- Atlas
- Natural Sciences
- Neuroimaging
Neuroimaging
Neuroimaging is the set of techniques that let us see the living brain—its structure, its chemistry, and its activity—without opening the skull.
16 frameworks, from 1918 to today · Neuroscience · Natural Sciences
- Pneumoencephalography and Ventriculography 1918-1975
- Cerebral Angiography 1927-1985
- Electroencephalography 1929-Present
- Electrophysiological and Magnetoencephalographic Imaging 1930-Present
- Autoradiography and Metabolic Tracer Mapping 1955-1985
- Magnetoencephalography 1968-Present
- CT scan 1972-Present
- Positron Emission Tomography 1975-Present
- Magnetic Resonance Imaging 1977-Present
- Calcium Imaging and Optogenetics 1990-Present
- Functional near-infrared spectroscopy 1990-Present
- Statistical Parametric Mapping 1991-Present
- Blood-oxygenation-level–dependent imaging 1992-Present
- Diffusion MRI and Tractography 1994-Present
- Multimodal Neuroimaging 2000-Present
- Brain Network Analysis 2005-Present
Overview
Neuroimaging is the set of techniques that let us see the living brain—its structure, its chemistry, and its activity—without opening the skull. What is at stake is direct: for most of history, knowledge of the human brain came from dissection of the dead and from the misfortune of the injured. Neuroimaging turned the working brain into an observable object. It made it possible to ask where in the brain a mental operation happens, how regions connect, and how activity changes with disease, development, and learning. The history below follows the instruments and the ideas they carried, from X-ray shadows of the skull to the maps of brain networks made today.
Shadows of the Skull: Contrast Agents and the First Living Images
The first step was not seeing the brain but seeing through it. Wilhelm Röntgen's discovery of X-rays in 1895 produced the first images of the living human interior, but the skull is dense and the brain soft, so plain radiographs showed bone and little else. The solution was to introduce something that absorbs X-rays differently from tissue.
In 1918 the American neurosurgeon Walter Dandy injected air into the ventricles of the brain—the fluid-filled chambers—and X-rayed the head1. Air is far less dense than tissue, so the ventricles appeared as dark outlines. This was pneumoencephalography (air injected into the spinal fluid spaces) and ventriculography1. It was painful and risky, but for decades it was the only way to see the brain's internal shape in a living patient, and it could reveal a tumor by displacing the ventricles.
A parallel line came from angiography. In 1927 the Portuguese neurologist Egas Moniz injected a contrast agent into the carotid artery and X-rayed the head, making the brain's blood vessels visible. Cerebral angiography showed aneurysms, malformations, and the displacement of vessels by masses. Both techniques were invasive and uncomfortable, but they established a principle that would recur: make the invisible visible by introducing a signal the body does not produce on its own.
Listening to the Brain's Electricity
A different tradition asked not what the brain looks like but what it does. In 1924 the German psychiatrist Hans Berger recorded electrical potentials from the human scalp and found a rhythmic oscillation—the alpha rhythm—that changed with eyes open and closed. He published the electroencephalogram (EEG) in 19292. EEG measures the summed electrical activity of large populations of neurons, and it gave clinicians a way to detect epilepsy and sleep stages and gave researchers a window on brain dynamics, though with poor spatial resolution because the skull blurs the signal.
Magnetoencephalography (MEG) came much later, in the late 1960s, when physicists led by David Cohen showed that the tiny magnetic fields generated by neural currents could be detected with sensitive superconducting sensors. MEG offered better spatial localization than EEG and, like it, millisecond timing. Both remain the tools of choice when the question is when, not just where1.
Mapping Metabolism: Radiotracers and Autoradiography
The next conceptual leap was to image not anatomy or electricity but chemistry—which parts of the brain are using energy. In the 1950s and 1960s, researchers developed radioactive tracers that cells take up in proportion to their metabolic activity. A landmark was the 2-deoxyglucose method, developed by Louis Sokoloff and colleagues in the 1970s: a glucose analog labeled with a radioisotope is taken up by active neurons but not fully metabolized, so it accumulates where energy demand is high. In animal studies, autoradiography—placing brain slices against film—produced detailed maps of activity.
This work established the metabolic logic that functional imaging would later exploit: active brain regions consume more glucose and oxygen and receive more blood. It also introduced the idea of cognitive subtraction, which would become central: if you image the brain during task A and during task A plus an added mental operation, the difference isolates the added operation.
