Neuroimaging Notes PDF

Neuroimaging notes L1: Introduction to Neuroimaging Explain what neuroimaging is, and what it's not. Explain how neuroimaging techniques are classified. Give examples of neuroimaging research. Understand the importance of philosophy of mind for studying mind-brain relation...

Neuroimaging notes L1: Introduction to Neuroimaging Explain what neuroimaging is, and what it's not. Explain how neuroimaging techniques are classified. Give examples of neuroimaging research. Understand the importance of philosophy of mind for studying mind-brain relationships. Explain forward and reverse inferences. Understand what is expected of you in this module. Experiments all start with a research question Appropriate experiments are designed to address this question Neuroimaging is a way of addressing this research question What is neuroimaging? Use of techniques that portray the structure and function of the human brain Imaging techniques measure different variables – not a way to assign locations to specific functions in the brain (e.g. fear in the amygdala, social processing in the prefrontal cortex etc) How are neuroimaging techniques classified? Structure vs function Structural: analysis of anatomical/locational properties of the brain Morphological Localisation is not the primary goal of imaging – aids with developing theories about how the brain works as a system Functional: identify areas and processes associated with performing 1a cognitive/behavioural task ‘Imaging the brain during psychological activity’ Temporal vs spatial resolution Spatial resolution: ability to distinguish between different locations in the brain High spatial resolution instrument would differentiate regions in the brain close together Temporal resolution: ability to measure changes in brain activity over time / How process occur in the brain over time Refers the smallest unit of time that can be differentiated by a method Invasiveness Methods are either fully invasive (entering the skull), or non-invasive (most neuroimaging techniques!) Correlation vs causation Philosophy of mind Philosophical theory that deals with the mind and mental states, and the relationship between the physical and mental Brain vs Mind: How are these related to one another? Dualism: mind and brain are separate entities Monism: mind and brain are the same Physicalism: mental experiences are the product of neurons and synapses in the brain, the metaphysical view that all mental processes are ultimately physical processes or necessitated by physical phenomena. Reductive physicalism: mental states can be reduced to physical states. When you fully understand the brain, you also understand the mind Non-reductive physicalism: mental and physical states are related but they are not the same Cognitive neuroscience Assumes that there are 1-to-1 mapping between brain patterns and cognitive functions Types of inferences Inferences are steps in reasoning to reach a conclusion based on information Forward Inferences ‘Using qualitatively different patterns of activity across the brain to distinguish between competing cognitive theories’  Assessing the similarities and differences in the brain across different conditions E.g. an autobiographical memory task and a recognition memory task Autobiographical memory task: activated default mode brain network Different areas – there are qualitative differences meaning not all memories are the same Reverse Inferences When you infer a particular cognitive function from a brain pattern See this pattern, therefore this previously described function is responsible  N400 in semantic processing (previously described) Seeing the brain pattern N400 in schizophrenic patients Infer that schizophrenic patients have deficits in semantic processing One-to-one mapping Assuming that one specific cognitive function may cause one specific brain pattern, and vice versa This is not always an assumption, especially if we consider reductive physicalism (mental processes can be reduced to a physical state) – this means that L2: Overview of methods You will learn about structural and functional magnetic resonance imaging, optical imaging, and electromagnetic imaging. Describe the basic workings of each technique, including what brain activity is measured. Describe the pros and cons of each technique.  