.1,2 It is estimated that up to 5% of adults and 3% of all children in the United States are affected by OSA.1,3 In children, OSA is associated with excessive daytime sleepiness, hyperactivity, attention deficit disorder, failure to thrive, cardiometabolic problems, and neurocognitive deficits.4 As part of the evaluation of OSA, imaging of the UA in patients with OSA has traditionally been performed with magnetic resonance imaging (MRI), computed tomography (CT), and/or fluoroscopy during natural sleep, drug-induced sleep, or wakefulness to evaluate airway size, configuration, and dynamics. Commercially available MRI scanners produce static images that are used to identify UA soft-tissue targets amenable to surgical modification, such as enlarged adenoid or lingual tonsils, macroglossia, elongated soft palate, or enlarged turbinates.5 Additionally, current MRI scanners can provide detailed
dynamic single-slice images of the airway during respirations that can be used to create a cine image of airway showing movement dynamics and pattern of collapse.6,7 Overall, MRI provides detailed information,6,8–10 but is limited to only single-slice cines. Although three-dimensional (3D) cine images obtained with compressed sensing techniques11,12 or multiple two-dimensional planes obtained simultaneously13 are available at advanced research sites, they are not yet available clinically. 189
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Table 1—Polysomnography data for study subjects. Subject Number 1 2 3 4 5 6 7 8
Type of Study P P P P C C C C
AHI 27.0 4.4 5.3 4.1 10.8 9.8 11.7 17.5
Lowest O2 Saturations (%) 78.0 84.0 90.1 92.2 85.0 92.6 88.0 79.9
% Time CO2 > 50% NR 0.7 0.0 8.4 0.4 0.0 NR 5.8
% REM Sleep 0.0 20.0 22.7 17.6 20.6 28.7 25.0 0.0
Sleep Efficiency 39.0 76.0 66.4 78.1 77.4 77.4 89.1 37.5
Cardiac Disease Yes No Yes Yes Yes No No Yes
AHI = apnea-hypopnea index, C = clinical, NR = not recorded, P = prospective, REM = rapid eye movement.
CT has the same anatomical coverage as dynamic 3D CT and second, facial CT is often used for planning craniofacial surgery, which is sometimes a consideration in patients with OSA. The required information for calculating radiation dose and area of coverage were recorded. The control group was divided into subgroups based on age, and effective dose was calculated for comparison to individual age-matched study subjects.
CT of the UA and cine CT have also been described previously.8,14,15 However, with the advent of slip-ring and multidetector CT technology, cine CT covering a wider anatomic area is now possible.16,17 Wide-detector scanners, with 256–320 detector rows, are sufficiently wide to image the UA from the nasal passages to the glottis during each rotation with high temporal resolution to stop motion and create a dynamic 3D dataset. For the current study, we modified and applied this method to the UA patients undergoing an imaging study during a sleep study for OSA aimed at creating a realistic model of UA dynamics. The purpose of this study was to describe a dynamic 3D CT technique for the UA, set a benchmark for radiation exposure requirements for this technique, and compare the radiation dose to that used for common clinical studies of an identical anatomical area (e.g., clinical facial CT).
Computed Tomography
All study subjects (both prospective and clinical) were anesthetized with dexmedetomidine and permitted to breathe spontaneously via the native airway using a protocol described previously.18 Depth of sedation was assessed using the University of Michigan Sedation Scale.19 After MRI was completed, the prospective subjects were transferred to the CT scanner and positioned with their head and neck in neutral position while anesthesia and adequate oxygenation were maintained. Patients with a contraindication to MRI (the clinical subjects) were transferred directly from induction to the CT scanner. All imaging was performed on a 320-slice volume CT scanner (Aquilion ONE, Toshiba, Otawara, Japan) and reconstructed using AIDR 3D with a FC18 reconstruction kernel.20 After scout views of the head and neck were obtained, the scan length was set to extend from the nasal air passages to the subglottic region, thus covering the entire nasal air passages, nasopharyngeal airway (NPA), retroglossal airway (RGA), and hypopharynx. The number of gantry rotations was determined by the respiratory interval in milliseconds (ms) divided by the gantry rotation speed (350msec/rotation) such that the period of imaging covered one respiratory cycle. Because the respiratory cycle was relatively stable, the scan duration was calculated by dividing the minute respiratory rate into 60 sec. Therefore, if the respiratory rate was 24 breaths per minute, scanning was performed for 2.5 sec. The maximum z-axis coverage was 16 cm and could be decreased in 2-cm increments to fit each patient’s requirements. Low-dose principles based on noise tolerance by radiologists were used when determining the dynamic 3D CT parameters.21 The tube voltage was determined as either 80-kVp or 100-kVp depending on each patient’s age and size (cranial diameter). The cranial diameter was measured from the sagittal scout views as the distance from the soft tissues over the frontal bone to the soft tissues of the occiput. The tube current was based on the
METHODS
Subjects
This evaluation of the UA in OSA was approved by our institutional review board; informed consent was obtained for four subjects for prospective study and four additional subjects were retrospectively obtained from the clinical picture-archiving computer system. We performed a Health Information Portability and Accountability Act-compliant prospective study of dynamic 3D CT of the UA in four children with Down syndrome with OSA [subset of participants in the DYMOSA study (R01HL105206-01)] who were under drug-induced sleep for MRI evaluation between February 2013 and June 2013 (prospective subjects group). The four retrospective, clinical patients all had a contraindication to MRI (clinical subjects group), but did have dynamic 3D CT only between July 2012 and June 2013. Prior to imaging, all study subjects had undergone adenotonsillectomy and postoperative polysomnography to document persistent OSA (Table 1). The study subjects did not have concomitant pulmonary disease. The control subjects (n = 41) were obtained from a retrospective review of the picture-archiving computer system to identify age-matched patients undergoing facial CT between August 2012 to June 2013 for consideration of cranial facial surgery and trauma. The reason for using facial CT is twofold; first, facial Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
190
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Table 2—Comparison of study subjects with age- and size-matched control subjects. Subject Number 1 2 3 4 5 6 7 8
Type of Study P P P P C C C C
Reference Control Group Adult 5 y old 5 y old 10 y old Adult 5 y old Adult 5 y old
kVp 100 80 100 100 100 80 100 100
mA 60 40 20 30 60 80 30 30
CTDIvol 12.35 2.3 b 2.9 b 6.22 b 12.43 4.2 b 5.1 b 4.9 b
DLP 197.6 23.1 b 46.4 b 99.5 b 198.9 66.8 a 81.6 b 77.8 a
Effective Dose (mSv) 0.35568 0.08085 b 0.1624 b 0.26865 b 0.37791 0.2338 a 0.15504 b 0.2723 a
Relation to Control Doses Equivalent Lower Lower Lower Equivalent Lower Lower Lower
Falls outside of the lower and upper quartiles for controls. b Falls outside of the 95% confidence interval for controls. C = clinical, CTDIvol = volume computed tomography dose index, DLP = dose length product, kVp = kilovoltage, mA = milliampere, mSv = milliSievert, P = prospective.
a
frontal occipital diameter obtained from the scout views and determined by the formula:
where Dmax is the maximum dimension and Dmin is the minimum dimension. FC is expressed as the degree of collapse in the Tx, AP, or cross-sectional area values. Collapse is defined as FC greater than 0.80, because it would greatly limit airflow from a fluid mechanics perspective, and dynamic motion is defined as FC greater than 0.50 based on observations in adults that the airway in patients with OSA is more dynamic than patients without OSA.8,24 Stable is defined as FC less than 0.50, which is an unproven, arbitrary limit.
mA = [(Frontal occipital diameter in cm) × 3] The effective dose for all studies was calculated using the method of Deak and colleagues.22 For comparison of radiation doses, at least 10 age- and size-matched facial CT images were used as controls for each age group.
