A new method for determining the optimal CT threshold for extracting the upper airway

An adjoint-based approach for the surgical correction of nasal septal deviations

Deviations of the septal wall are widespread anatomic anomalies of the human nose; they vary significantly in shape and location, and often cause the obstruction of the nasal airways. When severe, septal deviations need to be surgically corrected by ear-nose-throat (ENT) specialists. Septoplasty, however, has a low success rate, owing to the lack of suitable standardized clinical tools for assessing type and severity of obstructions, and for surgery planning. Moreover, the restoration of a perfectly straight septal wall is often impossible and possibly unnecessary. This paper introduces a procedure, based on advanced patient-specific Computational Fluid Dynamics (CFD) simulations, to support ENT surgeons in septoplasty planning. The method hinges upon the theory of adjoint-based optimization, and minimizes a cost function that indirectly accounts for viscous losses. A sensitivity map is computed on the mucosal wall to provide the surgeon with a simple quantification of how much tissue removal at each location would contribute to easing the obstruction. The optimization procedure is applied to three representative nasal anatomies, reconstructed from CT scans of patients affected by complex septal deviations. The computed sensitivity consistently identifies all the anomalies correctly. Virtual surgery, i.e. morphing of the anatomies according to the computed sensitivity, confirms that the characteristics of the nasal airflow improve significantly after small anatomy changes derived from adjoint-based optimization.

The Effect of Segmentation Threshold on Computational Fluid Dynamic Analysis of Nasal Airflow

BACKGROUND

The objective analysis of nasal airflow stands to benefit greatly from the adoption of computational fluid dynamic (CFD) methodologies. In this emerging field, no standards currently exist in regard to the ideal modeling parameters of the nasal airway. Such standards will be necessary for this tool to become clinically relevant.

METHODS

Human nasal airways were modeled from a healthy control, segmented, and analyzed with an in-house immersed boundary method. The segmentation Hounsfield unit (HU) threshold was varied to measure its effect in relation to airflow velocity magnitude and pressure change.

FINDINGS

Surface area and volume have a linear relationship to HU threshold, whereas CFD variables had a more complex relationship.

INTERPRETATION

The HU threshold should be included in nasal airflow CFD analysis. Future work is required to determine the optimal segmentation threshold.

Comparison of rhinomanometric and computational fluid dynamic assessment of nasal resistance with respect to measurement accuracy

PURPOSE

Computational fluid dynamics (CFD)-based calculation of intranasal airflow became an important method in rhinologic research. Current evidence shows weak to moderate correlation as well as a systematic underprediction of nasal resistance by numerical simulations. In this study, we investigate whether these differences can be explained by measurement uncertainties caused by rhinomanometric devices and procedures. Furthermore, preliminary findings regarding the impact of tissue movements are reported.

METHODS

A retrospective sample of 17 patients, who reported impaired nasal breathing and for which rhinomanometric (RMM) measurements using two different devices as well as computed tomography scans were available, was investigated in this study. Three patients also exhibited a marked collapse of the nasal valve. Agreement between both rhinomanometric measurements as well as between rhinomanometry and CFD-based calculations was assessed using linear correlation and Bland-Altman analyses. These analyses were performed for the volume flow rates measured at trans-nasal pressure differences of 75 and 150 Pa during inspiration and expiration.

RESULTS

The correlation between volume flow rates measured using both RMM devices was good (R² > 0.72 for all breathing states), and no relevant differences in measured flow rates was observed (21.6 ml/s and 14.8 ml/s for 75 and 150 Pa, respectively). In contrast, correlation between RMM and CFD was poor (R² < 0.5) and CFD systematically overpredicted RMM-based flow rate measurements (231.8 ml/s and 328.3 ml/s). No differences between patients with and without nasal valve collapse nor between inspiration and expiration were observed.

CONCLUSION

Biases introduced during RMM measurements, by either the chosen device, the operator or other aspects as for example the nasal cycle, are not strong enough to explain the gross differences commonly reported between RMM- and CFD-based measurement of nasal resistance. Additionally, tissue movement during breathing is most likely also no sufficient explanation for these differences.

Anatomic development of the upper airway during the first five years of life: A three-dimensional imaging study

Purpose

Normative data on the growth and development of the upper airway across the sexes is needed for the diagnosis and treatment of congenital and acquired respiratory anomalies and to gain insight on developmental changes in speech acoustics and disorders with craniofacial anomalies.

