Abstract

In the cardiovascular and respiratory control system, slowly adapting pulmonary stretch receptors (SARs) provide the cardiovascular and respiratory control systems with information regarding the rate and depth of breathing. Previous studies have shown that the encoding and decoding functions of these mechanoreceptors are very reliable. However, it is important to ascertain whether information may also be transmitted by spike patterns. This dissertation evaluates the SAR firing characteristics and their physiological stimulus-response relationship.

Conventional spike train analyses are based on either spike quantity or predefined statistical models to differentiate responses to different stimuli. The SAR firing characteristics investigated in this dissertation depend upon neither. Firstly, two methods for classifying individual spike train responses under different stimulus amplitudes are introduced. One is based on the joint probability of observing the same spike/non-spike events in the single response and the model; the other is a simplified and faster method derived from the widely-used peri-stimulus time histogram (PSTH). Both use temporal discretization. Results presented show that both methods achieve good classification accuracy. Various discretization parameters were systematically varied and used to compare these methods’ performance. Specifically, effects of varying bin width, response epoch duration and two different response alignment strategies on classification accuracy are discussed.

Secondly, by means of these methods, it is feasible to study whether the temporal patterns of SAR spikes in a response carry information about different stimuli in addition to the number of spikes. Artificial spike trains are constructed with the same features as real spike trains except possible patterns. An evaluation standard is developed to demonstrate the presence of spike patterns by comparing the accuracy of classifying artificial spike trains vs. real ones. Differences of classification accuracy prove that spike patterns play an important role in stimulus information transmission.

In the third topic, a model spike train using K-means clustering inspired from relatively faithful SAR responses is proposed to provide the potential of describing event variance under different stimuli. The relation between SAR firing variance and different amplitude lung distension stimuli are studied. The trend of modeled firing variances illustrates that SARs fire more reliable while the stimulus is stronger.

Fourthly, SAR activity during bronchial constriction is explored with the methods and models above. Compared to normal respiration, different spike patterns are observed when the constriction agent methacholine is applied. Modeled firing variances show better firing reliability during bronchial constriction. These help us understand the relation between SAR firing consistency and their physical coupling with the airway tissue.

The research presented elucidates the potential neural code of SARs with respect to their physiological stimulus-response relationship. The classification method and the modeled spike train based on clustering demonstrated in this research is also useful for other spike train analyses, especially those used in analyzing firing responses to periodic stimuli.