How MRI Machines Create Detailed Images of Soft Tissue Without Radiation

An MRI Scanner Can Detect a Tumor 1mm Across Without Using a Single X-Ray Photon

Magnetic resonance imaging uses no ionizing radiation. Instead, it exploits a property of atomic nuclei discovered in 1946 — nuclear magnetic resonance — to create images of astonishing anatomical detail. A 3 Tesla MRI system can distinguish white matter from gray matter in the brain, identify a 1 mm herniated disc in the spine, and map blood flow through individual arteries. All using radio waves and magnetic fields.

The physics relies on the quantum mechanical property of protons — the nuclei of hydrogen atoms — called spin. The human body is approximately 60% water, and water contains two hydrogen atoms per molecule. This makes the body an exceptionally rich source of NMR signal. MRI essentially reads out information about where hydrogen atoms are and what molecular environment surrounds them.

Nuclear Magnetic Resonance: Protons as Tiny Magnets

Protons have a quantum property called spin, which gives them a magnetic moment — they behave like tiny bar magnets. Normally, these magnetic moments point in random directions and their fields cancel. When placed in a strong external magnetic field B₀, protons align either parallel (low energy) or antiparallel (high energy) to the field. A slight excess aligns parallel — creating a net magnetization M₀ in the direction of B₀.

These aligned protons precess (wobble) around B₀ at the Larmor frequency: ω₀ = γ × B₀, where γ is the proton gyromagnetic ratio (42.58 MHz/Tesla). For a 1.5 T scanner, protons precess at 63.87 MHz — in the radio frequency range.

When a radio frequency (RF) pulse is applied at exactly the Larmor frequency, it resonates with the precessing protons, tipping the net magnetization away from B₀. This is excitation. When the RF pulse stops, the magnetization recovers — relaxation — and the precessing protons emit detectable RF signals.

The Magnet: Superconducting Coils at 4 Kelvin

Clinical MRI scanners use superconducting electromagnets to generate the strong, uniform magnetic fields required. Coils of niobium-titanium (NbTi) wire are cooled to 4.2 Kelvin (−269°C) using liquid helium. At this temperature, the wire becomes superconducting — electrical resistance drops to exactly zero — and the coil carries thousands of amperes with no energy loss after the initial charging. The magnet field persists indefinitely.

Relaxation: T1 and T2 — The Contrast Mechanisms

After excitation, magnetization returns to equilibrium through two independent relaxation processes. These times differ dramatically between tissue types — providing the contrast that makes MRI diagnostically valuable.

• T1 (longitudinal relaxation): The time for magnetization to recover to 63% of its equilibrium value along B₀. T1 reflects how efficiently protons exchange energy with surrounding molecules (the "lattice"). Fat has a short T1 (~260 ms at 1.5T); cerebrospinal fluid has a long T1 (~4,000 ms). T1-weighted images make fat appear bright and fluid dark.
• T2 (transverse relaxation): The time for the transverse magnetization to decay to 37% of its initial value. T2 reflects phase coherence loss among precessing protons due to local magnetic field inhomogeneities. Fluid has a long T2 (~2,000 ms); muscle a short T2 (~50 ms). T2-weighted images make fluid bright — useful for detecting edema, tumors, and inflammation.

Spatial Encoding: Turning NMR Signals Into Images

The key challenge: the received RF signal comes from the entire body simultaneously. How does the scanner know which proton is where? Gradient magnetic fields solve this.

Three sets of gradient coils produce small linear variations in B₀ along x, y, and z axes. Because the Larmor frequency depends on B₀, protons at different positions precess at slightly different frequencies. Selective RF excitation with a slice-selective gradient excites only one slice at a time. Frequency encoding gradients during signal readout encode position in one in-plane direction. Phase encoding gradients — applied briefly before readout — encode the perpendicular in-plane direction as phase differences in the signal.

The received signal is a 2D Fourier transform of the image — called k-space. Applying the inverse 2D Fourier transform to the collected k-space data yields the final image. This mathematical relationship between k-space and image space is what makes MRI reconstruction possible.

The Pulse Sequence: Orchestrating the Acquisition

A pulse sequence specifies the timing and parameters of RF pulses, gradient applications, and signal readouts. Sequence design directly determines what tissue contrast is achieved:

• Spin echo (SE): A 90° excitation pulse followed by a 180° refocusing pulse. Produces T1 or T2 contrast depending on repetition time (TR) and echo time (TE).
• Gradient echo (GRE): Uses a smaller flip angle and gradient reversal instead of an RF refocusing pulse. Faster acquisition, useful for angiography and dynamic imaging.
• Inversion recovery: A 180° inversion pulse before the excitation allows selective nulling of specific tissues. FLAIR (Fluid Attenuated Inversion Recovery) suppresses CSF signal to reveal periventricular lesions clearly.
• Diffusion-weighted imaging (DWI): Adds pulsed gradient pairs that sensitize the signal to proton diffusion. Restricted diffusion in acute stroke appears bright — allowing stroke detection within minutes of onset.

Functional MRI: Mapping Brain Activity

Blood-oxygen-level-dependent (BOLD) fMRI uses an indirect marker of neural activity. When neurons fire, blood flow increases in the region, bringing oxygenated hemoglobin (diamagnetic — no effect on MRI signal) and washing out deoxygenated hemoglobin (paramagnetic — shortens T2*). The signal increase in active areas is only 1–4%, but it can be detected across thousands of brain voxels simultaneously. BOLD fMRI has become the primary tool for mapping cognitive functions and identifying eloquent cortex before neurosurgery.

MRI Safety: What the Magnetic Field Can Do

The static magnetic field attracts ferromagnetic objects with dangerous force. A 1.5 T scanner exerts sufficient force to accelerate an oxygen cylinder across a room at lethal speed. The FDA has classified MRI facilities as zone 1–4 environments with progressively stricter metal screening requirements. Patients with pacemakers, cochlear implants, and certain vascular clips may be contraindicated. The time-varying gradient fields generate loud acoustic noise (up to 130 dB) and can induce peripheral nerve stimulation. These hazards are manageable with proper protocols — but the magnetic field is always on.