Methods

The growth of the upper airway in children ages birth to 5 years, as compared to adults, was quantified using an imaging database with computed tomography studies from typically developing individuals. Methodological criteria for scan inclusion and airway measurements included: head position, histogram-based airway segmentation, anatomic landmark placement, and development of a semi-automatic centerline for data extraction. A comprehensive set of 2D and 3D supra- and sub-glottal measurements from the choanae to tracheal opening were obtained including: naso-oro-laryngo-pharynx subregion volume and length, each subregion’s superior and inferior cross-sectional-area, and antero-posterior and transverse/width distances.

Results

Growth of the upper airway during the first 5 years of life was more pronounced in the vertical and transverse/lateral dimensions than in the antero-posterior dimension. By age 5 years, females have larger pharyngeal measurement than males. Prepubertal sex-differences were identified in the subglottal region.

Conclusions

Our findings demonstrate the importance of studying the growth of the upper airway in 3D. As the lumen length increases, its shape changes, becoming increasingly elliptical during the first 5 years of life. This study also emphasizes the importance of methodological considerations for both image acquisition and data extraction, as well as the use of consistent anatomic structures in defining pharyngeal regions.

Agreement between rhinomanometry and computed tomography-based computational fluid dynamics

Purpose

Active anterior rhinomanometry (AAR) and computed tomography (CT) are standardized methods for the evaluation of nasal obstruction. Recent attempts to correlate AAR with CT-based computational fluid dynamics (CFD) have been controversial. We aimed to investigate this correlation and agreement based on an in-house developed procedure.

Methods

In a pilot study, we retrospectively examined five subjects scheduled for septoplasty, along with preoperative digital volume tomography and AAR. The simulation was performed with Sailfish CFD, a lattice Boltzmann code. We examined the correlation and agreement of pressure derived from AAR (RhinoPress) and simulation (SimPress) and these of resistance during inspiration by 150 Pa pressure drop derived from AAR (RhinoRes150) and simulation (SimRes150). For investigation of correlation between pressures and between resistances, a univariate analysis of variance and a Pearson’s correlation were performed, respectively. For investigation of agreement, the Bland–Altman method was used.

Results

The correlation coefficient between RhinoPress and SimPress was r = 0.93 (p < 0.001). RhinoPress was similar to SimPress in the less obstructed nasal side and two times greater than SimPress in the more obstructed nasal side. A moderate correlation was found between RhinoRes150 and SimRes150 (r = 0.65; p = 0.041).

Conclusion

The simulation of rhinomanometry pressure by CT-based CFD seems more feasible with the lattice Boltzmann code in the less obstructed nasal side. In the more obstructed nasal side, error rates of up to 100% were encountered. Our results imply that the pressure and resistance derived from CT-based CFD and AAR were similar, yet not same.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11548-021-02332-1.

Decreased maxillary sinus volume is a potential predictor of obstructive sleep apnea

OBJECTIVES

To investigate the associations between nasal airway volume and the presence and severity of obstructive sleep apnea (OSA) in adults.

MATERIALS AND METHODS

The medical records of adult patients who visited the sleep clinic at University Hospital between June 2013 and April 2017 and underwent overnight polysomnography for the diagnosis of obstructive sleep apnea were reviewed retrospectively. Using computed tomography, the volumes of the nasal airways and maxillary sinuses were measured, and associations with the presence and severity of OSA were analyzed while controlling for the effects of possible confounders such as lateral cephalometric variables, maxillary widths, tongue/hyoid position, and soft palate dimensions.

RESULTS

Comparison between normal subjects and patients with OSA revealed that the latter had decreased ratios of maxillary sinus volume to whole nasal airway volume (P = .029) than normal subjects. OSA severity was greater in those with inferior positions of the hyoid (P = .010), in older patients (P = .011), and in those with high body mass index (P = .001). The volume of the total nasal airway or maxillary sinuses were not associated with OSA severity.

CONCLUSIONS

A decreased ratio of maxillary sinus volume to whole nasal airway volume is associated with adult OSA. However, OSA severity is not associated with either maxillary sinus volume or whole nasal airway volume.

