Upper airway segmentation and dimensions estimation from cone-beam CT image datasets

Two-stage contextual transformer-based convolutional neural network for airway extraction from CT images

Accurate airway segmentation from computed tomography (CT) images is critical for planning navigation bronchoscopy and realizing a quantitative assessment of airway-related chronic obstructive pulmonary disease (COPD). Existing methods face difficulty in airway segmentation, particularly for the small branches of the airway. These difficulties arise due to the constraints of limited labeling and failure to meet clinical use requirements in COPD. We propose a two-stage framework with a novel 3D contextual transformer for segmenting the overall airway and small airway branches using CT images. The method consists of two training stages sharing the same modified 3D U-Net network. The novel 3D contextual transformer block is integrated into both the encoder and decoder path of the network to effectively capture contextual and long-range information. In the first training stage, the proposed network segments the overall airway with the overall airway mask. To improve the performance of the segmentation result, we generate the intrapulmonary airway branch label, and train the network to focus on producing small airway branches in the second training stage. Extensive experiments were performed on in-house and multiple public datasets. Quantitative and qualitative analyses demonstrate that our proposed method extracts significantly more branches and longer lengths of the airway tree while accomplishing state-of-the-art airway segmentation performance. The code is available at https://github.com/zhaozsq/airway_segmentation.

Artificial Intelligence as an Aid in CBCT Airway Analysis: A Systematic Review

BACKGROUND

The use of artificial intelligence (AI) in health sciences is becoming increasingly popular among doctors nowadays. This study evaluated the literature regarding the use of AI for CBCT airway analysis. To our knowledge, this is the first systematic review that examines the performance of artificial intelligence in CBCT airway analysis.

METHODS

Electronic databases and the reference lists of the relevant research papers were searched for published and unpublished literature. Study selection, data extraction, and risk of bias evaluation were all carried out independently and twice. Finally, five articles were chosen.

RESULTS

The results suggested a high correlation between the automatic and manual airway measurements indicating that the airway measurements may be automatically and accurately calculated from CBCT images.

CONCLUSIONS

According to the present literature, automatic airway segmentation can be used for clinical purposes. The main key findings of this systematic review are that the automatic airway segmentation is accurate in the measurement of the airway and, at the same time, appears to be fast and easy to use. However, the present literature is really limited, and more studies in the future providing high-quality evidence are needed.

Automatic segmentation of the pharyngeal airway space with convolutional neural network

OBJECTIVES

This study proposed and investigated the performance of a deep learning based three-dimensional (3D) convolutional neural network (CNN) model for automatic segmentation of the pharyngeal airway space (PAS).

METHODS

A dataset of 103 computed tomography (CT) and cone-beam CT (CBCT) scans was acquired from an orthognathic surgery patients database. The acquisition devices consisted of 1 CT (128-slice multi-slice spiral CT, Siemens Somatom Definition Flash, Siemens AG, Erlangen, Germany) and 2 CBCT devices (Promax 3D Max, Planmeca, Helsinki, Finland and Newtom VGi evo, Cefla, Imola, Italy) with different scanning parameters. A 3D CNN-based model (3D U-Net) was built for automatic segmentation of the PAS. The complete CT/CBCT dataset was split into three sets, training set (n = 48) for training the model based on the ground-truth observer-based manual segmentation, test set (n = 25) for getting the final performance of the model and validation set (n = 30) for evaluating the model's performance versus observer-based segmentation.

RESULTS

The CNN model was able to identify the segmented region with optimal precision (0.97±0.01) and recall (0.96±0.03). The maximal difference between the automatic segmentation and ground truth based on 95% hausdorff distance score was 0.98±0.74mm. The dice score of 0.97±0.02 confirmed the high similarity of the segmented region to the ground truth. The Intersection over union (IoU) metric was also found to be high (0.93±0.03). Based on the acquisition devices, Newtom VGi evo CBCT showed improved performance compared to the Promax 3D Max and CT device.

CONCLUSION

The proposed 3D U-Net model offered an accurate and time-efficient method for the segmentation of PAS from CT/CBCT images.

