An MRI scanner creates images by exploiting a simple fact about your body: it is mostly water, and every water molecule contains hydrogen atoms whose nuclei behave like tiny spinning magnets. The scanner uses a powerful magnetic field to align those hydrogen nuclei, briefly disturbs them with a pulse of radio energy, and then listens as they snap back into alignment. Different tissues snap back at different rates, and those differences in timing become the contrast you see in the final image. No ionizing radiation is involved, which is one reason MRI became the go-to tool for visualizing soft tissues like the brain, spinal cord, and joints.
The Static Magnetic Field
At the heart of every MRI scanner is a large magnet that produces a strong, steady magnetic field. Clinical scanners typically operate at 1.5 or 3 tesla, tens of thousands of times stronger than the Earth’s own magnetic field. Most high-field systems achieve this with superconducting wire coils cooled to extremely low temperatures, traditionally using liquid helium. Newer designs are moving toward helium-free conduction cooling, which can still achieve the field uniformity needed for clinical imaging at 1.5 to 3 tesla.1PubMed Central. A Narrative Review of Advancements in Magnetic Resonance Imaging (MRI) Technology: Evaluating the Shift From Helium-Cooled to Helium-Free Systems
When you lie inside the scanner, this field does something specific to the hydrogen nuclei throughout your body. Normally, the tiny magnetic moments of hydrogen protons point in random directions. Under a strong external field, they preferentially align along the field’s direction. Not all of them line up perfectly in the same orientation, but a slight statistical majority do, creating a small net magnetization in your tissues. That net magnetization is what the scanner works with. It is faint, which is why the magnetic field needs to be so strong in the first place.
The Radiofrequency Pulse
Once the hydrogen nuclei are aligned, the scanner sends a brief burst of radiofrequency energy, tuned precisely to the resonant frequency of hydrogen in that particular magnetic field strength. This frequency falls in the radio band of the electromagnetic spectrum, which is why the technique was eventually called magnetic resonance imaging rather than nuclear magnetic resonance, even though the physics is identical. The resonant frequency is determined by the strength of the magnetic field and a property of the nucleus itself. At 1.5 tesla, hydrogen resonates at about 64 megahertz; at 3 tesla, roughly 128 megahertz.
The radiofrequency pulse tips the net magnetization away from its resting alignment along the main field. Think of it like pushing a spinning top off-balance. The hydrogen nuclei absorb the energy and begin to precess, spiraling around the direction of the magnetic field at that resonant frequency. As they precess, they generate their own tiny oscillating magnetic signal, and that signal is what the scanner detects.
Relaxation and Tissue Contrast
The real magic of MRI is what happens after the radiofrequency pulse ends. The disturbed hydrogen nuclei gradually return to their equilibrium state, and they do so by two independent processes that occur simultaneously but at different rates depending on the tissue.
The first process involves the nuclei realigning with the main magnetic field. The time this takes is called the T1 relaxation time, and it varies by tissue type. Fat, for instance, has a short T1 because its hydrogen nuclei recover quickly. Fluid like cerebrospinal fluid has a long T1 and recovers slowly. The second process involves the nuclei losing coherence with each other. Right after the pulse, they are all spinning in sync, but they gradually fall out of step. The time this takes is called the T2 relaxation time, and it also differs between tissues.
In practice, the loss of coherence happens even faster than the intrinsic T2 time because slight irregularities in the magnetic field cause some nuclei to precess a little faster or slower than their neighbors. This faster decay is called T2* relaxation. It is a combination of the inherent spin-spin relaxation and the effects of magnetic field inhomogeneity, and it forms the basis for several advanced applications including perfusion imaging and functional MRI.2PubMed Central. Principles, techniques, and applications of T2*-based MR imaging and its special applications
By choosing when to collect the signal after the radiofrequency pulse, the scanner operator can emphasize either T1 or T2 differences between tissues. A T1-weighted image, collected quickly after the pulse, highlights tissues that recover fast (like fat, which appears bright). A T2-weighted image, collected later, highlights tissues with long T2 values (like fluid, which appears bright). This is how a single machine can make the same body part look dramatically different depending on the scan settings. No physical change happens to the patient; only the timing of signal collection changes.
Gradient Coils and Spatial Encoding
Knowing that hydrogen nuclei in your body are emitting a signal is not enough to build an image. The scanner also needs to know exactly where each bit of signal is coming from. This is the job of the gradient coils, a separate set of electromagnets inside the bore of the scanner that slightly vary the magnetic field strength across the body in three directions.
The principle is straightforward. If the magnetic field is slightly stronger at your head than at your feet, the hydrogen nuclei near your head precess at a slightly higher frequency. By applying these gradients in carefully timed sequences, the scanner encodes spatial information into the frequency and phase of the returning signal. One gradient selects a specific slice through your body. Another encodes position along one axis of that slice by frequency. A third encodes position along the perpendicular axis by phase. The combined effect is that every small volume element in the selected slice contributes a signal with a unique combination of frequency and phase, which the scanner’s computer can decode.
