Minimally interactive segmentation of 4D dynamic upper airway MR images via fuzzy connectedness

A systematic review on 4D images of the upper airway in patients with OSA

AIM-BACKGROUND

In the treatment of obstructive sleep apnoea (OSA), oral appliances are now being recognized as a valuable alternative to continuous positive airway pressure (CPAP). Various static imaging techniques of the upper airways allow for assessment of bone and soft tissue structures. However, static images do not capture dynamic airway characteristics. The aim of this paper was to review 4D imaging techniques in patients with OSA.

METHODS

PubMed/MEDLINE, Web of Science and Embase were systematically searched for studies published before June 2022. The review was compliant with the PRISMA guidelines. The quality of each eligible study was critically evaluated by all four authors independently. Four unique articles with qualitative analyses were retrieved. All included studies had a clear objective/aim, an appropriate endpoint and sufficiently described eligibility criteria.

RESULTS

With dynamic imaging (4D) evaluation of the upper airway, the incidence of upper airway collapsibility due to use of a mandibular advancement device (MAD) was reduced, extraluminal tissue pressure was decreased and the space in the upper airway was increased, notably in the retropalatal and retroglossal areas of the airway. These findings suggest that MADs may be effective for OSA regardless of whether or not the obstruction site is in the velopharynx or oropharynx. However, further investigation of dynamic changes in the upper airway is required to explain the efficacy of OSA treatment and the underlying mechanisms.

Automatic upper airway segmentation in static and dynamic MRI via anatomy-guided convolutional neural networks

PURPOSE

Upper airway segmentation on MR images is a prerequisite step for quantitatively studying the anatomical structure and function of the upper airway and surrounding tissues. However, the complex variability of intensity and shape of anatomical structures and different modes of image acquisition commonly used in this application makes automatic upper airway segmentation challenging. In this paper, we develop and test a comprehensive deep learning-based segmentation system for use on MR images to address this problem.

MATERIALS AND METHODS

In our study, both static and dynamic MRI data sets are utilized, including 58 axial static 3D MRI studies, 22 mid-retropalatal dynamic 2D MRI studies, 21 mid-retroglossal dynamic 2D MRI studies, 36 mid-sagittal dynamic 2D MRI studies, and 23 isotropic dynamic 3D MRI studies, involving a total of 160 subjects and over 20 000 MRI slices. Samples of static and 2D dynamic MRI data sets were randomly divided into training, validation, and test sets by an approximate ratio of 5:2:3. Considering that the variability of annotation data among 3D dynamic MRIs was greater than for other MRI data sets, we increased the ratio of training data for these data to improve the robustness of the model. We designed a unified framework consisting of the following procedures. For static MRI, a generalized region-of-interest (GROI) strategy is applied to localize the partitions of nasal cavity and other portions of upper airway in axial data sets as two separate subobjects. Subsequently, the two subobjects are segmented by two separate 2D U-Nets. The two segmentation results are combined as the whole upper airway structure. The GROI strategy is also applied to other MRI modes. To minimize false-positive and false-negative rates in the segmentation results, we employed a novel loss function based explicitly on these rates to train the segmentation networks. An inter-reader study is conducted to test the performance of our system in comparison to human variability in ground truth (GT) segmentation of these challenging structures.

RESULTS

The proposed approach yielded mean Dice coefficients of 0.84±0.03, 0.89±0.13, 0.84±0.07, and 0.86±0.05 for static 3D MRI, mid-retropalatal/mid-retroglossal 2D dynamic MRI, mid-sagittal 2D dynamic MRI, and isotropic dynamic 3D MRI, respectively. The quantitative results show excellent agreement with manual delineation results. The inter-reader study results demonstrate that the segmentation performance of our approach is statistically indistinguishable from manual segmentations considering the inter-reader variability in GT.

CONCLUSIONS

The proposed method can be utilized for routine upper airway segmentation from static and dynamic MR images with high accuracy and efficiency. The proposed approach has the potential to be employed in other dynamic MRI-related applications, such as lung or heart segmentation.

Segmentation of 4D images via space-time neural networks

Medical imaging techniques currently produce 4D images that portray the dynamic behaviors and phenomena associated with internal structures. The segmentation of 4D images poses challenges different from those arising in segmenting 3D static images due to different patterns of variation of object shape and appearance in the space and time dimensions. In this paper, different network models are designed to learn the pattern of slice-to-slice change in the space and time dimensions independently. The two models then allow a gamut of strategies to actually segment the 4D image, such as segmentation following just the space or time dimension only, or following first the space dimension for one time instance and then following all time instances, or vice versa, etc. This paper investigates these strategies in the context of the obstructive sleep apnea (OSA) application and presents a unified deep learning framework to segment 4D images. Because of the sparse tubular nature of the upper airway and the surrounding low-contrast structures, inadequate contrast resolution obtainable in the magnetic resonance (MR) images leaves many challenges for effective segmentation of the dynamic airway in 4D MR images. Given that these upper airway structures are sparse, a Dice coefficient (DC) of ~0.88 for their segmentation based on our preferred strategy is similar to a DC of >0.95 for large non-sparse objects like liver, lungs, etc., constituting excellent accuracy.

Interactive iterative relative fuzzy connectedness lung segmentation on thoracic 4D dynamic MR images

Lung delineation via dynamic 4D thoracic magnetic resonance imaging (MRI) is necessary for quantitative image analysis for studying pediatric respiratory diseases such as thoracic insufficiency syndrome (TIS). This task is very challenging because of the often-extreme malformations of the thorax in TIS, lack of signal from bone and connective tissues resulting in inadequate image quality, abnormal thoracic dynamics, and the inability of the patients to cooperate with the protocol needed to get good quality images. We propose an interactive fuzzy connectedness approach as a potential practical solution to this difficult problem. Manual segmentation is too labor intensive especially due to the 4D nature of the data and can lead to low repeatability of the segmentation results. Registration-based approaches are somewhat inefficient and may produce inaccurate results due to accumulated registration errors and inadequate boundary information. The proposed approach works in a manner resembling the Iterative Livewire tool but uses iterative relative fuzzy connectedness (IRFC) as the delineation engine. Seeds needed by IRFC are set manually and are propagated from slice-to-slice, decreasing the needed human labor, and then a fuzzy connectedness map is automatically calculated almost instantaneously. If the segmentation is acceptable, the user selects "next" slice. Otherwise, the seeds are refined and the process continues. Although human interaction is needed, an advantage of the method is the high level of efficient user-control on the process and non-necessity to refine the results. Dynamic MRI sequences from 5 pediatric TIS patients involving 39 3D spatial volumes are used to evaluate the proposed approach. The method is compared to two other IRFC strategies with a higher level of automation. The proposed method yields an overall true positive and false positive volume fraction of 0.91 and 0.03, respectively, and Hausdorff boundary distance of 2 mm.