Understand how each technique can be used to study mind-brain questions. Choose an appropriate technique to address a question of interest. Specific outcomes Explain what is meant by structural and functional imaging Describe the basic workings of MRI, fMRI, PET, fNIRS, EEG and MEG, including what brain response is measured. Describe the pros and cons of each technique. Understand how each technique can be used to study mind-brain questions. Choose/Justify an appropriate technique to address a question of interest (e.g., which technique is best to assess whether autistic children process the meaning of words). Structural imaging Investigates anatomy of the brain Detailed, static anatomical brain pictures Looks at morphology: the size, shape, density of grey and white matter Why do we need it? Crucial for later understanding the function of the brain Behavioural / observed psychological difference may be traced to subtle differences in neuroanatomy Types: X-rays, Computerised Tomography (CT) and MRI scans Structural MRI Neuroanatomical structures shown with clarity and precision Used to look or check for abnormal anatomy Produces T1-weighted or T2-weighted images T1W: describes the spin relaxation or rate of decay of applied magnetisation of excited protons of water molecules in the brain Provide good contrast between grey and white matter, and csf T2W: provides contrast between CSF (bright) and brain matter (dark) How does MRI work? Nuclear magnetism physics  Protons and neutrons have a quantum mechanical property of spin Spin is a form of angular momentum If there is an odd number of protons/neutrons, the spins do not cancel out The nucleus has a non-zero ‘magnetic momentum’ When this atom is placed in a magnetic field, the magnetic momentum aligns in the direction of the field Nuclei with odd spin, spin at a specific frequency, called a Larmor frequency When the magnetic field oscillates, the nuclei absorb energy from the field if the oscillation frequency is the same as the Larmor frequency This is called ‘resonance frequency’ The oscillation frequency is referred to a radio frequency (RF) pulse The oscillating radio frequency causes the H1 spin to go in ‘phase’ – the same position Spin direction changes and is flipped in the direction of the oscillating field + H1 nucleus absorbs energy When the RF is removed, these effects disappear (dephasing & realigning) Every time a nucleus realigns, it emits a small energy in the RF range B0: the  constant, static magnetic field use to polarise spins to create magnetisation. The direction of B0 is the longitudinal axis The B0 is the magnetic field that aligns the protons B1: RF energy field that is applied perpendicular to the longitudinal axis (B0) Physics of MRI MRI technique is based on the magnetisation properties of atomic nuclei Hydrogen atom is mainly used as it is abundant in the body, in water and fat An external, powerful magnetic field (from the MRI machine) aligns the protons that are normally randomly orders – this process is ‘magnetisation’ They protons will either be parallel or antiparallel to the direction of the magnetic field (B0) – the parallel state is the lower energy state Some of the low energy protons will change into the high energy state An external radio frequency (RF) is introduced from the scanner that disturbs the magnetisation The nuclei subatomic particles return to their resting (normal) alignment and emit RF energy that is measured The energy associated with realigning to the normal state is used to make an image Fourier transformation is used to convert the frequency information from the signal for each location in the imaged plane  These are displayed as grey pixels (which make up the images we see in MRIs!) Using multiple transmitted RF pulses in succession can show different tissues Different tissues can be emitted because they emit different energies Repetition time (TR): amount of time between consecutive pulse sequences Spin Echo: refocussing of spin Time to Echo (TE): Time measures Relaxation time: time taken for protons to return to their original state There are 2 relaxation times: T1 and T2 T1: (spin-lattice relaxation): time for protons to re-align to the static magnetic field (B0). T2 (spin-spin relaxation): time for protons to de-phase due to interactions between spins. Always quicker than T1 T2*: rapid loss in coherence of protons and faster de-phasing because of local magnetic field inhomogeneities (T2* < T2). T1 and T2: are for structural imaging T2* is for functional imaging  MRI sequences Diffusion tensor imaging (DTI) Visualises: white matter bundles/axons in the brain / observing impaired white matter integrity in diseases e.g. AD, MS Measures: strength and direction of water molecule diffusion on nerve fibres How does it work? Water molecules diffuse at different rates along nerve fibre tracts Where white matter is impaired/degraded, water diffusion is altered Measure this using fractional anisotropy (0 to 1 scale, where 1 is fully integral) Diffusion weighted imaging (DWI) Visualises: early pathological changes in tumours, vascular/ischaemic strokes in the brain, non-small cell lung cancers Measures: rate of diffusion of (water) molecules (e.g., for stroke). Arterial spin labelling (ASL) aka perfusion weighted imaging (PWI) Visualises:  