Image Reconstruction
Statistical Analysis
The images were reconstructed from half-scans, which results in a number of images that are two times the number of gantry rotations. A half-scan is a reconstruction using 180° of the gantry rotation data. These dynamic temporal phases were reviewed at a workstation (Vital Images, Inc., Minnetonka, MN, USA) that allowed for simultaneous viewing of multiplanar and 3D reformats in static phases and as cine loops in multiple planes. The anterior-posterior (AP), transverse (Tx), and crosssectional areas were measured at the narrowest section of the NPA and the RGA, above the epiglottis. Dynamic motion was defined by a greater than 50% reduction in the airway crosssectional area, and collapse was defined by a greater than 80% reduction in cross-sectional area.23 The images also were assessed independently for quality on a four-point Likert-type scale (0-non diagnostic, 1-adequate, 2-good, and 3-excellent image quality) by two radiologists with 10 and 15 y of experience. Motion blurring was assessed on individual dynamics as either present or not present and expressed as a percentage of dynamic temporal phases in the CT study.
The radiation doses of the control groups were calculated in relation to an eight-number summary for comparison to the individual doses calculated for the study subjects. Values outside of the lower and upper quartiles or the 95% confidence intervals for the controls were reported. Demographic differences between the study subjects and the respective control group for each were examined by the Wilcoxon Mann-Whitney test, and Fisher exact test for continuous and categorical variables, respectively. Values of p < 0.05 were used to determine significance throughout the analysis. For each subject, the actual airway dimensions (AP, Tx, and cross-sectional area) for both the RGA and NPA were plotted over time. A smoothing curve with a cubic B-spline basis function25 was applied to each measure to obtain an overall indication of the longitudinal course of the airway opening, closing, and reopening during the respiratory cycle. R ES U LT S
Airway Analysis
The NPA and RGA were measured transversely through the narrowest section when fully distended, and the same slice was measured throughout the respiratory cycle. The UA is a dynamic structure.8 To express this dynamic behavior, we introduce the term fractional collapse (FC), which yields a dimensionless fraction from 0 to 1.0 determined by the formula:
The first clinical study (Subject 5) was performed prior to the prospective group, and the additional clinical studies (Subjects 6, 7, and 8) were performed after the prospective patients were enrolled and imaged. The radiation doses for Subjects 1 (first prospective) and 5 (first clinical) were substantially more than the doses for the remaining six subjects (Table 2). This was due to subsequent modification of the dynamic 3D CT parameters to minimize the dose while maintaining adequate quality.
FC = (Dmax – Dmin ) / Dmax
191
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Figure 1—Volume-rendered surface display of air-filled structures and face in Subject 3, a 3-y-old boy with obstructive sleep apnea and Down syndrome.
AB
(A) Fully distended airway during expiration; retroglossal airway widely patent (arrowhead), nasopharyngeal airway widely patent (small arrow), soft palate outlined by air in the oral cavity (long arrow). (B) Airway collapse during inspiration; hypopharyngeal and retroglossal collapse (arrowheads), nasopharyngeal airway widely patent (small arrow), soft palate outlined by air in the oral cavity (long arrow), oral cavity cut off from the airway. The facial surface is in the image to show the location of the nose (N) and mouth (M).
Table 3—Demographics for obstructive sleep apnea study subjects. Subject Number 1 2 3 4 5 6 7 8
Type of Study P P P P C C C C
Age (years) 17.98 4.91 3.46 9.73 15.79 4.35 17.29 5.03
Height (cm) 156.8 100.2 92.6 117.5 166.9 87.7 143.0 NA
Weight (kg) 72.6 14.2 14.3 23.9 57.6 11.3 60.7 18.4
Sex Male Female Male Male Male Male Female Female
AnteriorTransverse Posterior Cranial Cranial Diameter (cm) Diameter (cm) 19.4 15.6 a a 15.9 13.5 a 15.9 a 14.3 17.6 14.1 18.3 15.5 a a 16.5 13.9 17.3 14.1 13.2 a 15.4 a
Respective median (interquartile range) and mean (95% confidence interval) for controls: anterior-posterior, 18.2 (16.95–19.4) and 22.35 (11.7–110.45); transverse, 14.4 (13.6–15.2) and 14.13 (8.3–17.15). a Falls outside of the interquartile range for controls. C = clinical, NA = not available, P = prospective.
study subjects are included in Table 3. Most of the subjects had smaller AP and Tx cranial diameters than the corresponding age-matched control group. The effective dose for each of these groups is reported in Table 4. Dynamic 3D CT studies were performed with a radiation dose that is within the range of radiation exposure used for facial CT, as shown by the effective dose calculated for each subject (Table 2). Multiphase, dynamic CT scans of the UA for all eight of the study subjects were performed at an effective dose equal to or lower than that for a standard facial CT. In fact, four of the eight were at a dose below the fifth percentile for the facial CT controls. Blurring due to motion was present in two subjects from the prospective group (Subjects
All studies were of diagnostic quality, and the airway was reformatted for viewing as both a multiplanar image (Figure 1) and a 3D-surface-rendered cine movie (Video 1 in the supplemental material) using a commercially available automated segmentation tool (Vital Images, Inc.). The demographic characteristics of each study subject are summarized in Table 3. There were no significant differences in any of the measured demographic parameters between the study subjects and the age-matched control groups (n = 41) used for determining radiation dose. However, because of incomplete height and weight data for the control group, the cranial AP and Tx diameters of each of the control age groups are reported in Table 4. The cranial AP and Tx diameters of the Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
192
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Table 4—Effective dose and cranial diameter for control groups. Controls Age Groups 1 y old 5 y old 10 y old Adult
n 10 10 11 10
Mean Effective Dose (mSv) 0.625 0.658 0.459 0.507
Standard Deviation 0.316 0.544 0.100 0.240
Lower Quartile 0.316 0.324 0.352 0.247
AnteriorTransverse Posterior Cranial Cranial Diameter Diameter (cm) (cm) 15.130 12.050 17.892 14.690 18.791 14.764 19.770 15.040
Upper Quartile 0.952 0.700 0.524 0.601
mSv = milliSievert.
Table 5—Nasopharyngeal airway measurements and fractional collapse for prospective subjects. Subject Number 1 2 3 4
AP (mm) 6.7 5.3 9.5 9.9
Maximum Tx (mm) Area (mm2) 22.6 94.0 10.2 47.0 10.8 75.0 23.6 164.0
Minimum Tx (mm) 0.0 9.2 7.6 4.9
AP (mm) 0.0 4.7 7.1 1.8
Area (mm2) 0.0 38.0 34.0 8.0
AP 1.00 0.11 0.25 0.82
Fractional Collapse Tx 1.00 0.10 0.30 0.79
Area 1.00 0.19 0.55 0.95
Fractional Collapse Tx 1.00 0.79 0.74 0.65
Area 1.00 0.95 0.89 0.84
AP = anterior-posterior, Tx = transverse.
Table 6—Retroglossal airway measurements and fractional collapse for prospective subjects. Subject Number 1 2 3 4
AP (mm) 8.5 3.7 10.5 13.7
Maximum Tx (mm) Area (mm2) 25.1 126.5 11.7 42.0 17.4 95.0 26.3 280.0
Minimum Tx (mm) 0.0 2.5 4.6 9.2
AP (mm) 0.0 1.2 3.2 5.7
Area (mm2) 0.0 2.0 10.0 44.0
AP 1.00 0.68 0.70 0.58
AP = anterior-posterior, Tx = transverse.
Table 7—Summary of dynamics of the nasopharyngeal and retroglossal airways for prospective subjects. Subject Number 1 2 3 4
Nasopharyngeal Airway Dynamic Collapse Y Y N N Y N Y Y
Retroglossal Airway Dynamic Collapse Y Y Y Y Y Y Y Y
Type of Collapse Complete multilevel collapse NP stable with RG collapse NP dynamic with RG collapse Complete multilevel collapse
NP = nasopharyngeal, RG = retroglossal.
of the airway (see Tables 5 and 6), such as the complete multilevel collapse observed in Subjects 1 and 4, the stable noncollapsing airway observed in Subject 2, or the dynamic but noncollapsing airway observed in Subject 3 (Video 1). Airway dynamics of the NPA and RGA of all eight study subjects are summarized in Table 7. The three patterns of airway motion (stable, dynamic motion, and collapse) are illustrated in Figure 2. Each graph summarizes the cross-sectional area of the NPA and RGA over the time of the dynamic 3D CT. The data were smoothed with a spline function. Subject 2 (Figure 2A) had a NPA that was relatively stable and RGA with intermittent collapse. Subject 3 (Figure 2B) had a paradoxical movement of the NPA and RGA relationship due to occlusion of the retro-palatal airway during
2 and 3), but only visible on lung window. Reconstruction and measurements were not affected because the true surface was evident, utilizing full width at half-maximum measurement and display techniques established for CT.26 However, 61% of Subject 1 images and 86% of Subject 4 images showed no blurring. In all cases, image quality was judged by both readers as excellent. The measures of the NPA and RGA are summarized in Tables 5 and 6, respectively, and include the FC. We found that the motion of the airway was strongly dynamic. Three patterns of airway motion were noted: stable (FC < 0.50), dynamic motion (FC > 0.50), and collapse (FC > 0.80). FC is used to express the degree of collapse in the Tx, AP, or cross-sectional area values. It was possible to identify different dynamic behavior 193
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Figure 2—Fractional collapse of the NPA and the RGA over a respiratory cycle.