Computed Tomography: Reconstructing a Slice
The decisive break came from mathematics and computing. In 1972 Godfrey Hounsfield, working at EMI, and independently Allan Cormack, showed that a series of X-ray measurements taken from many angles around the head could be combined by computer into a cross-sectional image—a slice through the body. This was computed tomography (CT)1. Hounsfield's first clinical scans of the brain revealed tumors and hemorrhages with a clarity that plain X-rays never approached, and CT quickly displaced pneumoencephalography. The key idea—reconstructing an image from projections—was general, and it opened the door to every subsequent tomographic method.
Positron Emission Tomography: Imaging Function in Humans
CT showed structure. The question of function in the living human brain was answered by positron emission tomography (PET). PET detects pairs of gamma rays emitted when a positron—a positively charged electron—annihilates with an electron in the tissue3. By labeling a molecule with a positron-emitting isotope, researchers could trace its distribution1. In the late 1970s and 1980s, groups led by Michel Ter-Pogossian, Michael Phelps, and others built PET systems, and researchers such as Marcus Raichle used labeled water and oxygen to measure regional blood flow and metabolism in the human brain during tasks.
The 1980s brought a wave of PET studies mapping language, memory, and attention. The method's spatial resolution was modest and its temporal resolution poor, and it required a cyclotron and radioactive tracers, but it was the first way to watch a healthy human brain perform a mental task. PET also enabled molecular imaging—receptor maps, dopamine studies—that remains one of its strengths1.
Magnetic Resonance Imaging: From Nuclei to Anatomy
Magnetic resonance imaging (MRI) grew from nuclear magnetic resonance, a physics technique for studying atomic nuclei in magnetic fields4. In 1973 Paul Lauterbur showed that magnetic field gradients could encode spatial position into the NMR signal, producing an image; Peter Mansfield developed the mathematical and practical methods that made fast, efficient imaging possible. The first human MRI images followed in the late 1970s5.
MRI's advantage over CT was contrast: different tissues—gray matter, white matter, cerebrospinal fluid—could be distinguished by tuning the pulse sequence, and there was no ionizing radiation4. By the 1990s MRI had become the standard structural imaging method in neuroscience and medicine, revealing anatomy with a precision that dissection could only approximate in the living.
Blood-Oxygen-Level-Dependent fMRI: Watching Activity
The most consequential development for cognitive neuroscience came from an observation about blood. In 1990 Seiji Ogawa and colleagues showed that deoxygenated hemoglobin distorts the magnetic signal, so changes in blood oxygenation produce image contrast678. This is the blood-oxygen-level-dependent (BOLD) signal68. Because active brain regions receive more blood than they consume oxygen, local oxygenation rises with activity7.
In 1992, several groups—including those of Kenneth Kwong and colleagues at Massachusetts General Hospital and Ogawa's collaborators—demonstrated that BOLD contrast could map brain activity during tasks in humans. Functional MRI (fMRI) was born1. It required no tracer injection, offered millimeter spatial resolution, and could be repeated in the same subject17. Within a decade fMRI had transformed cognitive neuroscience, producing thousands of studies mapping perception, language, memory, and emotion.
The method's limitations shaped its history too: the BOLD signal is slow (seconds), indirect (it measures blood, not neurons), and noisy7. These constraints drove the development of statistical methods7. Statistical parametric mapping (SPM), developed by Karl Friston and colleagues from the early 1990s, provided a framework for comparing images voxel by voxel—each small volume element—and correcting for the thousands of statistical tests involved. Cognitive subtraction, inherited from the metabolic tradition, remained the workhorse design, though it was increasingly criticized for assuming that adding a task adds only the operation of interest.
Diffusion MRI and Tractography: Mapping Connections
Structural MRI showed where tissue was; diffusion MRI showed how water moves through it. In the mid-1990s, researchers including Peter Basser and colleagues showed that the directionality of water diffusion reveals the orientation of white-matter fibers, because water diffuses more freely along axons than across them. This is diffusion tensor imaging9. From these measurements, tractography algorithms reconstruct the pathways connecting brain regions—a noninvasive approximation of the wiring diagram109. Tractography became central to studies of development, aging, and disconnection in disease, though its reconstructions are indirect and debated.