Measures: tissue perfusion or cerebral blood flow MRI machine administers RF and gradient magnetisation to magnetise water molecules in blood to measure perfusion Strengths and weakness of structural techniques Functional Imaging Looking at brain function Based on physiological change Haemodynamic methods Haemodynamic: dynamics of blood flow through structures in the brain Functional imaging relies on the brain’s blood supply Relationship between neural activity and blood supply Blood circulation is responsible for providing neuronal tissue with energy Neurons require a continuous supply of oxygen and glucose – blood enters the brain through arteries Oxygen molecules are removed from haemoglobin where it becomes deoxyhaemoglobin A local increase in energy consumption – blood oxygenation decreases  Immediately after neuronal activity, there is local oxygen consumption – decrease in blood oxygenation, and decrease in the measured signal General idea Brain is metabolically demanding When a region on the brain is active, it uses more glucose and oxygen Blood volume and oxygenation would increase More blood is supplied than used – we can measure the flow and oxygenation levels to find which regions are activated Positron emission tomography (PET) Used for: in vivo imaging of biological, physiological and biochemical processes Uses radioactive tracers which are injected into the body The tracers are attached to molecules with a specific biological action The tracer will spread to a broader areas The tracers have a short half-life – will transform into a more common non- radioactive form in a process called 'positron emission decay’ The positive positron is the antiparticle of the negatively charged electron – they get attracted to each other This results in annihilation – photons are produced which travel in opposite directions Gamma rays are released in either direction at the time of annihilation More blood means more gamma rays Two photons are detected at the same time (coincidence detection) Thousands of coincidences are detected and localised which allow for a reconstruction of an image Signals are visualised as an intensity map where regions with abnormal deposition and metabolism can be seen Negatives of PET Not good at structural images – lower spatial resolution Tracer molecules have a short half-life – might be difficult Small risk of radiation Functional MRI (fMRI) Why use fMRI? Real-time depiction of brain activity during a task or stimulus Physics Blood oxygenation level dependent (bold) fMRI Oxygenation is important for fMRI signals because deoxyhaemoglobin is paramagnetic, which oxyhaemoglobin isn’t The paramagnetic nature of deoxyHb means spin-spin interactions are altered – faster T2 decay An increase in oxygenation causes an increased fMRI signal after a RF is added in a sequence that is sensitive to T2 decay Field inhomogeneity Even if the scanner creates a perfectly uniform magnetic field, placing biological tissue in the magnetic field causes small spatial variations  Different nuclei have difference field strengths, will spin at different frequencies and result in differences in phase and dephasing T2* decay Always faster than T2 decay – meaning there is more dephasing Measuring the BOLD contrast Pulse signals that are T2* weighed give more of a signal when blood is oxygenated Depends on deoxyhb – distorts magnetic field Molecules of deoxyhb align with the applied magnetic field of the MRI machine Concentration of deoxyhb is measured, resulting in the BOLD signal Method Usually involves a paradigm (task-evoked) that causes changes in cerebral blood flow Oxygen gets delivered to neurons through haemoglobin When neuronal activity increases, there is a local increase in demand for oxygen Activated brain regions experience a greater degree of change in deoxyHb concentration (more oxygen is used up)/ more deoxyHb is present Since deoxyHb is paramagnetic but Haemoglobin is magnetic, the difference in magnetic properties can be measured There is a greater BOLD signal relative to the surrounding tissues BOLD response corresponds to a local, increase in blood flow and volume Local response is to increase the blood flow to regions of increased neural activity Remember, the BOLD response is an indirect measure of neuronal activity The ‘initial dip’ After neural activity increases, there is a small momentary decrease in blood oxygenation called the ‘initial dip’ This is followed by a period where blood flow increases more than is required – attributed to neuronal activity Peak lasts for around 6 seconds before returning to the baseline There is an ‘undershoot’ which is the post stimulus Optical neuroimaging fNIRS Optical technique that measures changes of (HbO2) and deoxygenated (HbR) haemoglobin following neuronal activation in brain tissue Shine