Table 8—Scan settings for Toshiba Aquilion ONE (Toshiba, Otawara, Japan) scanner. Age 0–2 y >2y
A
kVp 80 100
mA 40 30
kVp = kilovoltage, mA = milliampere.
D I SCUS S I O N The primary purpose of this report is to describe a dynamic 3D CT technique that can be used to evaluate the entire UA, including the dynamic movement in patients with OSA, and establish the effective radiation dose comparable to that used for control subjects undergoing standard CT studies. We think that having a benchmark radiation dose and a description of the technique is important for clinical and research communities treating OSA. We demonstrate the feasibility of acquiring high-quality dynamic images of the UA by CT at relatively low radiation dose. Furthermore, the image resolution was sufficient to delineate the UA contour and quantify the dynamic profile of the UA. In this study, four subjects (the prospective subjects) were undergoing a clinically indicated sleep MRI for OSA as described by Donnelly7; cine MRI is the primary imaging modality for evaluation of children with persistent OSA at our institution. However, some patients cannot undergo cine sleep MRI because of artifact caused by metal-containing oral appliances or implanted devices (e.g., pacemakers) that are not magnetic resonance compatible.7 In this study, we included an additional four patients who were scanned by dynamic 3D CT only because of contraindications for magnetic resonance (the clinical subjects). Dynamic 3D CT of the UA is a viable alternative to cine sleep MRI and has several advantages; (1) the examination is brief, (2) spatial resolution is better than MRI, (3) temporal resolution is better at 175 msec per dynamic temporal phase, (4) data are much easier to segment, (5) bony structures are amenable to automated segmentation for 3D images and cephalometric measurement, and most importantly, (6) a dynamic 3D dataset can be obtained of the entire UA simultaneously with isotropic resolution for later reconstruction and analysis. This last aspect is something that is not currently available with clinical MRI. Other advantages for dynamic 3D CT of the UA exist. The CT scanner is quiet compared to a magnetic resonance scanner so that imaging can be performed during natural sleep without arousing the patient. However, MRI has been used during natural sleep in many settings13 and magnetic resonance scanners are becoming equipped with novel “silent modes” of scanning that are substantially quieter.28 The acquisition of a 3D dataset of the UA, inclusive of the nasal air passages, can be used for computational fluid dynamics modeling of air flow in the UA29–31 and ultimately modeling of fluid structure interaction, which is a method of combining computational fluid dynamics scans and interactions with the surrounding structures to predict the movement of the airway during breathing.
B
C
NPA = nasopharygeal airway (represented by the solid blue line), RGA = retroglossal airway (represented by the dashed yellow line). The respiratory cycle was repeated three times to show the pattern repetition, but it was only imaged once. (A) Subject 2, with a relatively stable NPA, but very dynamic collapse of the RGA. (B) Subject 3, with a relatively dynamic NPA with fractional collapse of 0.55 and a very dynamic collapse of the RGA (see also Video 1 in the supplemental material). (C) Subject 4, with a very dynamic collapse of both the NPA and RGA.
expiration.27 Subject 4 (Figure 2C) had a very dynamic concomitant collapse of both the NPA and RGA. Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
194
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
The goal of such a model would be to predict change in airway dynamics based on an intervention, such as a surgical procedure. Ultimately, this could lead to new and better procedures to remedy OSA. Use of dynamic 3D CT of the UA does have disadvantages, the most obvious of which is exposure to ionizing radiation. However, we have demonstrated that this type of study can be performed with a radiation dose comparable to that used during routine indications for imaging the same anatomic area of the body. Facial CT and this type of dynamic upper airway CT both contain the facial bones and glottis in all cases and therefore is directly comparable. In light of the serious repercussions of recurrent OSA and its effects on the patient’s heart, vascular and metabolic systems, and mental status,1,2 we would argue that dynamic 3D CT utilizes a reasonable radiation dose that is comparable to that used for anatomic imaging obtained to guide surgical treatment when MRI cannot be employed.5 Initially, the radiation dose for Subject 5, the first subject on whom this technique was used, and Subject 1, the first prospective subject, was extrapolated from experience in lowering the radiation dose in abdomen and chest studies and converted to a formula based on AP cranial diameter.21,32 However, because the goal was to extract information about the airway soft-tissue interface and bony structure, we pushed the dose lower for the remaining subjects. Image quality was favorably received and all cases were easily viewed and reconstructed using automatic segmentation at a commercially available workstation. Our current technique is similar to that described in the Methods section, but we now are using fixed voltage (kVp) and electric current (mA) as depicted in Table 8. Limitations of this report include the small number of study subjects enrolled and no direct comparison to the MRI findings. Also, inspiration and expiration were tracked, but were not measured at the time of CT. Airflow and mask pressure were measured in the induction room, which is something we plan to do in future research studies. Additionally, the generalizability of the information presented here is limited to wide detector-array CT scanners. Performance of dynamic imaging on other multidetector CT scanners is technically possible, but the extent of z-axis coverage would be limited by the number of detectors and their configuration, and radiation exposure and temporal resolution may differ between vendors. As well, the number of gantry rotations and scanning parameters would be different based on gantry rotation speed. In summary, dynamic 3D CT of the UA provides a unique 3D dataset with high spatial and temporal resolution of the dynamic motion of the entire UA simultaneously, which can provide important information about the dynamic patterns of collapse in OSA. The ease of rendering the bony structures and soft-tissue structure into multiplanar and 3D displays of the airway motion is invaluable for identifying structures that contribute to airway collapse to aid surgical planning. Additionally, these airway data can be segmented into a model that utilizes both computational fluid dynamics and fluid structure interaction modeling for virtual surgical planning. Dynamic 3D CT of the upper airway can be performed with a radiation dose less than that of a facial CT and provides a unique 3D dataset with high spatial and temporal resolution
depicting dynamic motion of the entire upper airway simultaneously, which can provide important information about dynamic patterns of collapse in OSA. A B B R E V I AT I O N S 3D, three-dimensional AP, anteriorposterior CT, computed tomography Dmax, diameter maximum Dmin, diameter minimum FC, fractional collapse kVp, kilovoltage potential mA, milliampere MRI, magnetic resonance imaging ms, millisecond mSv, millisievert NPA, nasopharyngeal airway OSA, obstructive sleep apnea RGA, retroglossal airway Tx, transverse UA, upper airway R E FE R E N CES 1. Nevin MA. Pediatric obesity, metabolic syndrome, and obstructive sleep apnea syndrome. Pediatr Ann. 2013;42(10):205–210. 2. Ryan S, Crinion SJ, McNicholas WT. Obesity and sleep-disordered breathing--when two ‘bad guys’ meet. QJM. 2014;107(12):949–954. 3. Lumeng JC, Chervin RD. Epidemiology of pediatric obstructive sleep apnea. Proc Am Thorac Soc. 2008;5(2):242–252. 4. Donnelly LF, Casper KA, Chen B, Koch BL. Defining normal upper airway motion in asymptomatic children during sleep by means of cine MR techniques. Radiology. 2002;223(1):176–180. 5. Shott SR, Donnelly LF. Cine magnetic resonance imaging: evaluation of persistent airway obstruction after tonsil and adenoidectomy in children with Down syndrome. Laryngoscope. 2004;114(10):1724–1729. 6. Arens R, McDonough JM, Costarino AT, et al. Magnetic resonance imaging of the upper airway structure of children with obstructive sleep apnea syndrome. Am J Respir Crit Care Med. 2001;164(4):698–703. 7. Donnelly LF. Obstructive sleep apnea in pediatric patients: evaluation with cine MR sleep studies. Radiology. 2005;236(3):768–778. 8. Schwab RJ, Gefter WB, Hoffman EA, Gupta KB, Pack AI. Dynamic upper airway imaging during awake respiration in normal subjects and patients with sleep disordered breathing. Am Rev Respir Dis. 1993;148(5):1385–1400. 9. Shorten GD, Opie NJ, Graziotti P, Morris I, Khangure M. Assessment of upper airway anatomy in awake, sedated and anaesthetised patients using magnetic resonance imaging. Anaesth Intensive Care. 1994;22(2):165–169. 10. Gibson SE, Myer CM, Strife JL, O’Connor DM. Sleep fluoroscopy for localization of upper airway obstruction in children. Ann Otol Rhinol Laryngol. 1996;105(9):678–683. 11. Kim YC, Lebel RM, Wu Z, Ward SL, Khoo MC, Nayak KS. Real-time 3D magnetic resonance imaging of the pharyngeal airway in sleep apnea. Magn Reson Med. 2014;71(4):1501–1510. 12. Zhu Y, Kim YC, Proctor MI, Narayanan SS, Nayak KS. Dynamic 3-D visualization of vocal tract shaping during speech. IEEE Trans Med Imaging. 2013;32(5):838–848. 13. Shin LK, Holbrook AB, Capasso R, et al. Improved sleep MRI at 3 tesla in patients with obstructive sleep apnea. J Magn Reson Imaging. 2013;38(5):1261–1266.
