Accuracy, reliability and limitations of functional and structural imaging data
Daniela Seixas Dept. of Imaging, Centro Hospitalar de Vila Nova de Gaia Faculty of Medicine of Porto University Portugal
• The accuracy, reliability and limitations of neuroimaging depend on the purpose of its use • • • • •
Neuroscience research Medical Marketing Law Military
Structural vs Functional brain imaging • Structural brain imaging – – – – –
Anatomy Volume Area Thickness Spatial resolution • 1x1x1 mm3
• Functional brain imaging – Brain function • • • •
fMRI Perfusion imaging Diffusion imaging Molecular imaging – MR spectroscopy – SPECT – PET
• Spatial and temporal resolution
Structural imaging
Structural imaging • Knowledge of – Techniques – Neuroanatomy – Pathology • Meaning of disease?
Technical knowledge
Meaning of disease • Newborn male diagnosed with a brain tumour before birth with fetal MRI
Meaning of disease • Should pregnancy be interrupted? – What is the effect of the disease on the brain development and function?
We don’t know!
Born 1932
Born 1932
Who has dementia?
Structural imaging • Incidental findings – A brain aneurism is found incidentally in a non clinical scan • High risk of rupture • The patient has surgery:
Survives with brain damage
Survives
Dies
• Informed consent • Should all scans be read by a radiologist? – Costs?
Functional imaging
BOLD BOLD fMRI fMRI Neuronal Neuronal Acoplamento Acoplamento activity activity neurovascular neurovascular
Hemodynamic Hemodynamic response response
Stimulus Stimulus
Sensory
Cognitive
Motor …
Indirect measure of brain activity
Limitations and validity of fMRI • Theory behind the study & nature of the research question • Design and logistics of experiment • Reliability of task • Subject’s collaboration – Head movement – Task performance
• Field strength of the MRI scanner (and other hardware)
• Due to the nature of research questions, planning fMRI tasks and logistics can become challenging
“Castle” design; each sequential block represents increasing bladder water volume and correspondent visceral sensations
Positive linear trend of brain areas active during filling of the bladder
Hyper scanning • Allows subjects in different fMRI scanners to interact in real time – Hardware and software that link scanners through the internet – Behavioral experiments in which participants interact with each other while fMRI is acquired in synchrony in both scanners (neuroeconomics)
Montague R, et al. Neuroimage 2002;16:1159-64
Subject collaboration • Paraplegic patient (unable to move both legs) due to spine trauma – Investigate location of cortical motor areas for right inferior limb – Patient could not collaborate in motor task • Passive movement of right inferior limb
Limitations and validity of fMRI • Number of subjects • Software – Registration
• Statistics – – – – – –
Smoothing Type of test False positives & false negatives Thresholding Region-of-interest analysis Inferences to the population
Number of subjects Clinical setting Forensic use Marketing
Research Marketing
Single subject
Group
Seixas D, Ayres M, Andersson J. Neuroradiology 2009;51(S1):S93-94
The effect of the number of participants
Increasing spatial extent of clusters with increasing n
Thresholding • Several techniques can be used – Bonferroni correction – Permutation thresholding – Random field theory – False discovery rate
Designed to control for type 1 error ‘To make sure that we don’t say there is an activation when there isn’t one
Type 2 error is important when n=1! ‘To make sure that we don’t say there isn’t an activation when in fact there is one’ Seixas D, Ayres M, Andersson J. Neuroradiology 2009;51(S1):S93-94
• Prediction techniques – Multivariate analysis – Exploratory analysis
MIT Sloan Management Review 2009;50:32-34
Independent component analysis
Visual
Independent component analysis
Cognitive
Independent component analysis
Auditory & sensoriomotor
Limitations and validity of fMRI • Interpretation of data – The activation of a certain brain area does not mean it is essential to the process, but that is is part of it – Culture – Context – Beliefs
• No standardised tasks or normative data – What is normal or abnormal?
• Secondary use of brain images
• Quality of brain imaging research depends on the evolution of the field – Vision AND fMRI • 3478 results
– Taste AND fMRI • 261 results
– Marketing AND fMRI (and yes, Marketing is a MeSH word in PubMed) • 71 results
To think • Brain imaging is already being used in numerous fields, from Neuroscience to Marketing or Military • It is an ongoing process that cannot be stopped but should instead be supported by those who know its technical, ethical and legal limits
Seixas D, Ayres Basto M. Clin Neuroradiol 2008;18:79-87
Thank you!
[email protected] www.danielaseixas.pt
References • Andersson J, Hutton C, Ashburner J, Turner R, Frinston K (2001). Modeling geometric deformations in EPI time series. Neuroimage 13:903-19 • Kirschen MP, Jaworska A, lles J (2006). Subjects’ expectations in neuroimaging research. J Magn Reson Imaging 23:205-209 • Nichols TE, Holmes AP (2002). Nonparametric permutation tests for functional neuroimaging: a primer with examples. Hum Brain Mapp 15:125 • Seixas D, Ayres Basto M (2008). Ethics in fMRI studies: A review of the EMBASE and MEDLINE literature. Clin Neuroradiol 18:79-87 • Steinke EE (2004). Research ethics, informed consent, and participant recruitment. Clin Nurse Spec 18:88-95 • Zijlmans M, Buskens E, Hersevoort M, Huiskamp G, van Huffelen AC, Leijten FS (2008). Should we reconsider epilepsy surgery? The motivation of patients once rejected. Seizure 17:374-377