Font Size: a A A

Characterization Of Pain-selective Neural Activity In The Human Brain Using Functional MRI

Posted on:2020-12-13Degree:DoctorType:Dissertation
Country:ChinaCandidate:Q SuFull Text:PDF
GTID:1364330590466433Subject:Medical imaging and nuclear medicine
Abstract/Summary:PDF Full Text Request
[Objective] How pain emerges from cortical activities remains an unresolved question in pain neuroscience.A large number of neuroimaging studies have suggested that a set of brain areas,the so-called “pain matrix”,subserve pain perception because they can be reliably activated by transient painful stimuli and their activation correlates with perceived pain intensity.However,painful stimuli are often intense and highly salient;therefore,it is being debated whether the activation of the so-called “pain matrix” reflects pain perception per se or simply stimulus intensity/saliency regardless of stimulus modality.This controversy mainly stems from several limitations of previous studies: First,as stimulus intensity was not matched when comparing brain responses to painful stimuli with those to non-painful stimuli,it remains unclear whether previously identified brain areas truly responded to pain preferentially or simply because painful stimuli were more intense.Second,most previous studies performed traditional univariate analysis which has been shown to be very limited in the ability of data mining compared with multivariate analysis methods.Third,most of previous studies searching for pain-selective neural activity focused on particular brain regions but overlooked large-scale activity patterns across different brain regions.Therefore,in order to better understand the neural mechanisms of pain in human brain,we performed three studies using functional Magnetic Resonance Imaging(fMRI)and machine learning.Part 1: We rigorously matched the perceived intensity between painful and non-painful stimuli so that we could identify brain activities that occur preferentially in response to painful stimuli in comparison to non-painful stimuli.Part 2: We aimed to test whether pain-selective information can be isolated in the fMRI responses to painful stimuli within the so-called “pain matrix” by utilizing the advanced machine learning technique,i.e.,multivariate pattern analysis of fMRI responses elicited by intensity/saliency-matched painful and non-painful stimuli.Part 3: Our previous results(obtained in Part 2)suggested that pain perception might be encoded both within certain local brain regions in small spatial scales but also across multiple distant brain regions in large spatial scales,therefore we aimed to systematically investigate in what spatial scales “pain-selective” neural activities measured using fMRI can be detected in the human brain.[Materials and Methods] Part 1: Fifty-one healthy participants were scanned using fMRI while receiving painful and tactile stimuli.Importantly,intensity ratings were collected for each stimulus and a subset of stimuli were selected with rigorously matched intensity ratings between painful and tactile stimuli.We then compared the brain activity elicited by the selected iso-intense painful and tactile stimuli using voxel-wise general linear modeling(GLM)analysis and model-free regions of interest(ROI)analysis.Part 2: In the so-called “pain matrix”,we tested whether multivariate pattern analysis(MVPA)could identify patterns distinguishing the fMRI responses triggered by(1)intensity-matched painful vs.non-painful stimuli,and(2)high vs.low intensity stimuli regardless of whether they elicit pain.Importantly,two independent datasets collected from different participants in different countries were used to test the reproducibility and the generalizability of the identified response patterns.Part 3: fMRI data of 62 healthy young adults were collected during painful and tactile stimulation using an event-related design.fMRI responses to painful and tactile stimuli in different spatial scales were created by image rescaling – parcellating the whole brain into different regions using a pre-defined brain atlas and then gradually merging smaller regions into large regions.In this way,we created fMRI images of four spatial scales:(1)original voxel-level images,(2)merging voxels into 246 regions according to a brain atlas,(3)further merging regions into 48 larger regions,and(4)further merging regions into 14 large regions.We performed both within-subject and between-subject “painful vs.tactile” classifications using whole-brain fMRI images of every spatial scale.Similar analyses were also performed for every brain region to test whether local brain regions and which regions contain pain-selective information.To explicitly test whether pain may be generated through the coordination of neural activities across different brain regions,we also performed “painful vs.tactile” classification using brain functional network constructed using fMRI images of every spatial scale.[Results] Part 1: We found that all brain areas activated by painful stimuli were also activated by tactile stimuli,and vice versa.Neural responses in these areas were correlated with the perceived stimulus intensity,regardless of stimulus modality.More importantly,among these activated areas,we further identified several brain regions showing stronger responses to painful stimuli than to tactile stimuli when perceived intensity was carefully matched,including the bilateral opercular cortex,the left supplementary motor area and the right frontal middle and inferior areas.Among these areas,the right frontal middle area still responded more strongly to painful stimuli even when painful stimuli were perceived less intense than tactile stimuli,whereas in this condition other regions now showed stronger responses to tactile stimuli.In contrast,the left postcentral gyrus,the visual cortex,the right parietal inferior gyrus,the left parietal superior gyrus and the right cerebellum were found to have stronger responses to tactile stimuli than to painful stimuli when perceived intensity was carefully matched.When tactile stimuli were perceived less intense than painful stimuli,the left postcentral gyrus and the right parietal inferior gyrus still responded more strongly to tactile stimuli while other regions now showed similar responses to painful and tactile stimuli.Part 2: We observed that MVPA was able to isolate response patterns distinguishing intensity/saliency-matched painful versus non-painful stimuli and response patterns distinguishing high versus low intensity/saliency stimuli regardless of whether they elicit pain.These response patterns were widely distributed within the “pain matrix”,and were reproducible within the two datasets and also generalizable across the two datasets.Part 3: We found(1)that “pain-selective” information could be detected from fMRI responses in all four spatial scales,both within subjects and between subjects;(2)that “pain-selective” spatial patterns of neural activities could be detected within many local areas but these areas varied considerably across subjects;(3)that “pain-selective” information could also be detected from functional connectivity patterns in two spatial scales(Scales 2 and 3).[Conclusions] Part 1: These results suggest that different brain areas may be engaged differentially when processing painful and tactile information,although their neural activities are not exclusively dedicated to encoding information of only one modality but are also strongly determined by perceived stimulus intensity regardless of stimulus modality.Part 2: These results indicate that neural activity in the so-called “pain matrix” is functionally heterogeneous,and carries information related to both painfulness and saliency.The response patterns distinguishing these aspects are spatially distributed and cannot be ascribed to specific brain structures.Part 3: Our results suggest that pain-specific neural coding is represented in the human brain across multiple spatial scales,both in small-scales(within local areas)and large-scales(across distant brain areas),suggesting a “network” representation of pain in the brain.The inter-subject variability in the brain regions containing pain-selective information might reflect variability in pain perception across different individuals.
Keywords/Search Tags:Pain, Tactile, Pain-selective, functional Magnetic Resonance Imaging, Machine Learning, Multivariate Pattern Analysis
PDF Full Text Request
Related items