Loading field map
Imaging physics is the branch of medical physics concerned with the physical principles, instrumentation, and quantitative methods that underlie the production of medical images. Its practitioners design, characterize, and optimize the machines that create images of the human body, and they develop the mathematical procedures that turn raw measured signals into diagnostically useful pictures. The field sits at the intersection of fundamental physics, engineering, and clinical medicine: it asks how radiation or other physical energy interacts with tissue, how that interaction can be detected and encoded, and how the resulting data can be reconstructed into an image that faithfully represents anatomy or function.
The central stakes of imaging physics are twofold. First, image quality must be sufficient for the clinical question at hand—whether that means resolving a small tumor, measuring blood flow, or tracking a catheter in real time. Second, the imaging process must be safe: ionizing radiation doses must be kept as low as reasonably achievable, and non-ionizing modalities must respect tissue heating and other biological limits. These two goals often conflict, and much of the field's practical work involves balancing them through careful system design, protocol optimization, and quality assurance.
All medical imaging modalities share a common logical structure. A physical agent—X-rays, gamma rays, sound waves, radiofrequency electromagnetic fields, or magnetic field gradients—is directed into the body. The agent interacts with tissue in a way that depends on some property of that tissue, such as electron density, acoustic impedance, or proton concentration. Detectors placed around or outside the body record the transmitted, emitted, or reflected signal. Finally, a reconstruction algorithm converts the detected signals into a spatial map of the tissue property, which is displayed as an image.
The differences among modalities arise from which physical property is mapped and how the signal is encoded with spatial information. In X-ray radiography and computed tomography (CT), the property is the attenuation of X-rays along straight-line paths through the body. In nuclear medicine, including single-photon emission computed tomography (SPECT) and positron emission tomography (PET), the property is the concentration of a radioactive tracer that has been injected into the patient. In ultrasound, the property is the reflection of sound waves at boundaries between tissues with different acoustic impedances. In magnetic resonance imaging (MRI), the property is the density and relaxation behavior of hydrogen nuclei in water and fat, encoded through their response to radiofrequency pulses in a strong magnetic field.
The physics of each modality determines its fundamental capabilities and limitations. X-ray attenuation depends strongly on atomic number, which is why bone appears bright on CT and why iodine-based contrast agents can highlight blood vessels. Ultrasound cannot penetrate bone or air-filled structures well, but it images soft tissue in real time without ionizing radiation. MRI provides exquisite soft-tissue contrast and can be sensitized to physiological parameters like blood flow or tissue perfusion, but it is slow, expensive, and contraindicated for patients with certain metallic implants. PET detects pairs of gamma rays emitted in opposite directions when a positron annihilates with an electron, allowing sensitive detection of molecular tracers but with relatively poor spatial resolution.
Imaging physics emerged gradually from the physics of radiation and its medical applications. The discovery of X-rays in 1895 and of natural radioactivity in 1896 immediately suggested medical uses, and within a few years physicians were using X-ray tubes to image bones and foreign objects. The early practitioners were physicists and engineers who understood the vacuum tubes, high voltages, and photographic plates involved. Their work was largely empirical, guided by trial and error rather than by a systematic theory of image formation.
The field matured during the twentieth century as new physical principles were harnessed for imaging. The development of nuclear medicine in the 1950s and 1960s required physicists to design scintillation detectors, collimators, and the first gamma cameras. The invention of CT in the early 1970s was a landmark: it required solving the mathematical problem of reconstructing a cross-sectional image from a large set of line-integral measurements, a problem with roots in the Radon transform developed in 1917. MRI, which emerged in the 1970s and 1980s, drew on nuclear magnetic resonance physics that had been studied in chemistry and physics laboratories for decades. Ultrasound imaging developed from sonar and industrial nondestructive testing, with medical applications beginning in the 1950s.
Throughout this history, imaging physics has been characterized by a close and productive relationship between fundamental physics and clinical need. The field has never been a purely academic discipline; its problems are posed by the demands of diagnosis and therapy, and its solutions are validated by their performance in clinical settings. At the same time, it has consistently drawn on deep physics: the quantum theory of radiation detection, the statistical mechanics of signal and noise, the electromagnetic theory of radiofrequency coils, and the mathematics of inverse problems.
The field is not organized into rival schools or paradigms in the sense that, say, particle physics has competing theoretical frameworks. Rather, it is organized around the distinct physical principles of each modality, each with its own instrumentation, physics, and reconstruction mathematics. These modality-based traditions are the natural units of the field, and they coexist rather than compete. A medical physicist typically specializes in one or two modalities, though the underlying concepts of signal detection, noise, and image quality unify the field.
