Computer-controlled stimulation for functional magnetic resonance imaging studies of the neonatal olfactory system

Abstract

Aim: Olfactory sensation is highly functional early in human neonatal life, with studies suggesting that odours can influence behaviour and infant-mother bonding. Due to its good spatial properties, blood oxygen level-dependent (BOLD) contrast functional magnetic resonance imaging (fMRI) has the potential to rapidly advance our understanding of the neural activity which underlies the development of olfactory perception in this key period. We aimed to design an 'olfactometer' specifically for use with neonatal subjects for fMRI studies of odour perception.

Methods: We describe a fully automated and programmable, fMRI compatible system capable of presenting odorant liquids. To prevent contamination of the system and minimize between-subject infective risk, the majority of the olfactometer is constructed from single-use, readily available clinical equipment. The system was used to present the odour of infant formula milk in a validation group of seven neonatal subjects at term equivalent postmenstrual age (median age 40 weeks).

Results: A safe, reliable and reproducible pattern of stimulation was delivered leading to well-localized positive BOLD functional responses in the piriform cortex, amygdala, thalamus, insular cortex and cerebellum.

Conclusions: The described system is therefore suitable for detailed studies of the ontology of olfactory sensation and perception during early human brain development.

Keywords: Infant; Newborn; Olfactory; fMRI.

Figure 1

System architecture of the neonatal olfactometer: The olfactometer is composed of three subsystems, with the airflow preparation and odour-sourcing apparatus situated in the MR scanner console room. Medical grade breathable air is supplied to the system through the standard wall socket, with control and monitoring of air flow possible with a regulator and digital flow meter respectively. Air flow is then directed via the controlled opening of on/off valves into 1 of 3 odour chambers containing an odorant liquid. The vaporized odour is then delivered directly to the subject inside the scanner examination room via nasal cannulae (delivery apparatus).

Figure 2

The neonatal olfactometer was developed to be Functional magnetic resonance imaging (fMRI) compatible and to minimize between-subject infective risks: (A): The flow meter, valves (yellow arrow) and data acquisition card (National Instruments, Austin, TX USA) are housed inside a single control box that is situated in the scanner control suite; (B): To minimize infective risk, all components distal to the control box are single-use pieces of readily available clinical equipment such as mucous specimen traps (Pennine Healthcare, Derby, UK) (blue arrow), and antimicrobial respiratory filters are fitted (red arrow). (C): The delivery apparatus contains a manifold containing one-way valves (black arrow) to prevent the mixing of odours. These are then connected to nasal cannulae fitted to the subject prior to data acquisition.

Figure 3

Example olfactory responses to the odour of formula milk in two ex-preterm infants at term equivalent postmenstrual age (PMA). (A,B,C): Coronal (A), sagittal (B) and axial (C) T2-weighted images with an overlaid thresholded statistical map (corrected cluster significance of p < 0.05) showing well-localized functional responses in the piriform cortices bilaterally in an infant at 42 + 6 weeks PMA (blue arrows). (D,E,F): Well localized responses were also identified in the amygdalae as shown in an infant at 39 + 6 weeks PMA (yellow arrows). (G): In the example time series (averaged over the clusters of activation), the sampled Blood oxygen level dependent (BOLD) signal (red) can be seen to closely fit the predicted response as modelled by the pattern of stimulation and an age-appropriate Hemodynamic response function (HRF) model (green).

Figure 4

The average response to the odour of formula milk in seven infants at term equivalent postmenstrual age. The results of a one-sample nonparametric t-test (with a false discovery rate correction for multiple comparisons at threshold p < 0.05) have been overlaid on a custom T2-weighted template image in the coronal (A,B) and axial (C,D) planes. Clusters of activation are seen in the right piriform cortex and amygdala (blue arrows), the right thalamus (orange arrow), the left insular area (white arrow) and cerebellum (green arrow).