Fig 1.
Schematic for nasal cycle logger.
(A) Schematic of electronics. (B) Printed circuit board (PCB). Component-side in green, soldering-side in blue. (C) Picture of component-side. (D) Picture of soldering-side. (E) Component layout. (F) Illustration of low Pressure Sensor (1” H2O to 30”H2O). (G) The device in its assembled form. (H) Respiration cannulas positioned in subject’s nares.
Table 1.
The full bill of materials for building a nasal cycle monitor.
Fig 2.
(A) Raw data overlaid with Hilbert transform and its peaks during 2 minute time scales. (B) Hilbert transform overlaid with its peaks and average during 30 minute time scales.
Fig 3.
Example of processed data from a typical subject.
(A) Average filtered nasal airflow peaks over time for right (blue) and left (green) nostrils (smoothed with a 20 minute window for display). Large red rectangle highlights a portion of sleep with right dominance indicated by black bar and negative inter nostril correlation (r = -0.44). Small red rectangle highlights a portion of wake with left dominance indicated by black bar and positive inter nostril correlation (r = 0.7). (B) Laterality index graph calculated and aligned for the data presented above. Light blue shading highlights low LI amplitude in wake (mean = 0.19) and light green highlights high amplitude in sleep (mean = 0.68).
Fig 4.
The device data reflected nasal obstruction.
(A) Time Series of MRI scans showing the nasal turbinates. Time measured from application of Otrivin (0.1% Xylometazoline Hcl) to the right nostril. Note right nostril decongestion over time. (B) Larger image of nasal passage. (C) Airflow measured in right nostril. (D) Airflow measured in left nostril. (E) Laterality index calculated as: . (F) Airflow in one subject with flipping of left and right nasal cannulas every 1.5 minutes. One logger channel in red and one in blue.
Fig 5.
The nasal cycle was highly variable within and across subjects.
(A) All intervals across all subjects. X-axis describes each subject (sorted by increasing variability), Y-axis describes all intervals measured for each subject during 24 hours. Each dot is an interval. (B) Range in parameters across subjects. Each dot is a subject. (C) LI distribution across the population. (D) Mean right vs left dominance intervals during 24 hours, reflecting a small but significant tendency to spend more time in left dominant intervals than in right dominant intervals. Error bars are SE.
Fig 6.
Laterality index measures for all subjects.
(A) Mean laterality index over 24 hours for 33 all subjects. (B) Mean laterality index amplitude over 24 hours for all 33 subjects.
Fig 7.
The nasal cycle differed in wake and sleep.
(A) Mean interval length. Inset: Mean over population. (B) Distribution of wake and sleep interval lengths across subjects. (C) Inter nostril correlation. Inset: Mean over population. (D) Mean LI amplitude Inset: Mean over population. (E) Mean LI. Inset: Mean over population. (F) Distribution of wake and sleep mean LI across subjects. (G) Right and left interval means during wake and sleep. In scatter plots each dot is a subject and the diagonal line is the unit slope line (X = Y). Error bars are SE.
Fig 8.
The difference between sleep and wake was evident in 1-hour windows.
Distribution of nasal cycle characteristics calculated in 1-hour non-overlapping windows. (A) Mean Laterality Index. (B) Amplitude of laterality index. (C) Inter nostril correlation. This figure implies that the results did not reflect an averaging artifact.
Fig 9.
The nasal cycle is linked to overall respiratory frequency.
(A) LI amplitude as reflected during highest and lowest respiration pace at wake. (B) LI amplitude as reflected during highest and lowest respiration pace at sleep. Note: We chose violin presentation for A and B panels as it demonstrates not only population mean but also the median and distribution of LI during different respiration conditions. Note that although mean and median values are correct, violin presentation deforms the distribution to be greater than 1, a case clearly not feasible for LI. (C) Comparison of mean LI amplitude during lowest vs highest respiration pace for each subject during wake. (D) Comparison of mean LI amplitude during lowest vs highest respiration pace for each subject during sleep.
Fig 10.
Relation between nasal cycle and body posture.
(A) Mean LI amplitude directly before and after position change. Data shown for all types of position change (To right, to left, to stomach, to back etc) pooled over 23 subjects. No significant difference is observed between directly before and after position change, indicating that position change alone does not produce an artifact in LI amplitude during sleep. (B) Correlation between Laterality index and body posture. Left: mean LI for each measured body position. Right: Comparison between ‘on right’ and ‘on left’ position LI means.
Fig 11.
Autocorrelation revealing oscillations in single nostrils: Examples from 3 individual subjects.
Twenty-four hour respiration envelopes of right nostril (left) and left nostril (middle) aligned with the corresponding laterality index calculation (Right). Inset: Autocorrelation results for each graph (black) with the fitted model (red–see Methods for fit details). (A) Subject 14 demonstrating similar significant 1/f fluctuation frequency in both right and left nostrils as well as in Laterality Index. This fits with the classical model of both nostrils fluctuating in similar out of phase rhythm resulting in LI with the same rhythm. (B) Subject 17 is also showing similar 1/f fluctuation frequency in right and left nostrils but here LI frequency is obviously contributed from nostril 1 as they are identical. (C) Subject 26 is demonstrating a rather noisy airflow in both right and left nostrils with no distinct oscillation therefore no significant value is detected by the model. Yet, a significant oscillation is detected in the laterality index vector indicating of significant reciprocal changes between the nostrils.
Fig 12.
Distribution of single nostril frequencies.
(A) Distribution of all frequencies over all significantly oscillatory nostrils, (right and left nostrils pooled together, see Methods for significance threshold). (B) Right nostril frequencies. (C) Left nostril frequencies. (D) Dominant frequency of left nostril vs dominant frequency of right nostril for five subjects in which both nostrils passed significance level.