Nasal cavity airflow: Comparing laser doppler anemometry and computational fluid dynamic simulations

Objective parameters to assess the physical flow conditions of breathing are scarce and decisions for surgery, e.g. nasal septum correction, mainly rely on subjective surgeon judgment. To define decision supporting parameters, we compare laser Doppler anemometry (LDA) and numerical computational fluid dynamic simulations (CFD) of the airflow velocity vector fields in the nasal cavity, including lattice Boltzmann (LB) and finite volume methods (FVM). The simulations are based on an anonymous patient CT dataset with septal deviation. LDA measurements are preformed using a 3D printed model. Nasal airflow geometry is randomly deformed in order to approximate surgical changes. The root-mean-square velocity error near the nasal valve of laser Doppler anemometry and lattice Boltzmann simulations is 0.071. Changes in geometry similarly affect both measurement and simulation.

Attention-Guided Generative Adversarial Network to Address Atypical Anatomy in Synthetic CT Generation

Recently, interest in MR-only treatment planning using synthetic CTs (synCTs) has grown rapidly in radiation therapy. However, developing class solutions for medical images that contain atypical anatomy remains a major limitation. In this paper, we propose a novel spatial attention-guided generative adversarial network (attention-GAN) model to generate accurate synCTs using T1-weighted MRI images as the input to address atypical anatomy. Experimental results on fifteen brain cancer patients show that attention-GAN outperformed existing synCT models and achieved an average MAE of 85.223±12.08, 232.41±60.86, 246.38±42.67 Hounsfield units between synCT and CT-SIM across the entire head, bone and air regions, respectively. Qualitative analysis shows that attention-GAN has the ability to use spatially focused areas to better handle outliers, areas with complex anatomy or post-surgical regions, and thus offer strong potential for supporting near real-time MR-only treatment planning.

Comparison of CBCT based synthetic CT methods suitable for proton dose calculations in adaptive proton therapy

In-room imaging is a prerequisite for adaptive proton therapy. The use of onboard cone-beam computed tomography (CBCT) imaging, which is routinely acquired for patient position verification, can enable daily dose reconstructions and plan adaptation decisions. Image quality deficiencies though, hamper dose calculation accuracy and make corrections of CBCTs a necessity. This study compared three methods to correct CBCTs and create synthetic CTs that are suitable for proton dose calculations. CBCTs, planning CTs and repeated CTs (rCT) from 33 H&N cancer patients were used to compare a deep convolutional neural network (DCNN), deformable image registration (DIR) and an analytical image-based correction method (AIC) for synthetic CT (sCT) generation. Image quality of sCTs was evaluated by comparison with a same-day rCT, using mean absolute error (MAE), mean error (ME), Dice similarity coefficient (DSC), structural non-uniformity (SNU) and signal/contrast-to-noise ratios (SNR/CNR) as metrics. Dosimetric accuracy was investigated in an intracranial setting by performing gamma analysis and calculating range shifts. Neural network-based sCTs resulted in the lowest MAE and ME (37/2 HU) and the highest DSC (0.96). While DIR and AIC generated images with a MAE of 44/77 HU, a ME of -8/1 HU and a DSC of 0.94/0.90. Gamma and range shift analysis showed almost no dosimetric difference between DCNN and DIR based sCTs. The lower image quality of AIC based sCTs affected dosimetric accuracy and resulted in lower pass ratios and higher range shifts. Patient-specific differences highlighted the advantages and disadvantages of each method. For the set of patients, the DCNN created synthetic CTs with the highest image quality. Accurate proton dose calculations were achieved by both DCNN and DIR based sCTs. The AIC method resulted in lower image quality and dose calculation accuracy was reduced compared to the other methods.

Sensitivity of nasal airflow variables computed via computational fluid dynamics to the computed tomography segmentation threshold