This encoding process is also the main reason MRI scanners are so loud. The gradient coils carry rapidly alternating electrical currents while sitting inside the powerful static magnetic field. The interaction between the current and the field produces forces that vibrate the coils mechanically, and those vibrations produce the banging, knocking, and buzzing you hear during a scan.3PubMed Central. Acoustic Noise Levels in High‐field Magnetic Resonance Imaging Scanners Stronger magnets produce louder noise because the forces on the gradient coils increase with field strength.
Turning Raw Data into Pictures
The signal the scanner collects is not an image. It is a collection of frequency and phase data stored in a mathematical space called k-space. Each point in k-space represents a specific combination of spatial frequency information about the image. The standard way to convert k-space data into a viewable image is through a mathematical operation called the Fourier transform, which separates the jumbled frequency data into its spatial components.
Most clinical scans fill k-space along a regular grid, which makes the reconstruction fast and straightforward. But some advanced techniques sample k-space along non-standard paths, such as spirals or radial spokes. Converting those non-uniform samples into an image requires more sophisticated reconstruction methods.4PubMed Central. A note on the iterative MRI reconstruction from nonuniform k-space data The tradeoff is usually speed versus image quality. Radial sampling, for example, is more forgiving of patient motion but takes more computational effort to reconstruct cleanly.
The reconstruction step is where the raw physics ends and the computing begins. Modern scanners use powerful processors to turn k-space data into images in seconds, and machine learning is increasingly being used to improve image quality from faster, lower-data acquisitions. But the fundamental principle has not changed since the earliest MRI images: collect frequency-encoded signals, apply the Fourier transform, and display the result as a grayscale image where brightness corresponds to signal strength.
Contrast Agents and What They Do
For many scans, the natural contrast between tissues is enough. But sometimes a doctor needs to see features that are subtle on a standard image, such as a small tumor or inflammation. That is where contrast agents come in. The most widely used MRI contrast agents are based on gadolinium, a rare-earth metal that is strongly paramagnetic.
Gadolinium does not show up in the image directly. Instead, it alters the relaxation times of nearby water molecules. The high magnetic moment of gadolinium shortens both T1 and T2 relaxation times of surrounding hydrogen nuclei, but the dominant visible effect on most scans is T1 shortening, which makes tissues that have absorbed the contrast agent appear brighter on T1-weighted images.5European Journal of Radiology Open. Gadolinium contrast agents- challenges and opportunities of a multidisciplinary approach: Literature review The degree of T1 shortening depends on several factors, including how close the gadolinium ions get to the water protons, how long they interact, and the concentration of the agent.6PubMed. Mechanisms of contrast enhancement in magnetic resonance imaging
Because tumors and inflamed tissues tend to have leaky blood vessels that accumulate more contrast agent than healthy tissue, they “light up” after gadolinium injection. This enhancement helps radiologists distinguish a benign cyst (which does not enhance) from a solid tumor (which often does), or identify areas of active inflammation in conditions like multiple sclerosis. Gadolinium-based agents are injected intravenously and are cleared by the kidneys, usually within hours.
Functional MRI and Brain Activity
One of the more striking applications of MRI does not image anatomy at all. Functional MRI, or fMRI, maps brain activity by detecting changes in blood oxygenation. When a region of the brain becomes more active, local blood flow increases and delivers more oxygenated hemoglobin than the tissue is consuming. Deoxyhemoglobin is paramagnetic and disturbs the local magnetic field, while oxyhemoglobin is not. So when a brain area “lights up” with activity, the local ratio shifts toward oxyhemoglobin, reducing magnetic field disturbances and producing a slightly stronger MRI signal.7PubMed Central. Coupling mechanism and significance of the BOLD signal: a status report
This blood-oxygen-level-dependent signal is what fMRI detects. The effect is small, typically a change of just a few percent in signal intensity, and it lags behind actual neural firing by several seconds because it depends on blood flow changes rather than electrical activity directly. Despite these limitations, fMRI has become an indispensable research tool for mapping which brain regions are involved in tasks ranging from language processing to decision-making. It relies on T2* contrast, the same magnetic-field-inhomogeneity-sensitive signal described earlier, which makes gradient-echo sequences the workhorse of functional brain imaging.2PubMed Central. Principles, techniques, and applications of T2*-based MR imaging and its special applications
Why Metal Implants Cause Problems
If you have ever been asked about surgical implants before an MRI, this is the reason: metals interact with the magnetic field in ways that can distort images and, in rare cases, pose safety risks. The core issue is the difference in magnetic susceptibility between metal and the surrounding tissue. When a metal implant sits in the scanner’s field, it distorts the field locally, causing signal loss, geometric warping, and failure of fat-suppression techniques in the area around the implant.8PubMed Central. Managing hardware-related metal artifacts in MRI: current and evolving techniques
The severity depends on the metal. Titanium and certain cobalt-chromium alloys are weakly magnetic and produce modest artifacts. Stainless steel is more problematic. The distortion can extend well beyond the implant itself, obscuring the very anatomy the scan is trying to evaluate. Most medical implant metals have magnetic susceptibilities that lead to frequency offsets of up to about 150 parts per million in the tissue near the device, which is enough to dominate the image’s geometric accuracy in that region.9PubMed Central. 3D quantification of metal-induced geometric distortions in MRI
Specialized scan sequences have been developed to reduce these metal artifacts. They use faster, wider-bandwidth readouts and modified encoding strategies to compensate for the field distortions. These techniques have made it possible to scan patients with joint replacements and spinal hardware more effectively than a decade ago, though some signal loss near the implant is still difficult to eliminate entirely.
