EFFECTS OF FLUID AND COMPUTED TOMOGRAPHIC TECHNICAL FACTORS ON CONSPICUITY OF CANINE AND FELINE NASAL TURBINATES

Abstract

Turbinate destruction is an important diagnostic criterion in canine and feline nasal computed tomography (CT). However decreased turbinate visibility may also be caused by technical CT settings and nasal fluid. The purpose of this experimental, crossover study was to determine whether fluid reduces conspicuity of canine and feline nasal turbinates in CT and if so, whether CT settings can maximize conspicuity. Three canine and three feline cadaver heads were used. Nasal slabs were CT-scanned before and after submerging them in a water bath; using sequential, helical, and ultrahigh resolution modes; with images in low, medium, and high frequency image reconstruction kernels; and with application of additional posterior fossa optimization and high contrast enhancing filters. Visible turbinate length was measured by a single observer using manual tracing. Nasal density heterogeneity was measured using the standard deviation (SD) of mean nasal density from a region of interest in each nasal cavity. Linear mixed-effect models using the R package 'nlme', multivariable models and standard post hoc Tukey pair-wise comparisons were performed to investigate the effect of several variables (nasal content, scanning mode, image reconstruction kernel, application of post reconstruction filters) on measured visible total turbinate length and SD of mean nasal density. All canine and feline water-filled nasal slabs showed significantly decreased visibility of nasal turbinates (P < 0.001). High frequency kernels provided the best turbinate visibility and highest SD of aerated nasal slabs, whereas medium frequency kernels were optimal for water-filled nasal slabs. Scanning mode and filter application had no effect on turbinate visibility.

Keywords: CT; artifact; cat; conchae; dog; nose.

FIG. 1

Plastic container used for the experiment (feline part) containing fitted foam slabs (A). Sagittal and dorsal CT topograms of water filled containers with canine nasal slabs (B, C). Sagittally reconstructed CT image of the container with foam separated canine nasal slabs (D).

FIG. 2

Transverse CT images of a rostral (A, B, C) canine nasal slab at the level of the 106 tooth and caudal (D, E, F) canine nasal slab at the mesiobuccal and mesiopalatal roots of the 108 tooth. (A, D) Air-filled without tracing; (B, E) air-filled with tracing (white lines) of turbinate length; and (C, F) water-filled with tracing. There is a marked reduction of visible turbinates when they are surrounded by fluid and/or in the presence of turbinate mucosal edema, indicative of effacement.For clarity these images were edited with software (Adobe Photoshop CS5, version 12.01, Adobe Systems Inc., San Francisco, CA) including cropping, application of tracer lines, and removal of extranasal water background.

FIG. 3

Transverse CT images of air-filled canine and feline nasal slabs demonstrating placement of 0.5-cm²-sized regions of interest (ROI) for the calculation of standard deviation of mean density. (A) A rostral canine nasal slab image at the level of the mesial root of the 106 tooth, (B) a feline nasal slab image at the level of the mesial root of the 107 tooth and (C) a caudal canine nasal slab image at the level of the mesiobuccal and mesiopalatal roots of the 108 tooth. The ROIs were placed in the ventral aspect of each nasal cavity in (A) and (B). In the caudal canine nasal slab image (C) the ROI was centered at the crossing point of horizontal and vertical positioning lines which emanate from the half-length points of the ventral quadrant lines.

FIG. 4

Boxplots representing the log₁₀ transformed total length of nasal turbinate length visibility and the log₁₀ transformed standard deviation of mean nasal density in canine nasal slabs subdivided into whether air- or water-filled: (A) overall; (B) sub-divided by kernel (H30, 31, 40, 41, 50, 70); (C) subdivided H70 kernel by filter, subdivided H50 kernel by filter in water filled; (D) subdivided H70 kernel by mode, subdivided H50 kernel by mode; and (E) total length by kernel (U40, U70, U90) in high resolution mode. For all plots both caudal and rostral slab data are included and the horizontal line in each box delineates the median value, the shaded boxes the interquartile range and the whiskers the full data range. PFO: posterior fossa optimization image filter, HCE: high contrast enhancement filter.

FIG. 5

Boxplots representing the log₁₀ transformed total length of nasal turbinate length visibility and the log₁₀ transformed standard deviation of mean nasal density in feline nasal slabs subdivided into whether air- or water-filled: (A) overall; (B) sub-divided by kernel (H30, 31, 40, 41, 50, 70); (C) subdivided H70 kernel by filter; (D) subdivided H70 kernel by mode; and (E) total length by kernel (U40, U70, U90) in high resolution mode. For all plots the horizontal line in each box delineates the median value, the shaded boxes the interquartile range and the whiskers the full data range. PFO: posterior fossa optimization image filter, HCE: high contrast enhancement filter.