A new approach for PET quantification using multilevel Bayesian modelling, and its application to study the neurobiology of depression

Depression is suffered by over 264 million people, and is one of the leading causes of disability worldwide.

Positron emission tomography (PET) is an in vivo imaging method essential for studying the neurochemical pathophysiology of this disease. However, its high cost and exposure of participants to radiation make it unfeasible to employ large sample sizes.

As such, the major shortcoming of PET imaging is its lack of power for studying clinically relevant research questions, particularly for conditions whose effects tend to be subtle, such as MDD.

We propose to develop, generalize, validate and extend a fundamentally new approach to PET modelling, using cutting-edge Bayesian methods in which quantification is performed for all individuals simultaneously, allowing the model to exploit similarities between individuals to improve both accuracy and specificity.

Preliminary results using a pilot implementation suggest that this method could make it possible to study effects four times smaller than what was previously possible, statistically equivalent to examining over ten times as many individuals.

We will first validate this method using simulated data, and then apply it to five previously collected and two newly-collected PET studies of patients with major depressive disorder: we will investigate patient-control differences in serotonin 1A receptors and transporters, monoamine oxidase A and the kappa opioid receptor, and treatment effects on the serotonin 1B receptor.