Heating and Safety Considerations
MRI does not use ionizing radiation, but it is not entirely without physical effects. The radiofrequency pulses deposit energy into the body as heat, measured as the specific absorption rate (SAR). In a healthy person with no implants, the heating is minimal and tightly regulated by scanner safety limits. Regulatory guidelines cap the amount of energy the scanner can deliver per unit of body weight.
The picture changes with implants. Metallic components can concentrate the radiofrequency energy and heat up locally. Testing of implanted devices in 3-tesla scanners has shown that conventional clinical sequences keep temperature rises below about 2 degrees Celsius, but more aggressive research sequences that deliver higher RF energy can push temperatures closer to that threshold.10PubMed Central. Safety of 3 Tesla Magnetic Resonance Imaging in Active Peripheral Nerve Field Stimulation Device for Facial Pain The head-averaged SAR of each sequence turns out to be a strong predictor of how much heating occurs, which is why technologists pay close attention to SAR limits, especially when scanning patients with implanted electrical devices.
Beyond heating, the static magnetic field itself creates a strong pull on ferromagnetic objects. This is the reason loose metal items are strictly prohibited in the scanner room. The “projectile effect,” where a forgotten oxygen tank or wheelchair is accelerated toward the magnet bore, is responsible for most serious MRI safety incidents. The magnetic field is always on, even when no scan is running, which is something visitors to MRI suites sometimes do not realize.
Low-Field MRI and Expanding Access
Most hospitals use scanners at 1.5 or 3 tesla, but there is growing interest in low-field MRI systems operating below 1 tesla, and sometimes well below it. These systems use either permanent magnets or simple electromagnets that require no cryogenic cooling, which eliminates the need for expensive liquid helium infrastructure and the specialized quench piping that high-field systems demand.11PubMed Central. Low-Field Magnetic Resonance Imaging: Its History and Renaissance
The appeal is practical. A permanent-magnet MRI consumes very little power, can be installed in spaces where a superconducting system would not fit, and costs a fraction of the price. Some portable low-field systems can be wheeled to a patient’s bedside in an intensive care unit, which is impossible with a conventional scanner. The tradeoff is image quality: lower field strength means a weaker signal, which generally translates to lower resolution, longer scan times, or both. But for certain clinical questions, like checking for a large stroke or monitoring fluid buildup in the brain, the images are good enough to guide treatment decisions in settings where high-field MRI simply is not available.
This renaissance in low-field imaging is being driven partly by advances in computing. Better reconstruction algorithms and machine learning can extract more useful information from noisier data than was possible a decade ago, narrowing the quality gap between low-field and high-field systems for specific applications. It does not replace the high-resolution detail of a 3-tesla scan for complex neurological or musculoskeletal questions, but it opens MRI to clinics, rural hospitals, and low-resource settings that have never had access to the technology before.
Why Different Scan Sequences Exist
If you have ever had more than one MRI, you may have noticed that some scans take five minutes and others take forty, and the machine makes different patterns of sounds during each one. Each distinct sound pattern corresponds to a different pulse sequence, a specific recipe of radiofrequency pulses and gradient timing designed to highlight particular tissue properties.
A spin-echo sequence, one of the oldest and most reliable, uses a second radiofrequency pulse to refocus the hydrogen nuclei after they start to lose coherence. This refocusing eliminates the extra signal loss caused by magnetic field irregularities, giving you a “clean” T2 measurement that reflects true tissue properties. A gradient-echo sequence skips that refocusing step, making it faster but leaving the signal sensitive to field inhomogeneities, which is why T2*-weighted gradient-echo images are the foundation of functional MRI and susceptibility-weighted imaging.2PubMed Central. Principles, techniques, and applications of T2*-based MR imaging and its special applications
Beyond these two families, there are dozens of specialized sequences. Diffusion-weighted imaging tracks the random motion of water molecules and is exquisitely sensitive to acute stroke. Inversion recovery sequences suppress the signal from fat or fluid to make other structures stand out. Angiography sequences highlight flowing blood without needing an injected contrast agent. Each builds on the same underlying physics of magnetic alignment, radiofrequency excitation, and relaxation, just with different timing, different gradient patterns, and different post-processing. The versatility of MRI comes not from having many different machines but from the sheer number of ways you can interrogate the same hydrogen nuclei inside the same magnet.