CLINICAL SIGNIFICANCE

The proposed method can allow clinicians to accurately and efficiently diagnose, plan treatment and follow-up patients with dento-skeletal deformities and obstructive sleep apnea which might influence the upper airway space, thereby further improving patient care.

A consideration of factors affecting palliative oral appliance effectiveness for obstructive sleep apnea: a scoping review

STUDY OBJECTIVES

This scoping review allows physicians, researchers, and others interested in obstructive sleep apnea to consider effectiveness of oral appliances (OAs). The intent is to improve understanding of OA effectiveness by considering morphologic interaction in patients with obstructive sleep apnea.

METHODS

Morphologic and biomechanical criteria for positional alterations of the mandible assessed success rates of OA appliances. Searches of databases (Medline, PubMed, The Cochrane Library, EBSCO) using terms: OA treatment effectiveness and positive and/or negative outcome predictors. Craniofacial predictors of OAs and obstructive sleep apnea biomechanical factors of anatomical traits associated with OA effectiveness were included. Databases searched radiographic cephalometric imaging for morphology/phenotypes and apnea-hypopnea index responses. Articles were excluded if title or abstract was not relevant or a case report. If the analysis did not report mean or standard deviation for apnea-hypoxia index, it was excluded. No language, age, or sex restrictions were applied.

RESULTS

Analysis of 135 articles included in searched literature indicated alterations in musculature and pharyngeal airway structure through OA use. These alterations were individually unpredictable with wide variability 61.81% ± 12.29 (apnea-hypoxia index mean ± standard deviation). Morphologic variations as predictors were typically weak and idiosyncratic. Biomechanical factors and wide variations in the metrics of appliance application were unclear, identifying gaps in knowledge and practice of OAs.

CONCLUSIONS

An integrated basis to identify morphologic and biomechanical elements of phenotypic expressions of sleep-disordered breathing in the design and application of OAs is needed. Current knowledge is heterogeneous and shows high variability. Identification of subgroups of patients with obstructive sleep apnea responding to OAs is needed.

Comparison of automatic airway analysis function of Invivo5 and Romexis software

Background

Visualization and calculation of the airway dimensions are important because an increase of airway resistance may lead to life-threatening emergencies. The visualization and calculation of the airway are possible using radiography technique with their advance software. The aim of this study was to compare and to test the reliability of the measurement of the upper airway volume and minimum area using airway analysis function in two software.

Methods

The sample consisted of 11 cone-beam computed tomography (CBCT) scans data, evaluated using the Invivo5 (Anatomage) and Romexis (version 3.8.2.R, Planmeca) software which afford image reconstruction, and airway analysis. The measurements were done twice with one week gap between the two measurements. The measurement obtained was analyzed with t-tests and intraclass correlation coefficient (ICC), with confidence intervals (CI) was set at 95%.

Results

From the analysis, the mean reading of volume and minimum area is not significantly different between Invivo5 and Romexis. Excellent intrarater reliability values were found for the both measurement on both software, with ICC values ranging from 0.940 to 0.998.

Discussion

The results suggested that both software can be used in further studies to investigate upper airway, thereby contributing to the diagnosis of upper airway obstructions.

Matlab® toolbox for semi-automatic segmentation of the human nasal cavity based on active shape modeling

The nose is a complex and important organ with a multitude of functions. Computational fluid dynamics (CFD) has been shown to be a valuable tool to obtain a better understanding of the functioning of the nose. CFD simulations require a surface geometry, which is constructed from tomographic data. This can be a very time-consuming task when one chooses to exclude the sinuses from the simulation domain, which in general keeps the size of the CFD model more manageable. In this work, an approach for the semi-automatic construction of the human nasal cavity is presented. In the first part, limited manual interaction is needed to create a coarse surface model. In the next part, this result is further refined based on the combination of active shape modeling with elastic surface deformation. The different steps are bundled in a Matlab toolbox with a graphical interface which guides the user. This interface allows easy manipulation of the data during intermediate steps, and also allows manual adjustments of the reconstructed nasal surface at the end. Two results are shown, and the approach and its precision are discussed. These results demonstrated that the followed approach can be used for the semi-automatic segmentation of a human nasal cavity from tomographic data, substantially reducing the amount of operator time.