The physics of X-ray imaging begins with the production of X-rays in a vacuum tube, where electrons accelerated by a high voltage strike a metal anode and emit bremsstrahlung and characteristic X-rays. The energy spectrum of the beam, its filtration, and its collimation determine both the radiation dose to the patient and the quality of the resulting image. The interaction of X-rays with tissue is governed by the photoelectric effect and Compton scattering, whose relative probabilities depend on photon energy and tissue composition. The photoelectric effect dominates at lower energies and scales strongly with atomic number, providing the contrast between bone and soft tissue; Compton scattering dominates at higher energies and contributes background that degrades image contrast.
In CT, the X-ray source and detector array rotate around the patient, acquiring projections from many angles. The reconstruction problem is to estimate the spatial distribution of attenuation coefficients from these projections. The standard solution is filtered backprojection, which applies a mathematical filter to each projection and then smears it back across the image plane. More recently, iterative reconstruction methods have become common: they model the imaging process statistically, including the physics of photon detection and the geometry of the scanner, and solve for the image that best explains the measured data. These methods can reduce radiation dose for the same image quality, but they require careful tuning of regularization parameters and can introduce artifacts if the model is inaccurate.
The physics of X-ray detection is also central. Modern CT scanners use solid-state detectors that convert X-rays to light and then to electrical signals, with each detector element corresponding to a small angular and spatial bin. The efficiency, dynamic range, and noise characteristics of these detectors directly affect image quality. The same physics underlies digital radiography, where flat-panel detectors have largely replaced film-screen combinations.
Nuclear medicine imaging is fundamentally different from X-ray imaging because the radiation source is inside the patient. A radiopharmaceutical—a molecule labeled with a radioactive isotope—is administered, and its distribution in the body reflects physiological or biochemical processes. The imaging task is to map that distribution from the gamma rays that escape the body.
The basic instrument is the gamma camera, which uses a collimator—typically a thick lead plate with parallel holes—to ensure that detected gamma rays come from a known direction. The collimator rejects most of the emitted radiation, which makes the detection process inefficient and limits the spatial resolution. SPECT acquires multiple projections by rotating the gamma camera around the patient and reconstructs a three-dimensional image using methods analogous to CT.
PET exploits a different physical principle. Certain isotopes decay by positron emission; the positron travels a short distance and annihilates with an electron, producing two gamma rays traveling in opposite directions. If two detectors on opposite sides of the patient register gamma rays within a short time window, the annihilation must have occurred somewhere along the line connecting them. This electronic collimation is far more efficient than physical collimation, giving PET higher sensitivity than SPECT. The reconstruction problem is to estimate the tracer distribution from the set of coincidence lines, and modern PET scanners use iterative methods that model the physics of attenuation, scatter, and random coincidences. PET images are often combined with CT or MRI in hybrid scanners, which provide anatomical context and, in the case of CT, an attenuation map for correcting the PET data.
The physics of radioactive decay, including half-life, decay mode, and photon energy, determines which isotopes are useful for imaging. The short half-lives of many PET isotopes, such as fluorine-18, require on-site cyclotrons or nearby radiopharmaceutical production facilities. The radiation dose to the patient is determined by the total activity administered and the biological clearance of the tracer.
MRI is the most physically complex of the major imaging modalities. It exploits the fact that hydrogen nuclei (protons) have a magnetic moment and a spin angular momentum. In a strong static magnetic field, the protons align preferentially with the field, producing a net magnetization. A radiofrequency pulse at the Larmor frequency—the frequency of precession, proportional to the magnetic field strength—tips this magnetization away from equilibrium. As the magnetization returns to equilibrium, it induces a signal in a receiver coil. The rate of return is characterized by two relaxation times: T1, the longitudinal relaxation time, and T2, the transverse relaxation time. Different tissues have different T1 and T2 values, and the image contrast can be manipulated by choosing pulse sequence parameters such as the repetition time and echo time.
Spatial encoding in MRI is achieved with magnetic field gradients. By applying a gradient, the magnetic field—and hence the Larmor frequency—becomes a function of position. The received signal is then a sum of contributions from all locations, each with a frequency determined by its position. The image is reconstructed by taking the Fourier transform of the signal, which is sampled in a space called k-space. The relationship between k-space sampling and image properties is fundamental to MRI physics: the extent of k-space sampled determines spatial resolution, while the spacing between samples determines the field of view.
The physics of MRI also includes the design of radiofrequency coils, which must efficiently transmit the excitation pulse and receive the signal; the behavior of the static magnetic field, which must be highly uniform; and the safety considerations of strong magnetic fields, including the potential for tissue heating from radiofrequency energy and the hazards of ferromagnetic objects being attracted into the scanner. Advanced MRI techniques extend the basic physics to measure diffusion of water molecules, blood flow, brain function through blood-oxygen-level-dependent contrast, and chemical composition through spectroscopy.
