What Is Tomographic Imaging and How Does It Work?

Tomographic imaging is any technique that creates cross-sectional pictures of the inside of an object by collecting data from many different angles and then mathematically assembling those data into slices. The word “tomography” comes from the Greek tomos (slice) and graphein (to write), and the core idea is the same whether you are scanning a human chest, mapping the Earth’s mantle, or inspecting a turbine blade. What changes from one type of tomography to another is the kind of energy used, whether X-rays, gamma rays, sound waves, light, or even electrical current, and the math that stitches the raw measurements into an image you can interpret.

The Core Principle Behind Every Tomographic Image

A conventional photograph or X-ray projects a three-dimensional scene onto a flat surface. Everything along the line of sight piles on top of everything else, which is why a standard chest X-ray can show a rib overlapping a lung nodule and make it hard to see either clearly. Tomography solves this by gathering information from many directions around the object. In X-ray computed tomography (CT), a tube fires X-rays through the body while a detector array sits on the opposite side. The tube and detectors rotate, collecting projection data at hundreds of angles. Each projection records how much the X-rays were weakened along every path through the body. A computer then works backward from all those projections to calculate what must be at each point inside the slice, producing a clean cross-sectional image with no overlap.

This “work backward” step is the reconstruction, and it is mathematically grounded in the Radon transform, a function that describes all possible line integrals through a two-dimensional object. Inverting the Radon transform, going from projection data back to an image, was the conceptual breakthrough that made tomography practical. Researchers have developed fast algorithms that exploit symmetry properties of the Radon transform and its inverse to make reconstruction computationally efficient.¹ Once you grasp this basic loop of “collect projections from many angles, then reconstruct,” every tomographic modality is a variation on the theme.

X-Ray Computed Tomography

CT is by far the most widely used form of tomographic imaging. The first clinical CT scan was performed in 1971 at Atkinson Morley Hospital in London, and by 1979 the technique’s inventors, Allan Cormack and Godfrey Hounsfield, had received the Nobel Prize in Physiology or Medicine.² Modern CT scanners can capture a full chest or abdomen in seconds, producing stacks of thin slices that radiologists scroll through or render into three-dimensional volumes.

In a CT image, each tiny picture element (voxel) is assigned a number on the Hounsfield scale. Air sits near −1000, water is defined as 0, and dense bone lands around +1000 or higher. Soft tissues cluster between roughly −100 and +100. Researchers have worked to tie these Hounsfield values more precisely to fundamental physical units, achieving predictions within about ten percent accuracy of measured values.³ That consistency is what makes CT so useful: a radiologist anywhere in the world knows roughly what tissue type a given Hounsfield number represents.

Because X-ray CT fires ionizing radiation through the body, it carries a dose penalty. A single abdominal CT delivers a radiation dose roughly equivalent to a few years of natural background exposure. Managing that dose, especially for children and for patients who need repeated scans, has driven a large body of work on scanning technique, including tube current modulation, organ-specific shielding, and careful protocol design.⁴ Institutions that take dose reduction seriously have shown that combining protocol revision, automated dose modulation software, and education for ordering physicians can meaningfully lower the radiation their patients receive.⁵

How Raw Data Become Readable Images

The earliest CT scanners used a technique called filtered back projection (FBP). The idea is to take each angular projection and “smear” it back across the image plane, then apply a mathematical filter to sharpen the result. FBP is fast and predictable, and it remained the default reconstruction method for decades. Iterative reconstruction, which refines the image through repeated cycles of comparison between the computed and measured projections, was actually proposed back in the 1970s but was too computationally expensive for the hardware of the day. It took until 2009 for the first commercial iterative reconstruction algorithms to reach clinical use.⁶

More recently, deep-learning reconstruction has entered the picture. These algorithms train neural networks on large datasets of high-quality images and then apply the learned patterns to produce cleaner results from noisier or lower-dose input data. In abdominal CT, for example, deep-learning reconstruction has been shown to outperform both FBP and hybrid iterative methods in reducing artifacts caused by the patient’s arms being positioned at their sides during the scan.⁷ Deep learning has also been explored as a way to generate volumetric CT images from far fewer projection views than would normally be needed, potentially slashing scan time and dose even further.⁸

