Assessing chronic pain using brain entropy mapping

点击翻译按钮获取中文摘要

Chronic pain is one of the most prevalent health problems in the world but remains poorly understood and challenging to treat or manage. Two major barriers to progress are the unclear brain mechanism of chronic pain and the complexity to model the multifaceted individual differences in pain experience, making it difficult to accurately diagnose pain or monitor pain progression or treatment effects. To solve this problem, we need big data and sophistic models. In this novel project, we will use the large (n~100000) UK Biobank (UKB) data and deep machine learning (DL) to address the two critical problems in chronic pain research. We propose to use resting state fMRI (rsfMRI) because pain perception and processing are ongoing in the brain which can be characterized by the spontaneous brain activity measured by rsfMRI. We will focus on temporal coherence (TC) of rsfMRI given its fundamental role in brain functions and our leading expertise in this research topic. We initiated the concept of brain entropy (BEN) mapping as a tool to measure regional brain TC and our systematic work has demonstrated the high test-retest stability, sensitivity to causal effects, specificity to focal stimulation, different diseases, and drug states, as well as the potential as an intervention target through neuromodulations or medication. We also showed that TC/BEN contains unique information that can not be characterized by other neuroimaging measures. Aim 1 of this project will use TC/BEN mapping to find a potential chronic pain brain circuit where the resting TC is positively correlated with chronic pain and individual differences of pain experience. Aim 2 will use resting TC to build a multi-task DL-based pain prediction model. This project represents the first- of-its-kind to study TC in chronic pain. It will bring new knowledge about chronic pain brain mechanisms (resting TC alterations and associations) and a DL-based quantitative pain prediction model. Research rigor and method/finding generalizability will be ensured by the use of by far the largest rsfMRI data. These high innovations may lead to intervention targets for pain treatment or intervention development and a quantitative tool to evaluate individual differences in pain or pain progression. Feasibility of this project is guaranteed by the existing large data from UKB, our years of work experience in related research fields, the strong pilot data, and the strong team expertise. Success of this pilot project will immediately lead to large size future important studies in this new research direction.