Near-Infrared Spectroscopy: A Portable Window
A quieter line of development used light. Near-infrared spectroscopy (NIRS) shines near-infrared light through the scalp and measures how much is absorbed by oxygenated and deoxygenated hemoglobin, exploiting the fact that these molecules absorb light differently111. Because near-infrared light penetrates bone and tissue a few centimeters, NIRS can monitor cortical blood oxygenation with portable, inexpensive equipment. It lacks the spatial resolution of fMRI and reaches mainly the cortex, but it brought functional imaging to infants, bedside patients, and settings where a scanner cannot go.
Cellular Imaging and Optogenetics: Watching Neurons Themselves
Meanwhile, a different scale of imaging was advancing. Calcium imaging uses fluorescent indicators that light up when neurons fire, letting researchers watch hundreds of individual cells in awake animals. Optogenetics, developed in the mid-2000s by Karl Deisseroth and colleagues, made neurons controllable with light by inserting light-sensitive proteins—channelrhodopsin and others—into their membranes. Combined with imaging, it allowed causal tests: activate a specific cell population and observe the effect on activity and behavior. Two-photon microscopy, developed in 1990 by Winfried Denk, James Strickler, and Watt Webb, made deep-tissue imaging of individual neurons possible. This cellular tradition is distinct from human neuroimaging but increasingly informs it, as the BOLD signal is interpreted against what cellular imaging reveals about the underlying neural activity.
The Present: Networks, Multimodal Imaging, and Open Questions
Today neuroimaging is plural. Structural MRI and CT remain the clinical backbone. fMRI dominates human cognitive neuroscience, increasingly analyzed in terms of brain networks—correlated activity across distributed regions—rather than isolated spots. Diffusion imaging maps connections; PET provides molecular specificity; EEG and MEG supply timing; NIRS extends functional imaging beyond the scanner. Multimodal studies combine these to compensate for each method's limits.
The open questions are the ones the history has produced. What does the BOLD signal actually reflect about neural activity? How reliable are tractography reconstructions10? Can imaging support causal claims about brain-behavior relationships, or only correlations? And the oldest question remains: whether measuring blood, electricity, and chemistry in living tissue can explain how the brain makes a mind.
References
- Neuroimaging (History, Magnetoencephalography (MEG) and Electroencephalography (EEG), Positron emission tomography, Functional magnetic resonance imaging, Diffuse optical imaging), Wikipedia
- Default mode network (History), Wikipedia
- Positron emission tomography, Wikipedia
- Magnetic resonance imaging, Wikipedia
- Magnetic resonance imaging of the brain (History), Wikipedia
- Seiji Ogawa, Wikipedia
- Functional magnetic resonance imaging (introduction, History), Wikipedia
- Blood-oxygenation-level–dependent imaging (History, introduction), Wikipedia
- Diffusion-weighted magnetic resonance imaging (introduction, Diffusion tensor imaging), Wikipedia
- Tractography, Wikipedia
- Functional near-infrared spectroscopy, Wikipedia
The frameworks of Neuroimaging
Each framework is a way of working the field has used: when it held, and the question it answers.
- 1918-1975
Pneumoencephalography and Ventriculography
What shape do the brain's fluid spaces take?
- 1927-1985
- 1929-Present
- 1930-Present
Electrophysiological and Magnetoencephalographic Imaging
When do neural currents flow in the brain?
- 1955-1985
Autoradiography and Metabolic Tracer Mapping
Which regions consume the most energy?
- 1968-Present
Magnetoencephalography
Where do neural currents generate magnetic fields?
- 1972-Present
- 1975-Present
Positron Emission Tomography
Which brain regions are active during a task?
- 1977-Present
Magnetic Resonance Imaging
How can tissue contrast be imaged without radiation?
- 1990-Present
- 1990-Present
Functional near-infrared spectroscopy
How does blood oxygenation change with activity?
- 1991-Present
Statistical Parametric Mapping
Which voxels differ significantly between conditions?
- 1992-Present
Blood-oxygenation-level–dependent imaging
Where does blood oxygenation track mental activity?
- 1994-Present
Diffusion MRI and Tractography
How are white-matter fibers oriented and connected?
- 2000-Present
- 2005-Present
How the ideas connect
Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.