195
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA 14. Ell SR, Jolles H, Galvin JR. Cine CT demonstration of nonfixed upper airway obstruction. Am J Roentgenol. 1986;146(4):669–677. 15. Galvin JR, Rooholamini SA, Stanford W. Obstructive sleep apnea: diagnosis with ultrafast CT. Radiology. 1989;171(3):775–778. 16. Lee EY, Boiselle PM. Tracheobronchomalacia in infants and children: multidetector CT evaluation. Radiology. 2009;252(1):7–22. 17. Greenberg SB. Dynamic pulmonary CT of children. Am J Roentgenol. 2012;199(2):435–440. 18. Mahmoud M, Gunter J, Donnelly LF, Wang Y, Nick TG, Sadhasivam S. A comparison of dexmedetomidine with propofol for magnetic resonance imaging sleep studies in children. Anesth Analg. 2009;109(3):745–753. 19. Malviya S, Voepel-Lewis T, Tait A, Merkel S, Naughton N. Validity and reliability of the University of Michigan sedation scale (UMSS) in children undergoing sedation for computerized tomography [abstract]. Anesthesiology. 2000;93:A1259. 20. Podberesky D, Angel E, Yoshizumi T, et al. Comparison of radiation dose estimates and scan performance in pediatric high-resolution thoracic CT for volumetric 320-detector row, helical 64-detector row, and noncontiguous axial scan acquisitions. Acad Radiol. 2013;20(9):1152–1161. 21. Larson DB, Wang LL, Podberesky DJ, Goske MJ. System for verifiable CT radiation dose optimization based on image quality. Part I. Optimization model. Radiology. 2013;269(1):167–176. 22. Deak PD, Smal Y, Kalender WA. Multisection CT protocols: sex- and agespecific conversion factors used to determine effective dose from dose-length product. Radiology. 2010;257(1):158–166. 23. Isono S, Morrison D, Launois S, Feroah T, Whitelaw W, Remmers J. Static mechanics of the velopharynx of patients with obstructive sleep apnea. J Appl Physiol. 1993;75(1):148–154. 24. Donnelly LF, Surdulescu V, Chini BA, Casper KA, Poe SA, Amin RS. Upper airway motion depicted at cine MR imaging performed during sleep: comparison between young Patients with and those without obstructive sleep apnea. Radiology. 2003;227(1):239–245. 25. Eilers P, Marx B. Flexible smoothing with B-splines and penalties. Statist Sci. 1996;11(2):89–121. 26. Koehler PR, Anderson RE, Baxter B. The effect of computed tomography viewer controls on anatomical measurements. Radiology. 1979;130(1):189–194. 27. Buffington CW, Wells CM, Soose RJ. Expiratory upper airway obstruction caused by the soft palate during bag-mask ventilation. Open J Anesthesiol. 2012;2(2):38–43.
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
28. Wu Z, Kim YC, Khoo MC, Nayak KS. Evaluation of an independent linear model for acoustic noise on a conventional MRI scanner and implications for acoustic noise reduction. Magn Reson Med. 2014;71(4):1613–1620. 29. Chen J, Gutmark E. Numerical investigation of airflow in an idealized human extra-thoracic airway: a comparison study. Biomech Model Mechanobiol. 2014;13(1):205–214. 30. Mihaescu M, Murugappan S, Gutmark E, Donnelly LF, Khosla S, Kalra M. Computational fluid dynamics analysis of upper airway reconstructed from magnetic resonance imaging data. Ann Otol Rhinol Laryngol. 2008;117(4):303–309. 31. Mihaescu M, Murugappan S, Gutmark E, Donnelly LF, Kalra M. Computational modeling of upper airway before and after adenotonsillectomy for obstructive sleep apnea. Laryngoscope. 2008;118(2):360–362. 32. Larson DB, Malarik RJ, Hall SM, Podberesky DJ. System for verifiable CT radiation dose optimization based on image quality. Part II. Process control system. Radiology. 2013;269(1):177–185.
ACK N O W L E D G M E N T S The authors are grateful to J. Denise Wetzel, CCHMC Medical Writer, for critical review of the manuscript.
SUBM I SSI O N & CO R R ESPO NDENCE I NFO R M ATI O N Submitted for publication September, 2015 Submitted in final revised form September, 2016 Accepted for publication September, 2016 Address correspondence to: Robert J. Fleck, MD, Radiologist, Department of Radiology and Medical Imaging, Division Chief, Cardiac MRI, Associate Professor, UC Department of Radiology; Tel: (513) 636-9941; Fax: (513) 636-8145; Email: [email protected]
D I SCLO S U R E S TAT E M E N T This was not an industry supported study. This study was supported by National Institutes of Health grant (RO1HL105206-01). The authors have indicated no financial conflicts of interest.
196
S CI E NT IF IC IN VES TIGATIONS
Dynamic Volume Computed Tomography Imaging of the Upper Airway in Obstructive Sleep Apnea Robert J. Fleck, MD1; Stacey L. Ishman, MD, MPH2,3,4; Sally R. Shott, MD2,4; Ephraim J. Gutmark, PhD, DSc 4,5; Keith B. McConnell, MS3; Mohamed Mahmoud, MD1,6; Goutham Mylavarapu, PhD3; Dhananjay R. Subramaniam, MS5; Rhonda Szczesniak, PhD3,7; Raouf S. Amin, MD3,8 Division of Radiology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 2Division of Pediatric Otolaryngology - Head and Neck Surgery, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 3Division of Pulmonary Medicine, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 4Department of Otolaryngology - Head and Neck Surgery, University of Cincinnati School of Medicine, Cincinnati, OH; 5Department of Aerospace Engineering and Engineering Mechanics, CEAS, University of Cincinnati, Cincinnati, OH; 6Department of Anesthesia, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 7Division of Biostatistics and Epidemiology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH; 8Department of Pediatrics, University of Cincinnati School of Medicine, Cincinnati, OH
1
Study Objectives: To describe a dynamic three-dimensional (3D) computed tomography (CT) technique for the upper airway and compare the required radiation dose to that used for common clinical studies of a similar anatomical area, such as for subjects undergoing routine clinical facial CT. Methods: Dynamic upper-airway CT was performed on eight subjects with persistent obstructive sleep apnea, four of whom were undergoing magnetic resonance imaging and an additional four subjects who had a contraindication to magnetic resonance imaging. This Health Insurance Portability and Accountability Act-compliant study was approved by our institutional review board, and informed consent was obtained. The control subjects (n = 41) for comparison of radiation dose were obtained from a retrospective review of the clinical picture-archiving computer system to identify 10 age-matched patients per age-based control group undergoing facial CT. Results: Dynamic 3D CT can be performed with an effective radiation dose of less than 0.38 mSv, a dose that is less than or comparable to that used for clinical facial CT. The resulting data- set is a uniquely complete, dynamic 3D volume of the upper airway through a full respiratory cycle that can be processed for clinical and modeling analyses. Conclusions: A dynamic 3D CT technique of the upper airway is described that can be performed with a clinically reasonable radiation dose and sets a benchmark for future use. Keywords: Down syndrome, four-dimensional computed tomography, obstructive volumetric computed tomography, radiation dosage, sleep apnea, sleepdisordered breathing Citation: Fleck RJ, Ishman SL, Shott SR, Gutmark EJ, McConnell KB, Mahmoud M, Mylavarapu G, Subramaniam DR, Szczesniak R, Amin RS. Dynamic volume computed tomography imaging of the upper airway in obstructive sleep apnea. J Clin Sleep Med. 2017;13(2):189–196.