A pilot study for segmentation of pharyngeal and sino-nasal airway subregions by automatic contour initialization

PURPOSE

The objective of the present study is to put forward a novel automatic segmentation algorithm to segment pharyngeal and sino-nasal airway subregions on 3D CBCT imaging datasets.

METHODS

A fully automatic segmentation of sino-nasal and pharyngeal airway subregions was implemented in MATLAB programing environment. The novelty of the algorithm is automatic initialization of contours in upper airway subregions. The algorithm is based on boundary definitions of the human anatomy along with shape constraints with an automatic initialization of contours to develop a complete algorithm which has a potential to enhance utility at clinical level. Post-initialization; five segmentation techniques: Chan-Vese level set (CVL), localized Chan-Vese level set (LCVL), Bhattacharya distance level set (BDL), Grow Cut (GC), and Sparse Field method (SFM) were used to test the robustness of automatic initialization.

RESULTS

Precision and F-score were found to be greater than 80% for all the regions with all five segmentation methods. High precision and low recall were observed with BDL and GC techniques indicating an under segmentation. Low precision and high recall values were observed with CVL and SFM methods indicating an over segmentation. A Larger F-score value was observed with SFM method for all the subregions. Minimum F-score value was observed for naso-ethmoidal and sphenoidal air sinus region, whereas a maximum F-score was observed in maxillary air sinuses region. The contour initialization was more accurate for maxillary air sinuses region in comparison with sphenoidal and naso-ethmoid regions.

CONCLUSION

The overall F-score was found to be greater than 80% for all the airway subregions using five segmentation techniques, indicating accurate contour initialization. Robustness of the algorithm needs to be further tested on severely deformed cases and on cases with different races and ethnicity for it to have global acceptance in Katradental radKatraiology workflow.

Automatic Pharynx Segmentation from MRI Data for Analysis of Sleep Related Disorders

In our project, we analyse throat structures using magnetic resonance imaging (MRI) to associate anatomic risk factors with sleep related disorders. Pharynx segmentation is the first step in the three-dimensional analysis of throat tissues.

We present a pipeline for automatic pharynx segmentation. The automatic part of the approach consists of three steps: smoothing, thresholding, 2D and 3D connected component analysis. Whereas two first steps are rather common, the third step provides a set of general rules for the automatic extraction of the pharyngeal component. Our method requires less than one minute to extract the pharyngeal structures. The approach is evaluated quantitatively on 30 data sets using region-based and edge-based measures.

A novel protocol for three-dimensional reconstruction of pharyngeal airway on computed tomography

Three-dimensional measurement of the pharyngeal airway has been widely used, but the three-dimensional reconstruction of pharyngeal airway has been performed in various ways, especially during the anterior boundary demarcation of the nasopharyngeal airway and oropharyngeal airway. This would inevitably affect the measurement and lead to noncomparison between different studies. Our study provided a novel method for anterior boundary demarcation of pharyngeal airway that defined the anterior boundary of nasopharyngeal airway as the "choana" according to the anatomical definition and defined the anterior boundary of oropharyngeal airway as a plane perpendicular to the long axis of soft palate and through the intersections of the lateral space and inferior space to soft palate according to the physiologic characteristics of soft palate. By 2-step segmentation, a three-dimensional image of pharyngeal airway was eventually reconstructed.Ten computed tomographic scans of pharyngeal airway were included for the anterior boundary demarcation and three-dimensional reconstruction by a medical imaging software (Surgicase 5.0; Materialise Interactive Medical Image Control System, Leuven, Belgium), with the volume and surface area being calculated. By using intraclass correlation coefficient, the reliability between intra- and interobservers of this method was well tested.The method established in this study for anterior boundary demarcation and three-dimensional reconstruction of pharyngeal airway is highly reliable and could more veritably reflect the intrinsic anatomical characteristics of the pharyngeal airway.