NIR light (650–950 nm) into the head and, taking advantage of the relative transparency of the biological tissue within this NIR optical window, light will reach the brain tissue. Used to investigate neurodevelopment in kids Advantages: portability, tolerates movements, safety Electromagnetic brain activity Electrophysiological activity can be observed in the brain Map brain activity by detecting signals of ionic currents from biochemical processes occurring at cellular level during neural activation Sensor arrays record these ionic currents EEG: Electric fields MEG: Magnetic fields Oscillations: frequency, bandwidth and amplitude of electric or magnetic field signals Temporal measurements: deal with time!!! Not structure EEG and MEG: best real-time method to see EEG EEG is an electrophysiological technique for the recording of electrical activity arising from the human brain. Temporal sensitivity Used in evaluation of dynamic cerebral functioning EEG is thought to be primarily generated by cortical pyramidal neurons in the cerebral cortex that are oriented perpendicularly to the brain's surface. The neural activity detectable by the EEG is the summation of the excitatory and inhibitory postsynaptic potentials of relatively large groups of neurons firing synchronously. Conventional scalp EEG is unable to register the momentary local field potential changes arising from neuronal action potentials. The neurons involved need to also be in a specific configuration - need to be open field How to do EEG recording Electrodes are placed on the scalp (they measure the current) Record voltage potentials resulting from current flow in and around neurons  EEG recording as a graph voltages on the vertical domain and time on the horizontal domain, providing a near real-time display of ongoing cerebral activity ERPs (event-related potentials): derived from EEG, are short segments of EEG data that are time-locked to particular events of experimental interest, and typically averaged over many trials of an experiment International 10-20 system - an internationally recognised method to describe and apply the location of scalp electrodes in the context of an EEG exam Why do we use EEG? Direct and real-time measure of the brain’s neural activity Tool for studying the brain’s electrical activity e.g. in seizures May be able to characterise brain networks and connectivity Problems with EEG The neural activity that EEG measures is conducted through the brain volume – you cannot assume a signal recorded at one point in the skull has its source directly below Measuring a voltage requires a reference site, but there is nowhere on the scalp that is a perfect electrical locus Spatial Biological and environmental electrical artefacts: voltage potentials from other body parts frequently overwhelm cerebral activity MEG Measures the magnetic field generated by the electrical activity of neurons. Images are superimposed with the brain to help identify the source Detects tangential dipoles because the radial dipoles are still in the brain and can’t be detected outside MEG set up EEG vs MEG EEG MEG EEG: more sensitive to radial MEG: coils more sensitive to dipoles tangential dipoles (magnetic  field forms perpendicular to the current) Currents from other dipoles get Little mix between sources and mixed up sensor so, better at estimating source locations Less expensive More expensive due to magnetic shielding room Low spatial resolution High temporal and spatial resolution Advantage of MEG Magnetic fields pass through the head without any distortion, high spatial and temporal resolution EEG signal is bent and mixed The Inverse problem Source localisation: scalp fields do not directly reflect what part of the brain actually caused those signals The recording place may not be the place where the maximal activity is happening Inverse problem: ‘the process of calculating from a set of observations the causal factors that produced them’ Difficult to determine the site in the brain where magnetic L3: fMRI acquisition Learning objectives MRI safety Describe the practical stages of (f)MRI acquisition. Understand the dangers of (f)MRI. Understand the nature of (f)MRI data. Understand possible artefacts and how to avoid them. Choose appropriate acquisition parameters. Evaluate a typical (f)MRI acquisition Methods section. MRI safety – basically, no ferromagnetic metal in the scanner room Static magnetic field of the MRI scanner picks up ferromagnetic objects and can even pull them towards the scanner at high velocities (projectile effect) Ferromagnetic: contain iron, nickel or cobalt Potential dangers in MRI scans Static field dangers: short lived effects like dizziness, nausea Torsion: metal in the body will try and move parallel to the static magnetic field Heating - avoid body loops e.g. crossing