I N T RO D U C T I O N
BRIEF SUMMARY
Current Knowledge/Study Rationale: Dynamic 3D CT produces a unique 3D dynamic dataset for evaluation of the upper airway for clinical or research use. Dynamic 3D CT can be performed with a radiation dose less than that of a static examination of the exact same anatomy imaged for a facial CT for common clinical indications. Study Impact: Dynamic 3D CT is an alternative method capable of obtaining a uniquely dynamic 3D dataset for the evaluation of obstructive sleep apnea in patients with persistent, problematic sleep apnea and could be used to guide surgical treatment to result in better outcomes. The effective radiation dose for dynamic 3D CT is equal to or less than that of a facial CT, setting a benchmark for use of the technique.
Obstructive sleep apnea (OSA) is a common disorder that manifests as repeated upper airway (UA) obstruction during sleep. Increasing prevalence of OSA in both pediatric and adult populations has been associated with increasing obesity and metabolic syndrome in the overall population.1,2 It is estimated that up to 5% of adults and 3% of all children in the United States are affected by OSA.1,3 In children, OSA is associated with excessive daytime sleepiness, hyperactivity, attention deficit disorder, failure to thrive, cardiometabolic problems, and neurocognitive deficits.4 As part of the evaluation of OSA, imaging of the UA in patients with OSA has traditionally been performed with magnetic resonance imaging (MRI), computed tomography (CT), and/or fluoroscopy during natural sleep, drug-induced sleep, or wakefulness to evaluate airway size, configuration, and dynamics. Commercially available MRI scanners produce static images that are used to identify UA soft-tissue targets amenable to surgical modification, such as enlarged adenoid or lingual tonsils, macroglossia, elongated soft palate, or enlarged turbinates.5 Additionally, current MRI scanners can provide detailed
dynamic single-slice images of the airway during respirations that can be used to create a cine image of airway showing movement dynamics and pattern of collapse.6,7 Overall, MRI provides detailed information,6,8–10 but is limited to only single-slice cines. Although three-dimensional (3D) cine images obtained with compressed sensing techniques11,12 or multiple two-dimensional planes obtained simultaneously13 are available at advanced research sites, they are not yet available clinically. 189
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Table 1—Polysomnography data for study subjects. Subject Number 1 2 3 4 5 6 7 8
Type of Study P P P P C C C C
AHI 27.0 4.4 5.3 4.1 10.8 9.8 11.7 17.5
Lowest O2 Saturations (%) 78.0 84.0 90.1 92.2 85.0 92.6 88.0 79.9
% Time CO2 > 50% NR 0.7 0.0 8.4 0.4 0.0 NR 5.8
% REM Sleep 0.0 20.0 22.7 17.6 20.6 28.7 25.0 0.0
Sleep Efficiency 39.0 76.0 66.4 78.1 77.4 77.4 89.1 37.5
Cardiac Disease Yes No Yes Yes Yes No No Yes
AHI = apnea-hypopnea index, C = clinical, NR = not recorded, P = prospective, REM = rapid eye movement.
CT has the same anatomical coverage as dynamic 3D CT and second, facial CT is often used for planning craniofacial surgery, which is sometimes a consideration in patients with OSA. The required information for calculating radiation dose and area of coverage were recorded. The control group was divided into subgroups based on age, and effective dose was calculated for comparison to individual age-matched study subjects.
CT of the UA and cine CT have also been described previously.8,14,15 However, with the advent of slip-ring and multidetector CT technology, cine CT covering a wider anatomic area is now possible.16,17 Wide-detector scanners, with 256–320 detector rows, are sufficiently wide to image the UA from the nasal passages to the glottis during each rotation with high temporal resolution to stop motion and create a dynamic 3D dataset. For the current study, we modified and applied this method to the UA patients undergoing an imaging study during a sleep study for OSA aimed at creating a realistic model of UA dynamics. The purpose of this study was to describe a dynamic 3D CT technique for the UA, set a benchmark for radiation exposure requirements for this technique, and compare the radiation dose to that used for common clinical studies of an identical anatomical area (e.g., clinical facial CT).
Computed Tomography
All study subjects (both prospective and clinical) were anesthetized with dexmedetomidine and permitted to breathe spontaneously via the native airway using a protocol described previously.18 Depth of sedation was assessed using the University of Michigan Sedation Scale.19 After MRI was completed, the prospective subjects were transferred to the CT scanner and positioned with their head and neck in neutral position while anesthesia and adequate oxygenation were maintained. Patients with a contraindication to MRI (the clinical subjects) were transferred directly from induction to the CT scanner. All imaging was performed on a 320-slice volume CT scanner (Aquilion ONE, Toshiba, Otawara, Japan) and reconstructed using AIDR 3D with a FC18 reconstruction kernel.20 After scout views of the head and neck were obtained, the scan length was set to extend from the nasal air passages to the subglottic region, thus covering the entire nasal air passages, nasopharyngeal airway (NPA), retroglossal airway (RGA), and hypopharynx. The number of gantry rotations was determined by the respiratory interval in milliseconds (ms) divided by the gantry rotation speed (350msec/rotation) such that the period of imaging covered one respiratory cycle. Because the respiratory cycle was relatively stable, the scan duration was calculated by dividing the minute respiratory rate into 60 sec. Therefore, if the respiratory rate was 24 breaths per minute, scanning was performed for 2.5 sec. The maximum z-axis coverage was 16 cm and could be decreased in 2-cm increments to fit each patient’s requirements. Low-dose principles based on noise tolerance by radiologists were used when determining the dynamic 3D CT parameters.21 The tube voltage was determined as either 80-kVp or 100-kVp depending on each patient’s age and size (cranial diameter). The cranial diameter was measured from the sagittal scout views as the distance from the soft tissues over the frontal bone to the soft tissues of the occiput. The tube current was based on the
METHODS
Subjects
This evaluation of the UA in OSA was approved by our institutional review board; informed consent was obtained for four subjects for prospective study and four additional subjects were retrospectively obtained from the clinical picture-archiving computer system. We performed a Health Information Portability and Accountability Act-compliant prospective study of dynamic 3D CT of the UA in four children with Down syndrome with OSA [subset of participants in the DYMOSA study (R01HL105206-01)] who were under drug-induced sleep for MRI evaluation between February 2013 and June 2013 (prospective subjects group). The four retrospective, clinical patients all had a contraindication to MRI (clinical subjects group), but did have dynamic 3D CT only between July 2012 and June 2013. Prior to imaging, all study subjects had undergone adenotonsillectomy and postoperative polysomnography to document persistent OSA (Table 1). The study subjects did not have concomitant pulmonary disease. The control subjects (n = 41) were obtained from a retrospective review of the picture-archiving computer system to identify age-matched patients undergoing facial CT between August 2012 to June 2013 for consideration of cranial facial surgery and trauma. The reason for using facial CT is twofold; first, facial Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
190
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Table 2—Comparison of study subjects with age- and size-matched control subjects. Subject Number 1 2 3 4 5 6 7 8
Type of Study P P P P C C C C
Reference Control Group Adult 5 y old 5 y old 10 y old Adult 5 y old Adult 5 y old
kVp 100 80 100 100 100 80 100 100
mA 60 40 20 30 60 80 30 30
CTDIvol 12.35 2.3 b 2.9 b 6.22 b 12.43 4.2 b 5.1 b 4.9 b
DLP 197.6 23.1 b 46.4 b 99.5 b 198.9 66.8 a 81.6 b 77.8 a
Effective Dose (mSv) 0.35568 0.08085 b 0.1624 b 0.26865 b 0.37791 0.2338 a 0.15504 b 0.2723 a
Relation to Control Doses Equivalent Lower Lower Lower Equivalent Lower Lower Lower
Falls outside of the lower and upper quartiles for controls. b Falls outside of the 95% confidence interval for controls. C = clinical, CTDIvol = volume computed tomography dose index, DLP = dose length product, kVp = kilovoltage, mA = milliampere, mSv = milliSievert, P = prospective.