CT scan replaced Pneumoencephalography and Ventriculography
CT scans gave a non-invasive cross-sectional view of the brain's interior that displaced pneumoencephalography and ventriculography, which had required injecting air into the spinal fluid spaces. Within a few years of Hounsfield's 1972 clinical scans, the painful air studies were largely abandoned.
Carried by Godfrey Hounsfield.
Positron Emission Tomography influenced by Autoradiography and Metabolic Tracer Mapping
The 2-deoxyglucose autoradiography work of Sokoloff and colleagues established that active brain regions take up more glucose and receive more blood, the metabolic logic PET would exploit in humans. It also introduced cognitive subtraction, which PET studies adopted as their standard design.
Carried by Louis Sokoloff.
Blood-oxygenation-level–dependent imaging influenced by Positron Emission Tomography
PET's demonstration that regional blood flow and metabolism track mental activity supplied the physiological rationale for BOLD fMRI, and its cognitive-subtraction designs were carried over directly. fMRI replaced the radioactive tracer with an endogenous blood-oxygen signal.
Carried by Marcus Raichle; Seiji Ogawa.
Blood-oxygenation-level–dependent imaging grew out of Magnetic Resonance Imaging
BOLD fMRI is a direct outgrowth of MRI: Ogawa's 1990 observation that deoxygenated hemoglobin alters the magnetic signal was turned into a functional imaging method by Kwong and others in 1992. It uses the same scanners and pulse sequences as structural MRI.
Carried by Seiji Ogawa; Kenneth Kwong.
Diffusion MRI and Tractography grew out of Magnetic Resonance Imaging
Diffusion MRI is a branch of MRI that encodes the direction of water motion using magnetic field gradients, building on the same hardware and reconstruction methods. Basser and colleagues showed in the mid-1990s that this diffusion anisotropy reveals white-matter fiber orientation.
Carried by Peter Basser.
CT scan replaced Cerebral Angiography
CT provided a direct, non-invasive image of brain structure and hemorrhage that made the invasive carotid puncture of cerebral angiography unnecessary for many diagnostic questions. Angiography survived mainly for vascular detail that CT could not show.
Carried by Godfrey Hounsfield.
Magnetoencephalography grew out of Electroencephalography
MEG grew out of the electrophysiological program of measuring brain activity non-invasively from outside the head, replacing scalp electrodes with superconducting magnetic sensors. It answered the same timing questions as EEG with improved source localization.
Carried by David Cohen.
Magnetoencephalography grew out of Electrophysiological and Magnetoencephalographic Imaging
MEG is the magnetic counterpart of electrophysiological imaging, measuring the fields generated by the same neural currents that EEG records electrically. Cohen's work turned the general program of non-invasive electromagnetic recording into a magnetic method.
Carried by David Cohen.
Positron Emission Tomography influenced by CT scan
PET borrowed the tomographic reconstruction principle that CT had made practical, combining projections from many angles into cross-sectional images. The same computing approach was applied to positron annihilation data to map function rather than structure.
Magnetic Resonance Imaging influenced by CT scan
MRI adopted the tomographic idea of reconstructing a slice from measurements taken under varying spatial encoding, which CT had established as the general route to cross-sectional imaging. Lauterbur's field-gradient encoding was a magnetic analogue of CT's projection reconstruction.
Carried by Paul Lauterbur.
Brain Network Analysis influenced by Blood-oxygenation-level–dependent imaging
Resting-state and task fMRI time series provided the functional connectivity estimates from which brain networks were constructed. Correlation of BOLD signals across regions became the basis for describing large-scale networks.
Brain Network Analysis influenced by Diffusion MRI and Tractography
Tractography supplied the structural connectivity matrices that brain network analysis used to build graphs of regions and fiber pathways. Diffusion-derived wiring diagrams became a standard input to network studies of the brain.
Multimodal Neuroimaging influenced by Statistical Parametric Mapping
Multimodal neuroimaging combined functional maps with structural and diffusion data, and the statistical frameworks developed for SPM provided the common voxel space and inference machinery for integrating them. The practice grew from the need to align and jointly analyze images from different modalities.
Carried by Karl Friston.
Beyond the field
Where Neuroimaging's ideas came from in other fields, and what they led to there: the strongest 12 of 76 links.
Cerebral Angiography influenced Aneurysm clipping in Neurosurgery
Cerebral angiography, developed by Moniz from 1927, made it possible to locate an intracranial aneurysm before operation and so made clipping feasible. The diagnostic technique is the precondition for aneurysm surgery.