Automatic segmentation of the nasal cavity and paranasal sinuses from cone-beam CT images

PURPOSE

A patient-specific upper airway model is important for clinical, education, and research applications. Cone-beam computed tomography (CBCT) is used for imaging the upper airway but automatic segmentation is limited by noise and the complex anatomy. A multi-step level set segmentation scheme was developed for CBCT volumetric head scans to create a 3D model of the nasal cavity and paranasal sinuses.

METHODS

Gaussian mixture model thresholding and morphological operators are first employed to automatically locate the region of interest and to initialize the active contour. Second, the active contour driven by the Kullback-Leibler (K-L) divergence energy in a level set framework to segment the upper airway. The K-L divergence asymmetry is used to directly minimize the K-L divergence energy on the probability density function of the image intensity. Finally, to refine the segmentation result, an anisotropic localized active contour is employed which defines the local area based on shape prior information. The method was tested on ten CBCT data sets. The results were evaluated by the Dice coefficient, the volumetric overlap error (VOE), precision, recall, and F-score and compared with expert manual segmentation and existing methods.

RESULTS

The nasal cavity and paranasal sinuses were segmented in CBCT images with a median accuracy of 95.72 % [93.82-96.72 interquartile range] by Dice, 8.73 % [6.79-12.20] by VOE, 94.69 % [93.80-94.97] by precision, 97.73 % [92.70-98.79] by recall, and 95.72 % [93.82-96.69] by F-score.

CONCLUSION

Automated CBCT segmentation of the airway and paranasal sinuses was highly accurate in a test sample of clinical scans. The method may be useful in a variety of clinical, education, and research applications.

Three-dimensional evaluation of upper airway in patients with different anteroposterior skeletal patterns

OBJECTIVES

To investigate variability in the upper airway of subjects with different anteroposterior skeletal patterns by evaluating the volume and the most constricted cross-sectional area of the pharyngeal airway and defining correlations between the different variables.

MATERIAL AND METHODS

The study sample consisted of 60 patients (29 boys, 31 girls) divided into three groups: Class I (1 ≤ ANB ≤ 3), Class II (ANB>3), and Class III (ANB<1), to evaluate how the jaw relationship affects the airway volume and the most constricted cross-sectional area (Min-CSA). Differences between groups were determined using the Tukey-Kramer test. Correlations between variables were tested using Pearson's correlation coefficient.

RESULTS

The volume and the Min-CSA of the pharyngeal airway (PA) were significantly related to anteroposterior skeletal patterns (p < 0.05). The nasopharyngeal airway (NA) volume of Class I and Class III subjects was significantly larger than that of Class II subjects (p < 0.05). The Min-CSA and the length of PA were significantly related to the volume of PA (p < 0.05). The site and the size of the Min-CSA varied among the three groups.

CONCLUSIONS

The volume and the most constricted cross-sectional area of the airway varied with different anteroposterior skeletal patterns. The NA volume of Class I and Class III subjects was significantly larger than that of patients with a Class II skeletal pattern.

Change in upper airway geometry between upright and supine position during tidal nasal breathing

BACKGROUND

As the upper airway is the most important limiting factor for the deposition of inhalation medication in the lower airways, it is interesting to assess how its morphology varies between different postures. The goal of this study is to compare the upper airway morphology and functionality of healthy volunteers in the upright and supine positions during tidal nasal breathing and to search for baseline indicators for these changes. This is done by performing three-dimensional measurements on computed tomography (CT) and cone beam computed tomography (CBCT) scans.

METHODS

This prospective study was approved by all relevant institutional review boards. All patients gave their signed informed consent. In this study, 20 healthy volunteers (mean age, 62 years; age range, 37-78 years; mean body mass index, 29.26; body mass index range, 21.63-42.17; 16 men, 4 women) underwent a supine low-dose CT scan and an upright CBCT scan of the upper airway. The (local) average (Savg) and minimal (Smin) cross-sectional area, the position of the latter, the concavity, and the airway resistance were examined to determine if they changed from the upright to the supine position. If changes were found, baseline parameters were sought that were indicators for these differences.