a
frontal occipital diameter obtained from the scout views and determined by the formula:
where Dmax is the maximum dimension and Dmin is the minimum dimension. FC is expressed as the degree of collapse in the Tx, AP, or cross-sectional area values. Collapse is defined as FC greater than 0.80, because it would greatly limit airflow from a fluid mechanics perspective, and dynamic motion is defined as FC greater than 0.50 based on observations in adults that the airway in patients with OSA is more dynamic than patients without OSA.8,24 Stable is defined as FC less than 0.50, which is an unproven, arbitrary limit.
mA = [(Frontal occipital diameter in cm) × 3] The effective dose for all studies was calculated using the method of Deak and colleagues.22 For comparison of radiation doses, at least 10 age- and size-matched facial CT images were used as controls for each age group.
Image Reconstruction
Statistical Analysis
The images were reconstructed from half-scans, which results in a number of images that are two times the number of gantry rotations. A half-scan is a reconstruction using 180° of the gantry rotation data. These dynamic temporal phases were reviewed at a workstation (Vital Images, Inc., Minnetonka, MN, USA) that allowed for simultaneous viewing of multiplanar and 3D reformats in static phases and as cine loops in multiple planes. The anterior-posterior (AP), transverse (Tx), and crosssectional areas were measured at the narrowest section of the NPA and the RGA, above the epiglottis. Dynamic motion was defined by a greater than 50% reduction in the airway crosssectional area, and collapse was defined by a greater than 80% reduction in cross-sectional area.23 The images also were assessed independently for quality on a four-point Likert-type scale (0-non diagnostic, 1-adequate, 2-good, and 3-excellent image quality) by two radiologists with 10 and 15 y of experience. Motion blurring was assessed on individual dynamics as either present or not present and expressed as a percentage of dynamic temporal phases in the CT study.
The radiation doses of the control groups were calculated in relation to an eight-number summary for comparison to the individual doses calculated for the study subjects. Values outside of the lower and upper quartiles or the 95% confidence intervals for the controls were reported. Demographic differences between the study subjects and the respective control group for each were examined by the Wilcoxon Mann-Whitney test, and Fisher exact test for continuous and categorical variables, respectively. Values of p < 0.05 were used to determine significance throughout the analysis. For each subject, the actual airway dimensions (AP, Tx, and cross-sectional area) for both the RGA and NPA were plotted over time. A smoothing curve with a cubic B-spline basis function25 was applied to each measure to obtain an overall indication of the longitudinal course of the airway opening, closing, and reopening during the respiratory cycle. R ES U LT S
Airway Analysis
The NPA and RGA were measured transversely through the narrowest section when fully distended, and the same slice was measured throughout the respiratory cycle. The UA is a dynamic structure.8 To express this dynamic behavior, we introduce the term fractional collapse (FC), which yields a dimensionless fraction from 0 to 1.0 determined by the formula:
The first clinical study (Subject 5) was performed prior to the prospective group, and the additional clinical studies (Subjects 6, 7, and 8) were performed after the prospective patients were enrolled and imaged. The radiation doses for Subjects 1 (first prospective) and 5 (first clinical) were substantially more than the doses for the remaining six subjects (Table 2). This was due to subsequent modification of the dynamic 3D CT parameters to minimize the dose while maintaining adequate quality.
FC = (Dmax – Dmin ) / Dmax
191
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Figure 1—Volume-rendered surface display of air-filled structures and face in Subject 3, a 3-y-old boy with obstructive sleep apnea and Down syndrome.
AB
(A) Fully distended airway during expiration; retroglossal airway widely patent (arrowhead), nasopharyngeal airway widely patent (small arrow), soft palate outlined by air in the oral cavity (long arrow). (B) Airway collapse during inspiration; hypopharyngeal and retroglossal collapse (arrowheads), nasopharyngeal airway widely patent (small arrow), soft palate outlined by air in the oral cavity (long arrow), oral cavity cut off from the airway. The facial surface is in the image to show the location of the nose (N) and mouth (M).
Table 3—Demographics for obstructive sleep apnea study subjects. Subject Number 1 2 3 4 5 6 7 8
Type of Study P P P P C C C C
Age (years) 17.98 4.91 3.46 9.73 15.79 4.35 17.29 5.03
Height (cm) 156.8 100.2 92.6 117.5 166.9 87.7 143.0 NA
Weight (kg) 72.6 14.2 14.3 23.9 57.6 11.3 60.7 18.4
Sex Male Female Male Male Male Male Female Female
AnteriorTransverse Posterior Cranial Cranial Diameter (cm) Diameter (cm) 19.4 15.6 a a 15.9 13.5 a 15.9 a 14.3 17.6 14.1 18.3 15.5 a a 16.5 13.9 17.3 14.1 13.2 a 15.4 a
Respective median (interquartile range) and mean (95% confidence interval) for controls: anterior-posterior, 18.2 (16.95–19.4) and 22.35 (11.7–110.45); transverse, 14.4 (13.6–15.2) and 14.13 (8.3–17.15). a Falls outside of the interquartile range for controls. C = clinical, NA = not available, P = prospective.
study subjects are included in Table 3. Most of the subjects had smaller AP and Tx cranial diameters than the corresponding age-matched control group. The effective dose for each of these groups is reported in Table 4. Dynamic 3D CT studies were performed with a radiation dose that is within the range of radiation exposure used for facial CT, as shown by the effective dose calculated for each subject (Table 2). Multiphase, dynamic CT scans of the UA for all eight of the study subjects were performed at an effective dose equal to or lower than that for a standard facial CT. In fact, four of the eight were at a dose below the fifth percentile for the facial CT controls. Blurring due to motion was present in two subjects from the prospective group (Subjects
All studies were of diagnostic quality, and the airway was reformatted for viewing as both a multiplanar image (Figure 1) and a 3D-surface-rendered cine movie (Video 1 in the supplemental material) using a commercially available automated segmentation tool (Vital Images, Inc.). The demographic characteristics of each study subject are summarized in Table 3. There were no significant differences in any of the measured demographic parameters between the study subjects and the age-matched control groups (n = 41) used for determining radiation dose. However, because of incomplete height and weight data for the control group, the cranial AP and Tx diameters of each of the control age groups are reported in Table 4. The cranial AP and Tx diameters of the Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
192
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Table 4—Effective dose and cranial diameter for control groups. Controls Age Groups 1 y old 5 y old 10 y old Adult
n 10 10 11 10
Mean Effective Dose (mSv) 0.625 0.658 0.459 0.507
Standard Deviation 0.316 0.544 0.100 0.240
Lower Quartile 0.316 0.324 0.352 0.247
AnteriorTransverse Posterior Cranial Cranial Diameter Diameter (cm) (cm) 15.130 12.050 17.892 14.690 18.791 14.764 19.770 15.040
Upper Quartile 0.952 0.700 0.524 0.601
mSv = milliSievert.
Table 5—Nasopharyngeal airway measurements and fractional collapse for prospective subjects. Subject Number 1 2 3 4
AP (mm) 6.7 5.3 9.5 9.9
Maximum Tx (mm) Area (mm2) 22.6 94.0 10.2 47.0 10.8 75.0 23.6 164.0
Minimum Tx (mm) 0.0 9.2 7.6 4.9
AP (mm) 0.0 4.7 7.1 1.8
Area (mm2) 0.0 38.0 34.0 8.0
AP 1.00 0.11 0.25 0.82
Fractional Collapse Tx 1.00 0.10 0.30 0.79
Area 1.00 0.19 0.55 0.95
Fractional Collapse Tx 1.00 0.79 0.74 0.65
Area 1.00 0.95 0.89 0.84
AP = anterior-posterior, Tx = transverse.