legs or body parts touching Acoustic noise caused by gradient switching, headphones used fMRI experiments Might takes months or years to plan Pilot study - to check that the behaviour you might want to measure is going to occur, or maybe to check the stimulus can be seen Acquisition key concepts The goal of fMRI is to relate changes in brain physiology over time to an experimental manipulation A single functional image gives no information about brain activity Only by examining changes across images over time can be see if our experimental manipulation has produces an effect fMRI data is collected as a time series We look at the time series to see if there are any changes that are associated with out experimental task The participants in as study are called subjects Each subject participates in a single experimental session around 1 hour long called a session Each session involves collecting anatomical images A run of images is a series of consecutive brain images  A single run can last a few minutes or longer. Multiple runs are collected in a single session Separating a session into runs stops subjects from being tired so they can focus on the tasks Each full 3D brain image acquired is called a volume Every run collects functional data that is a time series of volumes There can be hundreds of volumes collected in a single 10-minute run The first few volumes acquired are discarded as the net magnetisation has not yet become stable Each volume consists of slices which are designed to cover the whole brain Slices get built up to reconstruct the 3D volume of the brain Each slice is acquired at a different point in the time series (E.g. slice 1 and slice 2 at different times), but all the data points within one slice are collected at the same time  Each slice is made from thousands of voxels (tiny cubes) that are collectively make an image of the brain Usually, slices consist of a 2n x 2n matrix of voxels E.g. 64 x 64 Glossary What is an fMRI image? 3D matrix of data that is divided into voxels Each voxel has a value showing the signal at a specific point in space and time Slice acquisition 2 main methods: sequential or interleaved Sequential: adjacent slices are collected one after the other (consecutively) either from top-to-bottom or bottom-to-top E.g. slices 1, 2 3 ….12 However, boundaries between sequential slices can be blurred sometimes Why use sequential: less distortion if there is movement- different slice times - activity is measured at different times since they are one after the other Interleaved slice acquisition acquires every other slice, and then fills in the gaps on the second pass. E.g. scan from 1 to 4 upwards then 5 to 8 downwards - less interference between slices/ prevents cross-slice excitation Adjacent parts of the brain are acquired at different times – Pixels and voxels and spatial resolution Spatial resolution: ability to distinguish different locations in an image Voxels are the basic unit of MR images When you decrease voxel size, the ability to see finer structures in the brain improves Voxels in fMRI are around 3mm in length Effect of voxel size Larger voxels means more space is measured Field of View Distance over which a brain volume (1 image) is acquired Refers the distance (in cm or mm) over which an MR image is acquired or displayed How much of the brain is measured depends on the field of view Determined by the matrix size and thickness Each brain slice is a matrix that has a height, width and depth - is a 3D shape In-slice voxel size – FOV divided by matrix size Temporal resolution Many volumes (3D images) that takes time to acquire Lower (but okay) temporal resolution: fMRI is limited by the haemodynamic response time Typically, the BOLD response has a width of ~3s and a peak occurring ~5–6s after the onset of a brief neural stimulus Slower than the underlying neural processes, and temporal information is thereby heavily blurred Repetition time (TR) TR: the time between RF pulses that excite the same slice The amount of time that passes between consecutive acquire brain volumes. TR time: time taken to acquire a brain volume ~between 2 and 4 seconds Is the basic unit for temporal resolution EPI is an MRI acquisition method where a 2D image can be obtained by switching magnetic gradients back and forth after a single excitation pulse Echo time: Time interval between excitation and data acquisition Dummy scans: a few volumes in the beginning which are discarded as magnetisation equilibrium has not yet been reached. The data from the first few volumes can’t be used Position slice upwards - prevent artefacts from eye movements Electrical transient: e.g. if a phone is nearby Susceptibility artefacts: caused by field inhomogeneities Parts of the scan might be missing due to magnetic field distortions Answers A slice = m