Introduction to Computed Tomography

Fundamentals of Computed Tomography What is Computed Tomography? Computed tomography (CT) is an imaging technique that creates detailed cross-sectional pictures of the interior of an object—typically the human body. Unlike conventional X-ray images, which superimpose all structures in the path of the beam, CT eliminates this overlapping problem by acquiring images from multiple angles and reconstructing them mathematically into thin slices. This allows clinicians to visualize internal organs, bones, and tissues with exceptional clarity and without the confusion caused by overlying structures. Key Motivation: The fundamental challenge CT solves is that a single X-ray image is a 2D shadow of a 3D object. By acquiring projections from many different angles around the patient, CT provides enough information to determine what's actually inside—not just what's in front. How CT Works: The Physical Principle The core of CT is elegantly simple: an X-ray source and detector rotate around the patient while they remain in a fixed geometric relationship to each other. As they rotate, the source and detector acquire narrow X-ray beam projections from hundreds of different angles—often spanning 360 degrees. Here's what happens at each angle: X-ray Attenuation: The X-ray beam passes through tissues and is weakened (attenuated) as it travels. The amount of attenuation depends on two factors: Tissue density: Denser tissues (like bone) attenuate X-rays more than less dense tissues (like lung) Tissue composition: Different materials absorb X-rays differently based on their atomic structure Each projection records a one-dimensional profile showing how much X-ray intensity was lost as it traveled through the patient along that particular path. This is measurement data—not yet an image. The Mathematical Problem: Imagine you're trying to figure out what's inside a locked box by shining a light through it from different angles and measuring how much light comes out the other side. With enough angles, you can mathematically solve for what's inside. CT does exactly this: the collection of projections from many angles provides enough constraints to solve for the internal structure. From Projections to Images: Reconstruction After acquiring all the projection data, a computer must reconstruct the cross-sectional image. Two main approaches are used: Filtered Back-Projection Filtered back-projection is the classical reconstruction method, still widely used because it's fast and well-understood. The algorithm works in two steps: Back-projection: Imagine each projection is "smeared" back across the image space along the direction it came from. If a detector measured high attenuation at angle 0°, that high attenuation is smeared backward across the entire image at 0°. If another detector measured attenuation at 90°, it's smeared at 90°. By back-projecting all projections simultaneously, overlapping smears add together. Regions that truly had high density get reinforced by many projections saying "dense," while noise and artifacts cancel out somewhat. Filtering: Here's the critical part: back-projection alone produces blurry images. So the algorithm applies a mathematical filter to sharpen the image and suppress blur. Different filters can emphasize different features (bone detail versus soft tissue detail, for example). Why filter? Without filtering, back-projection creates a blurry image because each projection contributes the same intensity across its entire path. The filter corrects this by enhancing edges and removing the blur. Iterative Reconstruction Algorithms Modern CT systems increasingly use iterative reconstruction algorithms, which work fundamentally differently: Start with an initial guess at what the image looks like Mathematically simulate what projections that image would produce Compare the simulated projections to the actual measured projections Adjust the image estimate to better match the measured data Repeat steps 2-4 many times until convergence This approach has two major advantages: Better noise management: Iterative algorithms can reduce image noise while maintaining detail, allowing lower radiation doses Artifact reduction: The algorithm can be designed to suppress common artifacts The trade-off is computation time—iterative reconstruction requires many more calculations than filtered back-projection. Hounsfield Units: Quantitative Tissue Characterization One of CT's most valuable features is that it produces quantitative numbers, not just pictures. Each voxel (3D pixel) in a CT image is assigned a Hounsfield Unit (HU), a standardized scale that represents tissue density relative to water. The Hounsfield scale is