Table 6—Retroglossal airway measurements and fractional collapse for prospective subjects. Subject Number 1 2 3 4
AP (mm) 8.5 3.7 10.5 13.7
Maximum Tx (mm) Area (mm2) 25.1 126.5 11.7 42.0 17.4 95.0 26.3 280.0
Minimum Tx (mm) 0.0 2.5 4.6 9.2
AP (mm) 0.0 1.2 3.2 5.7
Area (mm2) 0.0 2.0 10.0 44.0
AP 1.00 0.68 0.70 0.58
AP = anterior-posterior, Tx = transverse.
Table 7—Summary of dynamics of the nasopharyngeal and retroglossal airways for prospective subjects. Subject Number 1 2 3 4
Nasopharyngeal Airway Dynamic Collapse Y Y N N Y N Y Y
Retroglossal Airway Dynamic Collapse Y Y Y Y Y Y Y Y
Type of Collapse Complete multilevel collapse NP stable with RG collapse NP dynamic with RG collapse Complete multilevel collapse
NP = nasopharyngeal, RG = retroglossal.
of the airway (see Tables 5 and 6), such as the complete multilevel collapse observed in Subjects 1 and 4, the stable noncollapsing airway observed in Subject 2, or the dynamic but noncollapsing airway observed in Subject 3 (Video 1). Airway dynamics of the NPA and RGA of all eight study subjects are summarized in Table 7. The three patterns of airway motion (stable, dynamic motion, and collapse) are illustrated in Figure 2. Each graph summarizes the cross-sectional area of the NPA and RGA over the time of the dynamic 3D CT. The data were smoothed with a spline function. Subject 2 (Figure 2A) had a NPA that was relatively stable and RGA with intermittent collapse. Subject 3 (Figure 2B) had a paradoxical movement of the NPA and RGA relationship due to occlusion of the retro-palatal airway during
2 and 3), but only visible on lung window. Reconstruction and measurements were not affected because the true surface was evident, utilizing full width at half-maximum measurement and display techniques established for CT.26 However, 61% of Subject 1 images and 86% of Subject 4 images showed no blurring. In all cases, image quality was judged by both readers as excellent. The measures of the NPA and RGA are summarized in Tables 5 and 6, respectively, and include the FC. We found that the motion of the airway was strongly dynamic. Three patterns of airway motion were noted: stable (FC < 0.50), dynamic motion (FC > 0.50), and collapse (FC > 0.80). FC is used to express the degree of collapse in the Tx, AP, or cross-sectional area values. It was possible to identify different dynamic behavior 193
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
Figure 2—Fractional collapse of the NPA and the RGA over a respiratory cycle.
Table 8—Scan settings for Toshiba Aquilion ONE (Toshiba, Otawara, Japan) scanner. Age 0–2 y >2y
A
kVp 80 100
mA 40 30
kVp = kilovoltage, mA = milliampere.
D I SCUS S I O N The primary purpose of this report is to describe a dynamic 3D CT technique that can be used to evaluate the entire UA, including the dynamic movement in patients with OSA, and establish the effective radiation dose comparable to that used for control subjects undergoing standard CT studies. We think that having a benchmark radiation dose and a description of the technique is important for clinical and research communities treating OSA. We demonstrate the feasibility of acquiring high-quality dynamic images of the UA by CT at relatively low radiation dose. Furthermore, the image resolution was sufficient to delineate the UA contour and quantify the dynamic profile of the UA. In this study, four subjects (the prospective subjects) were undergoing a clinically indicated sleep MRI for OSA as described by Donnelly7; cine MRI is the primary imaging modality for evaluation of children with persistent OSA at our institution. However, some patients cannot undergo cine sleep MRI because of artifact caused by metal-containing oral appliances or implanted devices (e.g., pacemakers) that are not magnetic resonance compatible.7 In this study, we included an additional four patients who were scanned by dynamic 3D CT only because of contraindications for magnetic resonance (the clinical subjects). Dynamic 3D CT of the UA is a viable alternative to cine sleep MRI and has several advantages; (1) the examination is brief, (2) spatial resolution is better than MRI, (3) temporal resolution is better at 175 msec per dynamic temporal phase, (4) data are much easier to segment, (5) bony structures are amenable to automated segmentation for 3D images and cephalometric measurement, and most importantly, (6) a dynamic 3D dataset can be obtained of the entire UA simultaneously with isotropic resolution for later reconstruction and analysis. This last aspect is something that is not currently available with clinical MRI. Other advantages for dynamic 3D CT of the UA exist. The CT scanner is quiet compared to a magnetic resonance scanner so that imaging can be performed during natural sleep without arousing the patient. However, MRI has been used during natural sleep in many settings13 and magnetic resonance scanners are becoming equipped with novel “silent modes” of scanning that are substantially quieter.28 The acquisition of a 3D dataset of the UA, inclusive of the nasal air passages, can be used for computational fluid dynamics modeling of air flow in the UA29–31 and ultimately modeling of fluid structure interaction, which is a method of combining computational fluid dynamics scans and interactions with the surrounding structures to predict the movement of the airway during breathing.
B
C
NPA = nasopharygeal airway (represented by the solid blue line), RGA = retroglossal airway (represented by the dashed yellow line). The respiratory cycle was repeated three times to show the pattern repetition, but it was only imaged once. (A) Subject 2, with a relatively stable NPA, but very dynamic collapse of the RGA. (B) Subject 3, with a relatively dynamic NPA with fractional collapse of 0.55 and a very dynamic collapse of the RGA (see also Video 1 in the supplemental material). (C) Subject 4, with a very dynamic collapse of both the NPA and RGA.
expiration.27 Subject 4 (Figure 2C) had a very dynamic concomitant collapse of both the NPA and RGA. Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
194
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA
The goal of such a model would be to predict change in airway dynamics based on an intervention, such as a surgical procedure. Ultimately, this could lead to new and better procedures to remedy OSA. Use of dynamic 3D CT of the UA does have disadvantages, the most obvious of which is exposure to ionizing radiation. However, we have demonstrated that this type of study can be performed with a radiation dose comparable to that used during routine indications for imaging the same anatomic area of the body. Facial CT and this type of dynamic upper airway CT both contain the facial bones and glottis in all cases and therefore is directly comparable. In light of the serious repercussions of recurrent OSA and its effects on the patient’s heart, vascular and metabolic systems, and mental status,1,2 we would argue that dynamic 3D CT utilizes a reasonable radiation dose that is comparable to that used for anatomic imaging obtained to guide surgical treatment when MRI cannot be employed.5 Initially, the radiation dose for Subject 5, the first subject on whom this technique was used, and Subject 1, the first prospective subject, was extrapolated from experience in lowering the radiation dose in abdomen and chest studies and converted to a formula based on AP cranial diameter.21,32 However, because the goal was to extract information about the airway soft-tissue interface and bony structure, we pushed the dose lower for the remaining subjects. Image quality was favorably received and all cases were easily viewed and reconstructed using automatic segmentation at a commercially available workstation. Our current technique is similar to that described in the Methods section, but we now are using fixed voltage (kVp) and electric current (mA) as depicted in Table 8. Limitations of this report include the small number of study subjects enrolled and no direct comparison to the MRI findings. Also, inspiration and expiration were tracked, but were not measured at the time of CT. Airflow and mask pressure were measured in the induction room, which is something we plan to do in future research studies. Additionally, the generalizability of the information presented here is limited to wide detector-array CT scanners. Performance of dynamic imaging on other multidetector CT scanners is technically possible, but the extent of z-axis coverage would be limited by the number of detectors and their configuration, and radiation exposure and temporal resolution may differ between vendors. As well, the number of gantry rotations and scanning parameters would be different based on gantry rotation speed. In summary, dynamic 3D CT of the UA provides a unique 3D dataset with high spatial and temporal resolution of the dynamic motion of the entire UA simultaneously, which can provide important information about the dynamic patterns of collapse in OSA. The ease of rendering the bony structures and soft-tissue structure into multiplanar and 3D displays of the airway motion is invaluable for identifying structures that contribute to airway collapse to aid surgical planning. Additionally, these airway data can be segmented into a model that utilizes both computational fluid dynamics and fluid structure interaction modeling for virtual surgical planning. Dynamic 3D CT of the upper airway can be performed with a radiation dose less than that of a facial CT and provides a unique 3D dataset with high spatial and temporal resolution