Ultrasound imaging uses high-frequency sound waves, typically in the range of 1 to 15 MHz, which are generated and detected by piezoelectric transducers. The transducer emits a short pulse of sound into the body and then listens for echoes reflected from tissue boundaries and scatterers. The time delay between transmission and reception gives the depth of the reflecting structure, and the amplitude of the echo gives information about the acoustic impedance mismatch at the boundary. The image is built up line by line as the transducer beam is swept across the region of interest.
The physics of ultrasound is governed by the wave equation and the acoustic properties of tissue. The speed of sound is roughly constant in soft tissue, which allows time-of-flight to be converted to distance. Attenuation of the sound beam increases with frequency, which limits the depth that can be imaged at higher frequencies; this is why high-frequency transducers give better resolution but shallower penetration. Reflection occurs at boundaries where the acoustic impedance changes, and scattering from small structures produces the speckle pattern that gives ultrasound its characteristic texture. The Doppler effect is used to measure blood flow: the frequency shift of echoes from moving red blood cells is proportional to their velocity.
Ultrasound physics also includes the design of transducer arrays, which use electronic phasing to steer and focus the beam without moving the transducer. Modern systems use thousands of elements and sophisticated beamforming algorithms. The safety considerations for ultrasound are different from those for ionizing radiation: the main concerns are tissue heating and mechanical effects such as cavitation, and the field has established indices to quantify these risks.
Despite the diversity of modalities, imaging physics is held together by a set of shared concepts. The first is the distinction between signal and noise. Every imaging system produces a signal that carries information about the tissue, and noise that corrupts that signal. The noise may arise from the random nature of photon detection, from electronic components, from patient motion, or from the reconstruction process. The signal-to-noise ratio (SNR) is a fundamental measure of image quality, and much of imaging physics is devoted to maximizing SNR for a given dose or acquisition time.
The second unifying concept is spatial resolution: the ability to distinguish two nearby structures as separate. Resolution is limited by the physics of the modality—the diffraction of sound, the size of detector elements, the extent of k-space sampled, the range of projection angles acquired. There is an unavoidable trade-off between resolution and SNR, because smaller image elements collect less signal.
The third concept is contrast: the difference in signal between two tissues or between a lesion and its background. Contrast is determined by the physical property being imaged and by the way the acquisition parameters are chosen. A modality may have excellent resolution but poor contrast for a particular clinical question, or vice versa.
The fourth concept is the imaging equation: the mathematical relationship between the measured data and the image. This equation is the basis for reconstruction algorithms, and it also determines the artifacts that arise when the assumptions of the reconstruction are violated. For example, CT reconstruction assumes that X-rays travel in straight lines and that the beam is monochromatic; beam hardening and scatter violate these assumptions and produce artifacts. Understanding the imaging equation is what allows physicists to design correction algorithms.
The current practice of imaging physics is shaped by several ongoing developments. The first is the trend toward hybrid and multimodality imaging. PET/CT and SPECT/CT are now standard, and PET/MRI is increasingly available. These systems require physicists who understand both modalities and the ways they interact—for example, how the CT attenuation map affects the PET reconstruction, or how the MRI radiofrequency field affects the PET detectors.
The second is the increasing role of quantitative imaging. Rather than simply producing a picture for visual interpretation, modern imaging aims to measure physical or biological quantities: the standardized uptake value in PET, the apparent diffusion coefficient in MRI, the attenuation coefficient in CT. Quantitative imaging requires careful calibration, standardization across scanners, and rigorous quality assurance, all of which are core activities of imaging physics.
The third is the growing influence of computational methods. Iterative reconstruction, machine learning for image denoising and reconstruction, and model-based approaches to image analysis are transforming the field. These methods are not replacements for physics; rather, they incorporate physical models of the imaging process into algorithms that can handle complex data. The imaging physicist's role is increasingly to understand both the physics and the mathematics of these methods, and to ensure that they are validated and safe for clinical use.
The fourth is the ongoing concern with radiation safety. The principle of ALARA (as low as reasonably achievable) drives continuous efforts to reduce dose in CT and nuclear medicine without compromising diagnostic quality. This involves optimizing acquisition protocols, using dose-reduction technologies such as automatic exposure control and iterative reconstruction, and auditing doses across institutions.
Imaging physics remains a fundamentally applied discipline, but its applied nature does not diminish its intellectual depth. The field requires a command of quantum mechanics, electromagnetism, acoustics, signal processing, and inverse problem theory, all in service of a single goal: producing images that help clinicians care for patients. Its practitioners work in hospitals, in industry, and in academia, and their work spans the entire chain from fundamental physical principles to the daily operation of clinical scanners.