When Metal and Physics Get in the Way

No imaging technique is perfect, and CT has a well-known set of image artifacts. The most visually dramatic is the metal artifact: bright and dark streaks radiating from dental fillings, joint replacements, or surgical hardware. These streaks are caused mainly by beam hardening. X-ray tubes produce a spectrum of photon energies, and as the beam passes through dense metal, the lower-energy photons are absorbed preferentially. The remaining beam is “harder” (higher in average energy) than the reconstruction algorithm expects, causing a mismatch that shows up as streaks and shadows.⁹

Correcting these artifacts is an active area of research. For lower-density implants like titanium, algorithms can estimate the beam hardening along each X-ray path and adjust the projection data accordingly. For very dense metals like gold or platinum dental fillings, the projections passing through the metal are sometimes treated as missing data entirely and filled in using iterative reconstruction techniques.¹⁰ Newer correction approaches estimate the mean photon energy along each ray path and use that to convert energy-dependent projections into corrected data, substantially reducing artifacts even in complex cases with multiple overlapping materials.¹¹

Nuclear Medicine Tomography

CT images anatomy by sending energy into the body from outside. Nuclear medicine flips the script: a radioactive tracer is injected into the patient, and the scanner detects radiation coming out. The two main nuclear tomographic techniques are positron emission tomography (PET) and single-photon emission computed tomography (SPECT).

In PET, the tracer contains an atom that decays by emitting a positron. The positron almost immediately collides with a nearby electron, and the two annihilate each other, producing a pair of high-energy photons that fly off in nearly opposite directions. The PET scanner is a ring of detectors that registers these photon pairs arriving at the same instant, a coincidence event, and uses the line between the two detectors as the projection data.¹² Millions of coincidence events are accumulated and then reconstructed into a three-dimensional map showing where the tracer concentrated. Because PET tracers can be designed to track specific biochemical processes, such as glucose consumption by cancer cells, PET excels at functional and metabolic imaging rather than pure anatomy.

SPECT uses tracers that emit single gamma-ray photons. A gamma camera, typically equipped with a collimator made of a dense material like tungsten or lead, rotates around the patient to collect projection views.¹³ Because there is no coincidence detection, SPECT generally has lower spatial resolution and sensitivity than PET, but it remains widely used for cardiac perfusion studies and certain neurological applications because the tracers are cheaper and more readily available.

Tomography Without Ionizing Radiation

Several tomographic methods avoid X-rays and radioactive tracers altogether, which makes them especially appealing for situations where radiation dose is a concern or where the physics of X-rays simply do not give you the contrast you need.

Optical coherence tomography (OCT) uses near-infrared light and a technique called low-coherence interferometry to create cross-sectional images of tissue microstructure. It works somewhat like ultrasound but uses light instead of sound, achieving spatial resolutions of just a few micrometers while detecting reflected signals as faint as one ten-billionth of the incident optical power.¹⁴ OCT cannot penetrate deeply, typically only a few millimeters into tissue, but within that range it produces exquisitely detailed images. Ophthalmologists rely on it daily to visualize the layers of the retina, and cardiologists thread OCT catheters into coronary arteries to inspect plaque.

Ultrasound tomography uses sound waves transmitted through tissue rather than the echo-based approach of conventional ultrasound. A ring of transducers surrounds the area of interest, such as a breast, and the speed and attenuation of sound passing through the tissue are reconstructed into quantitative maps.¹⁵ Because different tissue types transmit sound at different speeds, these images can help distinguish between cysts, solid masses, and normal tissue without any radiation exposure.¹⁶

Electrical impedance tomography (EIT) is perhaps the most unusual variant. Electrodes placed around the chest inject tiny electrical currents and measure the resulting voltage patterns. Because air-filled lung tissue conducts electricity very differently from fluid-filled or collapsed lung, EIT can create real-time images of how ventilation is distributed, which is valuable for managing patients with acute respiratory distress syndrome in intensive care units.¹⁷ EIT images are far coarser than CT, but the technique is portable, cheap, and produces continuous monitoring rather than a single snapshot.

Photoacoustic Tomography

Photoacoustic tomography (PAT) is a hybrid that exploits the strengths of both light and sound. Short pulses of laser light are fired into tissue. Molecules that absorb the light heat up by a tiny amount and expand, launching an ultrasonic pressure wave. Acoustic detectors pick up these waves and reconstruct an image that maps where optical absorption occurred.¹⁸ Because ultrasound scatters far less than light in tissue, PAT achieves much better resolution at depth than purely optical methods while retaining the rich contrast that comes from optical absorption.¹⁹

Hemoglobin is a strong natural absorber of laser light, so PAT can map blood vessel networks and even distinguish oxygenated from deoxygenated blood without any injected contrast agent. Recent work using clinically approved microbubbles as “virtual point sources” has extended the technique’s reach into deeper tissues, overcoming limitations that arise when detectors can only see part of the acoustic field.²⁰ PAT is still primarily a research tool, but its ability to provide functional and metabolic information without ionizing radiation has generated a lot of interest for applications in breast imaging, skin cancer detection, and small-animal research.