RESULTS

There were five dropouts due to movement artifacts in the CBCT scans. Savg and Smin were 9.76% and 26.90% larger, respectively, in the CBCT scan than in the CT scan, whereas the resistance decreased by 26.15% in the upright position. The Savg of the region between the hard palate and the bottom of the uvula increased the most (49.85%). In people with a high body mass index, this value changed the least. The airway resistance in men decreased more than in women.

CONCLUSIONS

This study demonstrated that there are differences in upper airway morphology and functionality between the supine and upright positions and that there are baseline indicators for these differences.

Three-dimensional airway evaluation in 387 subjects from one university orthodontic clinic using cone beam computed tomography

OBJECTIVE

To determine the linear, volumetric, and cross-sectional area measurements in a large sample of subjects seeking treatment in a university clinic of orthodontics.

MATERIALS AND METHODS

Cone-beam computed tomography (CBCT) scans from 387 patients were analyzed retrospectively. All scans were loaded into the 3dMDvultus program (Atlanta, Ga) for airway analysis using automated segmentation. Gender, age, height, weight, airway length, volume, and area of maximum constriction were collected.

RESULTS

The rostrocaudal length of the airway increased with age until the age of 15 years in female subjects but continued to increase in male subjects and ranged from 44 to 88 mm over the period of 8 to 18 years of age with the gender data combined. The volume of the airway increased through the entire 11-year age range from 2000 mm(3) to 27,000 mm(3). The rate of airway volume increased at a slower rate in female than in male subjects. The smallest cross-sectional area increased with age ranging from 20 to 250 mm(2) but at a slower rate with female subjects, while male subjects demonstrated a more exponential increase after the age of 13 years.

CONCLUSIONS

The human airway increases in length and volume during a rapid period of craniofacial growth in patients between the ages of 8 and 18 years, and the site of maximum cross-sectional area constriction can vary.

Three-dimensional segmentation of the upper airway using cone beam CT: a systematic review

The objectives of this study were to systematically review the literature for studies that used cone beam CT (CBCT) to automatically or semi-automatically model the upper airway (including the pharyngeal, nasal and paranasal airways), and to assess their validity and reliability. Several electronic databases (MEDLINE®, MEDLINE In-Process & Other Non-Indexed Citations, all evidence-based medicine reviews including the Cochrane database, and Scopus) were searched. Abstracts that appeared to meet the initial selection criteria were selected by consensus. The original articles were then retrieved and their references were searched manually for potentially suitable articles that were missed during the electronic search. Final articles that met all the selection criteria were evaluated using a customized evaluation checklist. 16 articles were finally selected. From these, five scored more than 50% based on their methodology. Although eight articles reported the reliability of the airway model generated, only three used intraclass correlation (ICC). Two articles tested the accuracy/validity of airway models against the gold standard, manual segmentation, using volumetric measurements; however, neither used ICC. Only three articles properly tested the reliability of the three-dimensional (3D) upper airway model generated from CBCT and only one article had sufficiently sound methodology to test the airway model's accuracy/validity. The literature lacks proper scientific justification of a solid and optimized CBCT protocol for airway imaging. Owing to the limited number of adequate studies, it is difficult to generate a strong conclusion regarding the current validity and reliability of CBCT-generated 3D models.

Three-dimensional analysis of the airway with cone-beam computed tomography

INTRODUCTION

In this study, using a cone-beam computed tomography system, we evaluated the airways of 30 adults.

METHODS

The shapes of the 3-dimensional volume of the airway were analyzed and compared among the subjects by using surface superimposition software techniques.

RESULTS

The airway had the greatest variability in the hypopharynx, in the region below the epiglottis, and above the vocal folds. Moderate variation was apparent at the nares, behind the soft palate, and at the base of the tongue. Conservation of form was seen at the central portion of the nasal airway surrounding the inferior turbinate.

CONCLUSIONS

The potential for comparing the shape of the airway among subjects is possible.

Two- and three-dimensional evaluation of the upper airway after bimaxillary correction of Class III malocclusion

OBJECTIVE

The aim the study was to evaluate area and volumetric changes in the upper airway after bimaxillary correction of Class III malocclusion by the means of computer tomography (CT), and to compare these to the changes in linear measurements from lateral cephalograms.

