Automated three-dimensional computed tomography analysis for surgical decisions in congenital nasal pyriform aperture stenosis

Abstract

Background: Congenital nasal pyriform aperture stenosis is a rare neonatal condition that causes respiratory distress and potentially requires surgery. Current diagnosis relies on clinical assessment and manual computed tomography (CT) measurements of the pyriform aperture width, which may not fully capture obstruction severity.

Objective: To evaluate the severity of pyriform aperture stenosis using automatic three-dimensional (3-D) analysis and identify parameters discriminating between conservative and surgical cases.

Materials and methods: This retrospective study analyzed CT scans of neonatal airways using a novel automated 3-D segmentation algorithm. We collected 22 CT scans (2010-2022) of newborns aged 0-35 days: 12 controls, four moderate cases treated conservatively, and six severe cases requiring surgery. The algorithm measured pyriform aperture width, nasal volumes, surface area, and cross-sectional areas.

Results: The algorithm achieved high accuracy (Dice coefficient, 0.961 ± 0.005) and aligned well with manual measurements of the pyriform aperture (average difference, -0.05 ± 0.77 mm, -0.7 ± 9.1%). All cases with stenosis showed anterior narrowing, while only severe cases exhibited stenosis along the entire mid-nasal cavity. Mid-nasal cavity volume, surface area, and cross-sectional areas at 50% and 75% of the mid-nasal cavity emerged as potential surgical predictors, with cross-sectional area at 75% being the most discriminating (moderate, 68.6 ± 7.5 mm²; severe, 33.5 ± 13.7 mm²; P<0.01).

Conclusion: Automated 3-D CT analysis quantifies pyriform aperture stenosis severity by measuring nasal airway dimensions. The study suggests objective parameters that may assist in surgical decisions and highlights the importance of considering the entire 3-D nasal cavity when planning surgical interventions. A multi-center study with a larger cohort is recommended to validate these findings.

Keywords: Airway management; Computed tomography; Computer-assisted; Computer-assisted decision-making; Congenital nasal pyriform aperture stenosis; Image interpretation; Three-dimensional imaging.

Fig. 1

Patient selection flowchart for congenital nasal pyriform aperture stenosis study. CT, computed tomography

Fig. 2

Axial computed tomography (CT) slices (without contrast) showing three examples of manual measurements of the pyriform aperture width using a bidirectional arrow. a A 4-day-old girl with normal pyriform aperture width (11 mm). b A 4-day-old girl with congenital nasal pyriform aperture stenosis treated conservatively (pyriform aperture width = 4.7 mm). c A 2-day-old boy with pyriform aperture stenosis who underwent surgery (pyriform aperture width = 6.7 mm)

Fig. 3

Automatic nasal airway segmentation process in an 8-day-old boy (computed tomography (CT) scan without contrast). a, b Selection of the volume of interest (VOI). a Reconstructed sagittal view showing the axial slices (black lines) that define the upper and lower boundaries of the nasal cavity and the nasopharynx, automatically selected by the algorithm. b Axial slice showing automatic cropping of the nasal airway region; a white box indicates the region of interest. c–e Nasal airway segmentation in axial view. c Nostril borders are defined by straight lines (white lines). d Initial segmentation using a global threshold of − 400 Hounsfield units (red, algorithm output) includes surrounding air, while narrow regions remain undetected. e Final segmentation after applying local adaptive thresholding and 3-connectivity filtering to include narrow regions (yellow indicates added areas, red indicates the initial segmentation; both are outputs of the automatic algorithm). The surrounding air was automatically removed

Fig. 4

Three-dimensional (D) models of the nasal airway in two newborns. a A 5-day-old girl with congenital pyriform aperture stenosis, demonstrating complete obstruction of one nostril. The 3-D model highlights connected airway regions in light blue and anatomically isolated regions in dark blue. In this case, the isolated nostril (dark blue) reflects a true lack of continuity with the main nasal airway, corresponding to complete obstruction of one side. This color distinction allows for intuitive identification of fully blocked regions. b A 6-day-old boy with no complete obstruction. Regional airway volumes are illustrated: nares (blue), mid-nasal region (light blue), and nasopharynx (dark blue)

Fig. 5

Comparison of the average cross-sectional area along the nasal cavity in three groups: normal (dark blue), moderate stenosis (blue), and severe stenosis (light blue). a Graph showing the combined cross-sectional area of both nasal cavities at each axial slice, averaged across subjects in each group. The approximate locations of the pyriform aperture and choanae are marked with dashed lines. b Group-wise comparison of cross-sectional area at representative locations in the mid-nasal cavity, demonstrating reduced airway dimensions in the severe group compared to the moderate and normal groups