Tomography Outside the Hospital

The principles behind tomographic imaging reach well beyond medicine. In geophysics, seismic tomography uses earthquake waves as the “beam” and the entire Earth as the patient. Seismometers distributed around the globe record the arrival times and amplitudes of seismic waves that have traveled through the deep interior. Because waves speed up in cold, rigid rock and slow down in hot, soft rock, researchers can invert those travel times to build three-dimensional maps of the mantle’s velocity structure. Over the past two decades, improvements in computation and methodology have revealed the fine-scale structure of mantle plumes rising from the core-mantle boundary and slabs of old ocean floor sinking into the lower mantle, providing the most important geophysical constraints on how material circulates inside the planet.²¹

In structural biology, cryo-electron tomography (cryo-ET) lets researchers visualize individual molecules and cellular structures in three dimensions at near-native conditions. A thin, flash-frozen biological sample is tilted in an electron microscope, and images are collected at a series of angles, much like a CT scanner rotating around a patient, except the sample rotates instead of the source. Reconstructed tomograms reveal the shapes of proteins, membranes, and organelles without the chemical fixation or staining that traditional electron microscopy requires.²² A persistent bottleneck has been the speed of data collection, since each tilt-series requires physically rotating the sample between exposures. Newer acquisition strategies, including beam-image-shift protocols that image multiple areas at each stage position, have accelerated collection by as much as an order of magnitude while also improving the resolution of the final maps, reaching around 3.6 angstroms for sub-volume averages.²³ At that resolution, individual amino acid side chains become visible, opening the door to studying drug targets and molecular machines in situ.

Phase-Contrast and Spectral Techniques

Conventional X-ray CT works by measuring how much the beam is absorbed. But absorption is not the only thing that happens when X-rays pass through matter. The waves also shift in phase, and for lightweight materials like polymers, biological soft tissues, and carbon-fiber composites, the phase shift can be much larger than the absorption signal. Phase-contrast tomography captures this phase information to visualize structures that would be nearly invisible on a standard CT scan.²⁴ The technique is especially valuable in materials science for detecting cracks, voids, and subtle boundaries between components that have similar densities. Using synchrotron light sources and interferometric setups, researchers have achieved phase-contrast tomograms with scan times as short as half a second, fast enough to watch dynamic processes in real time.²⁵

Spectral or multi-energy CT is another frontier. Traditional CT detectors count all incoming photons without distinguishing their energies, which limits the information you can extract. Newer photon-counting detectors register the energy of each individual X-ray photon, making it possible to decompose the image into maps of specific materials, for example separating iodine-based contrast agent from calcium in bone, or distinguishing kidney stones made of uric acid from those made of calcium oxalate.²⁶ This spectral decomposition adds a chemical dimension to what has traditionally been a purely anatomical picture, and the first clinical photon-counting CT scanners are already in use at major medical centers.

Why So Many Kinds of Tomography Exist

A reasonable question after surveying all these techniques is why there is not just one best version. The answer is that every type of energy interacts with matter differently, and each interaction reveals something the others miss. X-rays show density differences beautifully but tell you nothing about blood oxygenation. PET reveals metabolic activity but cannot resolve structures smaller than a few millimeters. OCT achieves micrometer resolution but only penetrates a couple of millimeters. Ultrasound tomography is safe and cheap but struggles with bone and air. Cryo-electron tomography reaches near-atomic detail but only works on tiny, flash-frozen samples. Each modality occupies a niche defined by its penetration depth, resolution, contrast mechanism, speed, cost, and safety profile.

The trend in both clinical and research settings is toward combining modalities. PET/CT scanners, which fuse a metabolic PET image with an anatomical CT scan, are already standard in oncology. PET/MRI systems, which pair metabolic data with the soft-tissue contrast of magnetic resonance, are becoming more common. Photoacoustic tomography fuses optical and acoustic information in a single measurement. As reconstruction algorithms grow more powerful and detectors grow more sensitive, the boundaries between modalities continue to blur, and the core idea of tomography, building a picture from many vantage points, keeps finding new ways to show us what is hidden inside.