STUDY DESIGN

This was a prospective clinical trial. Lateral cephalograms and CT scans of 10 Class III patients were evaluated 1 week before and 6 months after surgery. Wilcoxon matched pairs signed ranks test was used to determine the differences in measurements pre- and postoperatively. Spearman's rank correlation was used to test the association between the CT and cephalometric measurements.

RESULTS

CT measurements: The oropharyngeal and hypopharyngeal volumes increased by 3.98 +/- 4.18 cm(3) (P = .015) and 2.51 +/- 1.92 cm(3) (P = .021), respectively. The total volume of the posterior airway space increased, but the increase was not statistically significant. After surgery no change in the cross-sectional area of the upper airway was recorded at the retropalatal, oropharyngeal, or hypopharyngeal levels. Cephalometric measurements: The nasopharyngeal space increased 4.08 +/- 5.07 mm (P = .039) and the tongue increased in length by 4.84 +/- 5.93 mm (P = .22). No correlation was found between the measurements on CT scans and corresponding measurements on the lateral cephalograms.

CONCLUSIONS

Bimaxillary surgery for correction of Class III malocclusion did not cause decrease of the posterior airway space. Three-dimensional imaging techniques are preferable to 2-dimensional lateral cephalograms for evaluation of the upper airway after orthognathic procedures.

Evaluation of the human airway using cone-beam computerized tomography

OBJECTIVE

The goal of this project was to define and measure human airway space with radiographic volumetric 3-dimensional imaging and digital reconstruction of the pharynx using cone-beam computerized tomography.

STUDY DESIGN

This was a randomized retrospective study. Ten patient scans were selected randomly from a pool of 196 subjects seeking dental treatment at the University of California, San Francisco. Digital Imaging and Communications in Medicine-format volume images were captured using a low-radiation rapid-scanning cone-beam computerized tomography system (Hitachi MercuRay).

RESULTS

Detailed progressive rostrocaudal cross-sectional area histograms indicated that 8 of the 10 subjects demonstrated a region of maximum constriction near the oropharynx level. The most restricted cross-sectional area varied from 90 mm(2) to 360 mm(2).

CONCLUSIONS

The maximum constriction of the airway in 10 subjects quietly breathing for 10 seconds indicated variation in the level of the pharynx and the extent of the rostrocaudal zone of restriction.

The reliability of computed tomography (CT) values and dimensional measurements of the oropharyngeal region using cone beam CT: comparison with multidetector CT

OBJECTIVES

To evaluate the reliability of cone beam CT (CBCT) values and dimensional measurements of oropharyngeal air spaces as compared with those of multidetector row CT (MDCT).

METHODS

A soft tissue equivalent phantom with different sized holes was used. The phantom was scanned using CBCT and MDCT. In addition, a volunteer was examined using both methods for clinical comparison. The CT data were retrieved to a personal computer and treated using image processing software for three-dimensional rendering and dimensional measurements. The CT values of air, water and soft tissues were measured experimentally and the CT values of air, fat and muscle were measured in a clinical case. The diameter of each hole drilled in the phantom was measured on CBCT and MDCT images using the software tool and digimatic callipers.

RESULTS

The CBCT values were quite different from the Hounsfield units obtained with a typical MDCT system. The standard deviations were almost ten times larger with CBCT. In vivo assessment showed that the CBCT values for fat had a wide range that partially overlapped the values for muscle. The measurement of distances greater than 4 mm was consistent for all methods. The difference for holes was less than 0.2 mm.

CONCLUSIONS

The phantom study showed that CBCT provides limited quantitative CT values for each pixel on sliced images for differentiating air, water and soft tissues. However, the measurement of air spaces with CBCT was quite accurate.

Airway segmentation and measurement in CT images

In this paper we describe a methodology for constructing the airways from Cone Beam CT data and representing changes before and after a medical procedure. A seed region is automatically detected for the first CT slice using a heuristic algorithm incorporating morphological filtering. Our approach then extracts relevant contours on 3D slices by using gradient vector flow (GVF) snakes, modified by an edge detection and snake-shifting step. Following this, a 3D model is constructed. We then estimate the volume of the airway based on segmented 3D shape.