depicting dynamic motion of the entire upper airway simultaneously, which can provide important information about dynamic patterns of collapse in OSA. A B B R E V I AT I O N S 3D, three-dimensional AP, anteriorposterior CT, computed tomography Dmax, diameter maximum Dmin, diameter minimum FC, fractional collapse kVp, kilovoltage potential mA, milliampere MRI, magnetic resonance imaging ms, millisecond mSv, millisievert NPA, nasopharyngeal airway OSA, obstructive sleep apnea RGA, retroglossal airway Tx, transverse UA, upper airway R E FE R E N CES 1. Nevin MA. Pediatric obesity, metabolic syndrome, and obstructive sleep apnea syndrome. Pediatr Ann. 2013;42(10):205–210. 2. Ryan S, Crinion SJ, McNicholas WT. Obesity and sleep-disordered breathing--when two ‘bad guys’ meet. QJM. 2014;107(12):949–954. 3. Lumeng JC, Chervin RD. Epidemiology of pediatric obstructive sleep apnea. Proc Am Thorac Soc. 2008;5(2):242–252. 4. Donnelly LF, Casper KA, Chen B, Koch BL. Defining normal upper airway motion in asymptomatic children during sleep by means of cine MR techniques. Radiology. 2002;223(1):176–180. 5. Shott SR, Donnelly LF. Cine magnetic resonance imaging: evaluation of persistent airway obstruction after tonsil and adenoidectomy in children with Down syndrome. Laryngoscope. 2004;114(10):1724–1729. 6. Arens R, McDonough JM, Costarino AT, et al. Magnetic resonance imaging of the upper airway structure of children with obstructive sleep apnea syndrome. Am J Respir Crit Care Med. 2001;164(4):698–703. 7. Donnelly LF. Obstructive sleep apnea in pediatric patients: evaluation with cine MR sleep studies. Radiology. 2005;236(3):768–778. 8. Schwab RJ, Gefter WB, Hoffman EA, Gupta KB, Pack AI. Dynamic upper airway imaging during awake respiration in normal subjects and patients with sleep disordered breathing. Am Rev Respir Dis. 1993;148(5):1385–1400. 9. Shorten GD, Opie NJ, Graziotti P, Morris I, Khangure M. Assessment of upper airway anatomy in awake, sedated and anaesthetised patients using magnetic resonance imaging. Anaesth Intensive Care. 1994;22(2):165–169. 10. Gibson SE, Myer CM, Strife JL, O’Connor DM. Sleep fluoroscopy for localization of upper airway obstruction in children. Ann Otol Rhinol Laryngol. 1996;105(9):678–683. 11. Kim YC, Lebel RM, Wu Z, Ward SL, Khoo MC, Nayak KS. Real-time 3D magnetic resonance imaging of the pharyngeal airway in sleep apnea. Magn Reson Med. 2014;71(4):1501–1510. 12. Zhu Y, Kim YC, Proctor MI, Narayanan SS, Nayak KS. Dynamic 3-D visualization of vocal tract shaping during speech. IEEE Trans Med Imaging. 2013;32(5):838–848. 13. Shin LK, Holbrook AB, Capasso R, et al. Improved sleep MRI at 3 tesla in patients with obstructive sleep apnea. J Magn Reson Imaging. 2013;38(5):1261–1266.
195
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
RJ Fleck, SL Ishman, SR Shott, et al. 4D CT of Upper Airway in OSA 14. Ell SR, Jolles H, Galvin JR. Cine CT demonstration of nonfixed upper airway obstruction. Am J Roentgenol. 1986;146(4):669–677. 15. Galvin JR, Rooholamini SA, Stanford W. Obstructive sleep apnea: diagnosis with ultrafast CT. Radiology. 1989;171(3):775–778. 16. Lee EY, Boiselle PM. Tracheobronchomalacia in infants and children: multidetector CT evaluation. Radiology. 2009;252(1):7–22. 17. Greenberg SB. Dynamic pulmonary CT of children. Am J Roentgenol. 2012;199(2):435–440. 18. Mahmoud M, Gunter J, Donnelly LF, Wang Y, Nick TG, Sadhasivam S. A comparison of dexmedetomidine with propofol for magnetic resonance imaging sleep studies in children. Anesth Analg. 2009;109(3):745–753. 19. Malviya S, Voepel-Lewis T, Tait A, Merkel S, Naughton N. Validity and reliability of the University of Michigan sedation scale (UMSS) in children undergoing sedation for computerized tomography [abstract]. Anesthesiology. 2000;93:A1259. 20. Podberesky D, Angel E, Yoshizumi T, et al. Comparison of radiation dose estimates and scan performance in pediatric high-resolution thoracic CT for volumetric 320-detector row, helical 64-detector row, and noncontiguous axial scan acquisitions. Acad Radiol. 2013;20(9):1152–1161. 21. Larson DB, Wang LL, Podberesky DJ, Goske MJ. System for verifiable CT radiation dose optimization based on image quality. Part I. Optimization model. Radiology. 2013;269(1):167–176. 22. Deak PD, Smal Y, Kalender WA. Multisection CT protocols: sex- and agespecific conversion factors used to determine effective dose from dose-length product. Radiology. 2010;257(1):158–166. 23. Isono S, Morrison D, Launois S, Feroah T, Whitelaw W, Remmers J. Static mechanics of the velopharynx of patients with obstructive sleep apnea. J Appl Physiol. 1993;75(1):148–154. 24. Donnelly LF, Surdulescu V, Chini BA, Casper KA, Poe SA, Amin RS. Upper airway motion depicted at cine MR imaging performed during sleep: comparison between young Patients with and those without obstructive sleep apnea. Radiology. 2003;227(1):239–245. 25. Eilers P, Marx B. Flexible smoothing with B-splines and penalties. Statist Sci. 1996;11(2):89–121. 26. Koehler PR, Anderson RE, Baxter B. The effect of computed tomography viewer controls on anatomical measurements. Radiology. 1979;130(1):189–194. 27. Buffington CW, Wells CM, Soose RJ. Expiratory upper airway obstruction caused by the soft palate during bag-mask ventilation. Open J Anesthesiol. 2012;2(2):38–43.
Journal of Clinical Sleep Medicine, Vol. 13, No. 2, 2017
28. Wu Z, Kim YC, Khoo MC, Nayak KS. Evaluation of an independent linear model for acoustic noise on a conventional MRI scanner and implications for acoustic noise reduction. Magn Reson Med. 2014;71(4):1613–1620. 29. Chen J, Gutmark E. Numerical investigation of airflow in an idealized human extra-thoracic airway: a comparison study. Biomech Model Mechanobiol. 2014;13(1):205–214. 30. Mihaescu M, Murugappan S, Gutmark E, Donnelly LF, Khosla S, Kalra M. Computational fluid dynamics analysis of upper airway reconstructed from magnetic resonance imaging data. Ann Otol Rhinol Laryngol. 2008;117(4):303–309. 31. Mihaescu M, Murugappan S, Gutmark E, Donnelly LF, Kalra M. Computational modeling of upper airway before and after adenotonsillectomy for obstructive sleep apnea. Laryngoscope. 2008;118(2):360–362. 32. Larson DB, Malarik RJ, Hall SM, Podberesky DJ. System for verifiable CT radiation dose optimization based on image quality. Part II. Process control system. Radiology. 2013;269(1):177–185.
ACK N O W L E D G M E N T S The authors are grateful to J. Denise Wetzel, CCHMC Medical Writer, for critical review of the manuscript.
SUBM I SSI O N & CO R R ESPO NDENCE I NFO R M ATI O N Submitted for publication September, 2015 Submitted in final revised form September, 2016 Accepted for publication September, 2016 Address correspondence to: Robert J. Fleck, MD, Radiologist, Department of Radiology and Medical Imaging, Division Chief, Cardiac MRI, Associate Professor, UC Department of Radiology; Tel: (513) 636-9941; Fax: (513) 636-8145; Email: [email protected]
D I SCLO S U R E S TAT E M E N T This was not an industry supported study. This study was supported by National Institutes of Health grant (RO1HL105206-01). The authors have indicated no financial conflicts of interest.
196