Computational fluid dynamics (CFD) allows quantitative assessment of transport phenomena in the human nasal cavity, including heat exchange, moisture transport, odorant uptake in the olfactory cleft, and regional delivery of pharmaceutical aerosols. The first step when applying CFD to investigate nasal airflow is to create a 3-dimensional reconstruction of the nasal anatomy from computed tomography (CT) scans or magnetic resonance images (MRI). However, a method to identify the exact location of the air-tissue boundary from CT scans or MRI is currently lacking. This introduces some uncertainty in the nasal cavity geometry. The radiodensity threshold for segmentation of the nasal airways has received little attention in the CFD literature. The goal of this study is to quantify how uncertainty in the segmentation threshold impacts CFD simulations of transport phenomena in the human nasal cavity. Three patients with nasal airway obstruction were included in the analysis. Pre-surgery CT scans were obtained after mucosal decongestion with oxymetazoline. For each patient, the nasal anatomy was reconstructed using three different thresholds in Hounsfield units (-800HU, -550HU, and -300HU). Our results demonstrate that some CFD variables (pressure drop, flowrate, airflow resistance) and anatomic variables (airspace cross-sectional area and volume) are strongly dependent on the segmentation threshold, while other CFD variables (intranasal flow distribution, surface area) are less sensitive to the segmentation threshold. These findings suggest that identification of an optimal threshold for segmentation of the nasal airway from CT scans will be important for good agreement between in vivo measurements and patient-specific CFD simulations of transport phenomena in the nasal cavity, particularly for processes sensitive to the transnasal pressure drop. We recommend that future CFD studies should always report the segmentation threshold used to reconstruct the nasal anatomy.

Generating synthetic CTs from magnetic resonance images using generative adversarial networks

PURPOSE

While MR-only treatment planning using synthetic CTs (synCTs) offers potential for streamlining clinical workflow, a need exists for an efficient and automated synCT generation in the brain to facilitate near real-time MR-only planning. This work describes a novel method for generating brain synCTs based on generative adversarial networks (GANs), a deep learning model that trains two competing networks simultaneously, and compares it to a deep convolutional neural network (CNN).

METHODS

Post-Gadolinium T1-Weighted and CT-SIM images from fifteen brain cancer patients were retrospectively analyzed. The GAN model was developed to generate synCTs using T1-weighted MRI images as the input using a residual network (ResNet) as the generator. The discriminator is a CNN with five convolutional layers that classified the input image as real or synthetic. Fivefold cross-validation was performed to validate our model. GAN performance was compared to CNN based on mean absolute error (MAE), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR) metrics between the synCT and CT images.

RESULTS

GAN training took ~11 h with a new case testing time of 5.7 ± 0.6 s. For GAN, MAEs between synCT and CT-SIM were 89.3 ± 10.3 Hounsfield units (HU) and 41.9 ± 8.6 HU across the entire FOV and tissues, respectively. However, MAE in the bone and air was, on average, ~240-255 HU. By comparison, the CNN model had an average full FOV MAE of 102.4 ± 11.1 HU. For GAN, the mean PSNR was 26.6 ± 1.2 and SSIM was 0.83 ± 0.03. GAN synCTs preserved details better than CNN, and regions of abnormal anatomy were well represented on GAN synCTs.

CONCLUSIONS

We developed and validated a GAN model using a single T1-weighted MR image as the input that generates robust, high quality synCTs in seconds. Our method offers strong potential for supporting near real-time MR-only treatment planning in the brain.

Radiographic evaluation of the symphysis menti as a donor site for an autologous bone graft in pre-implant surgery

Purpose

This study was performed to obtain a quantitative evaluation of the cortical and cancellous bone graft harvestable from the mental and canine regions, and to evaluate the cortical vestibular thickness.

Materials and Methods

This study collected cone-beam computed tomographic (CBCT) images of 100 Italian patients. The limits of the mental region were established: 5 mm in front of the medial margin of each mental foramen, 5 mm under the apex of each tooth present, and above the inferior mandibular cortex. Cortical and cancellous bone volumes were evaluated using SimPlant software (SimPlant 3-D Pro, Materialize, Leuven, Belgium) tools. In addition, the cortical vestibular thickness (minimal and maximal values) was evaluated in 3 cross-sections corresponding to the right canine tooth (3_R), the median section (M), and the left canine tooth (3_L).

Results

The cortical volume was 0.71±0.23 mL (0.27-1.96 mL) and the cancellous volume was 2.16±0.76 mL (0.86-6.28 mL). The minimal cortical vestibular thickness was 1.54±0.41 mm (0.61-3.25 mm), and the maximal cortical vestibular thickness was 3.14±0.75mm(1.01-5.83 mm).

Conclusion

The use of the imaging software allowed a patient-specific assessment of mental and canine region bone availability. The proposed evaluation method might help the surgeon in the selection of the donor site by the comparison between bone availability in the donor site and the reconstructive exigency of the recipient site.