Carried by Egas Moniz.
Cerebral Angiography influenced Endovascular Neurosurgery in Neurosurgery
Cerebral angiography, introduced by Egas Moniz in 1927, was the diagnostic and access technique from which endovascular neurosurgery developed its catheter-based treatment of intracranial vascular lesions.
Carried by Egas Moniz.
Electroencephalography influenced Brain–computer interface in Neural Engineering
Electroencephalography supplied the non-invasive signal, the scalp electrode montage and the rhythmic features (notably sensorimotor rhythms) that the first brain-computer interfaces used to drive output. The BCI field's early demonstrations decoded EEG recorded from the human scalp.
Carried by Hans Berger; Jacques Vidal.
Electroencephalography influenced Epilepsy Genetics in Epilepsy
Berger's electroencephalography gave epilepsy genetics its key heritable phenotype: the epileptiform and dysrhythmic EEG patterns traced in families and twins by Lennox and others. EEG recording was the instrument that made genetic study of epilepsy possible.
Carried by Hans Berger; William Lennox.
Electrophysiological and Magnetoencephalographic Imaging influenced by Electroencephalography in Epilepsy
Clinical EEG provided the electrode montages, waveform interpretation and evoked-response paradigms that electrophysiological and magnetoencephalographic imaging extended to source localization.
Carried by Hans Berger.
Electrophysiological and Magnetoencephalographic Imaging influenced Neuroimaging of language in Psycholinguistics
Neuroimaging of language built directly on the electrophysiological tradition, especially event-related potentials, which supplied the millisecond-resolution methods and the component logic later combined with haemodynamic imaging. The ERP work on semantic anomalies carried into the neuroimaging programme.
Carried by Marta Kutas; Steven Hillyard.
Autoradiography and Metabolic Tracer Mapping influenced Metabolic Imaging with FDG in Molecular Imaging
FDG imaging directly translated the 2-deoxyglucose autoradiographic method for mapping local glucose metabolism in animals to the living human with a positron-labelled analogue. The Sokoloff method supplied the concept and the kinetic model.
Carried by Louis Sokoloff.
Autoradiography and Metabolic Tracer Mapping influenced by Tracer Principle in Molecular Imaging
The tracer principle, established by Hevesy's use of radioactive indicators, is the direct basis of autoradiographic metabolic mapping: a labeled substance traces its own distribution and fate in tissue.
Carried by George de Hevesy.
Autoradiography and Metabolic Tracer Mapping influenced Structural and Functional Neuroimaging in Epilepsy
Autoradiographic and metabolic tracer mapping, notably the 2-deoxyglucose method, established that regional brain metabolism could be measured and mapped, which directly inspired PET and later functional imaging. The tracer and metabolic-mapping approach passed into neuroimaging.
Carried by Louis Sokoloff.
Magnetoencephalography influenced by Electroencephalography in Epilepsy
MEG was built to measure the magnetic counterpart of the EEG signal; the known EEG phenomena and their generators set the target for the first SQUID-based magnetic recordings.
Carried by David Cohen.
CT scan influenced Three-Dimensional Conformal Radiotherapy in Radiation Oncology
Computed tomography, introduced in the early 1970s, gave radiotherapy the three-dimensional anatomical dataset needed to delineate target volumes and design conformal beams; conformal therapy is unthinkable without it.
Carried by Godfrey Hounsfield.
CT scan influenced by Tomographic reconstruction in Imaging Physics
CT scanning depends on the algebraic reconstruction techniques developed by Cormack and others; Hounsfield's scanner was built around reconstructing attenuation values from projection data.
Carried by Godfrey Hounsfield; Allan Cormack.

Neuroimaging is a planet in Nousfarer
Its city has a tower for each of these 16 frameworks. Land, walk in, and climb one concept at a time.
Land on NeuroimagingRelated fields
Fields its ideas are linked to
- Molecular Imaging 15 links
- Epilepsy 9 links
- Medical Imaging 5 links
- Imaging Physics 4 links
- Neurosurgery 4 links
- Bioelectronics 3 links
- Neurolinguistics 3 links
- Quantitative Imaging 3 links
- Behavioral Neuroscience 3 links
- Cardiology 2 links
- Cognitive Neuroscience 2 links
- Systems Neuroscience 2 links