defined as: $$HU = 1000 \times \frac{\mu{tissue} - \mu{water}}{\mu{water}}$$ where $\mu$ represents the X-ray attenuation coefficient. This means: Water = 0 HU (the reference point) Air ≈ -1000 HU (very low density) Bone ≈ +300 to +1000 HU (high density) Soft tissues ≈ +40 to +80 HU (slightly denser than water) Why this matters: Unlike conventional X-rays, which show only shadows, CT gives you actual numerical tissue density values. This enables: Precise tissue characterization (distinguishing fat from muscle, for example) Detection of subtle density changes that indicate disease Reproducible measurements over time for follow-up studies Image Quality Factors The quality of a reconstructed CT image is influenced by several factors: Reconstruction Filter Choice: Different filters optimize for different clinical goals. A "sharp" or "bone" filter emphasizes edges and detail but may increase noise, while a "smooth" filter reduces noise but sacrifices sharpness. The radiologist or protocol designer chooses the filter based on clinical need. Slice Thickness: Thinner slices provide better detail but require more radiation exposure and produce more noise. Modern multi-detector systems can acquire slices as thin as 0.5 mm. Noise and Contrast: These are always in tension. Noise (random fluctuations in pixel values) can be reduced by using more radiation, using smoother filters, or using iterative reconstruction. Contrast (the ability to distinguish different tissues) depends on tissue density differences and reconstruction settings. Radiation Dose and Safety CT delivers ionizing radiation to patients, and this is a critical consideration. Why CT Delivers More Dose Than Conventional X-rays: A single CT scan exposes the patient to more radiation than a conventional X-ray because: The X-ray beam is continuous (not a single pulse) The patient is scanned repeatedly from multiple angles Modern protocols often acquire overlapping slices for better 3D reconstruction A typical abdominal CT scan delivers roughly 5-10 mSv of radiation, compared to 0.1 mSv for a chest X-ray. Dose-Reduction Strategies: Lowering tube current: Reducing the electrical current in the X-ray tube decreases X-ray output proportionally. This reduces dose but increases image noise. Automatic Exposure Control (AEC): The scanner adjusts X-ray output in real time based on patient size and anatomy. A thinner patient requires less radiation to achieve the same image quality as a larger patient. AEC systems monitor the actual attenuation and adjust the tube current accordingly—delivering just enough dose for diagnostic quality without excess. Iterative reconstruction: As mentioned earlier, these algorithms can produce diagnostic images at lower doses than filtered back-projection, because they handle noise better. Protocol optimization: Using appropriate slice thickness and spacing for the clinical question—thinner slices than necessary waste dose. Risk-Benefit Balance: The ionizing radiation from CT carries a small but real risk of inducing cancer, particularly in younger patients. However, the diagnostic benefit of CT—detecting tumors, internal bleeding, fractures, and other serious pathologies—often far outweighs this risk. Clinical judgment is essential: CT should be used when needed, but not performed unnecessarily. Clinical Applications and Advantages CT excels at detecting a wide range of pathologies: Acute trauma: Fractures, internal bleeding, organ damage Oncology: Tumor detection, staging, and treatment planning Infectious disease: Pneumonia, abscess location Cardiovascular: Pulmonary embolism, aortic dissection, coronary artery disease Neuroimaging: Stroke, hemorrhage, tumor Key advantage: CT provides excellent spatial resolution and tissue density discrimination in a fast examination, making it the go-to modality for many acute and complex diagnostic problems. The quantitative Hounsfield units also enable specific diagnoses (for example, a kidney stone's density can help identify its composition). <extrainfo> Advanced Reconstruction and Analysis Modern CT systems include sophisticated post-processing software that goes beyond simple image reconstruction: 3D rendering: Volumetric data can be rendered into 3D views, showing complex anatomy from any angle Automatic segmentation: Software can automatically identify and isolate specific organs for analysis Quantitative analysis: Measurements of lesion size, volume, and density changes can be extracted automatically, useful for research and follow-up monitoring These tools are particularly valuable in surgical planning, radiation therapy planning, and research applications studying bone density, organ perfusion, and treatment response. </extrainfo>