2 Nov 2016
Working principles of CT and MRI
Olivier Taeymans looks at the similarities and differences between the two techniques – and explains why both are needed in a state-of-the-art imaging department.

A 3D volume rendered reconstruction of a dog.
Both CT and MRI acquire sectional (tomographic) slices through the body, avoiding superimposition of surrounding structures.

They also preserve a 3D representation of the patient, as opposed to traditional x-ray where the entire thickness of the body is displayed on a single 2D image.
Tomographic images, therefore, produce a much more detailed examination of internal organs, particularly when surrounded by dense bony structures, such as the rib cage.
Data acquisition is fundamentally different between CT and MRI; CT uses x-rays, whereas MRI uses a combination of a strong magnetic field and radio waves.
CT use and interpretation
For the sake of not overcomplicating matters, only helical/spiral scanners will be discussed in this article; this is almost the only type used in veterinary medicine nowadays.
A CT gantry looks like a large doughnut from the outside. It contains an x-ray tube that spins at phenomenal speed; three to four rotations per second. Opposite the x-ray tube are one (single slice CT) or multiple (2 to 320) rows of detectors (multi-detector row CT or multi-slice CT; abbreviated MDCT), aligned along an arch. These rows of detectors remain in a fixed apposition to the tube.
A fan-shaped, but very thinly collimated, x-ray beam is transmitted through the patient. It covers the radius of the detector arch and, therefore, the patient width. The collimation is in the z-axis (in the axis of the table) and determines the slice thickness of the images.
With the x-ray beam continuously on, the patient steadily progresses through the gantry on the table, resulting in a spiralling/helical data acquisition. This results in a volume of data, as opposed to sequential individual axial images.
Apart from increased speed of acquisition, the enormous advantage is the data volume can be reconstructed in any 3D plane, or represented as a large surface-rendered block of tissue (Figures 1 and 2).
The detectors, located in the shadow of the patient, measure the degree of the beam’s attenuation following transmission through the body. Interrogating structures from several angles, resulting from tube rotation, creates numerous attenuation spectra. These are then back-projected on to an image matrix, using complex mathematical algorithms. Spatial location and average attenuation value of each pixel in the matrix make up the image.
The degree of beam attenuation, in direct proportion to density of different tissue types, is measured in Hounsfield units (HU), with water having a default value of zero. Denser tissues have higher attenuation values – for example, cortical bone about 2,000HU – and are displayed in strong pixel brightness.
Gas, meanwhile, is at the opposite end of the spectrum – about -1,000HU – and is displayed in low pixel brightness; hypoattenuating structures are dark on CT images. Soft tissues – such as muscle, fat, normal organ parenchyma, fluid, connective tissue, nervous tissue and tumour tissue – all have similar medium pixel brightness (20HU to 80HU) and are, therefore, more difficult to further differentiate. The latter is referred to as poor contrast resolution.

CT interpretation, therefore, relies more on large differences in pixel values, so is preferred for assessing structures with high intrinsic contrast, such as bony (dense cortex versus sparse medullary cavity) and pulmonary (dense cardiovascular structures versus sparse air-containing lung) structures.
Differentiation among soft tissues can partially be remediated by windowing the images differently; for example, by:
- changing contrast and brightness (referred to as window width and window level)
- using different reconstruction algorithms
- (most efficiently) acquiring a second data set after intravenous injection of iodinated contrast medium
Vascularised tissues will attenuate more as a result of their higher contrast uptake, making them more distinct from other soft tissues.
Thanks to the rapid image acquisition speed of current scanners and using specific timings, or bolus-tracking technology, it is also possible to accurately display the contrast in arteries, portal vasculature and systemic veins separately.
A power injector is used to remotely inject the contrast from within the operator room, carefully synchronising the start of predefined countdowns, or bolus-tracking sequences, with the start of injection. Such angiography is particularly useful in diagnosing portosystemic shunts, neoplastic vascular invasion and pulmonary thromboembolism.
Image acquisition in CT is speedy – acquiring a thorax of a dog in less than 30 seconds – meaning some patients can be scanned under sedation. General anaesthesia remains a necessity when scanning the chest though, as lungs need to be inflated to assess pulmonary opacity and subtle lesions that might otherwise silhouette with the surrounding lung parenchyma. It also avoids pulmonary atelectasis and motion blur.
Notwithstanding being speedy, very large amounts of data are acquired, hence the high number of reconstructed images produced, both leading to long interpretation times and large storage needs.
Working with MRI

There are two main types of MRI equipment: low-field and high-field magnets. The magnet strength is expressed in Tesla (T), with 1T corresponding to 20,000 times the earth’s magnetic field strength.
High-field magnets have the appearance of a long gantry with narrow tunnel. Low-field magnets have an open design and resemble a big letter U on its side (Figure 3).
The magnetic field in high-field magnets is created by a large electromagnet, whereby an electrical current flows through a coil. The coil is cooled with helium to cryogenic temperatures (below -200°C), achieving superconductivity in the wirings. This increases the number of flowing electrons, thereby achieving field strengths above 1T.
Helium is an increasingly rare commodity, with regular helium refills considerably adding to running costs. New technologies, such as zero boil-off heads, markedly decrease the need for helium refills on the newer models.
Running costs will, however, always remain high due to their extremely high electricity consumption.
Low-field equipment is cheaper, both in running and purchase cost, and does not rely on superconductive coils. It consists of a large U-shaped permanent magnet, creating field strengths between 0.2T and 0.4T.
A disadvantage of using large permanent magnets is their weight (four tons to six tons), which is considerably more than high-field units. This requires floor reinforcement in almost all cases, adding to installation cost.
The physics behind image formation is extremely complex. Essentially, image creation relies on the large amount of hydrogen atoms (H+) present in living tissues. As these atoms are positively charged, they act as tiny magnets, which are getting aligned by the strong external magnetic field in the gantry.
After H+ atoms are aligned with the magnetic field, a radio frequency pulse is emitted, which temporarily tilts the axis of these small magnets. Once the pulse terminates, the H+ atoms slowly realign to their original axis.
Interactions among individual atoms and their surrounding, which depends on the molecular composition of different tissue types, influence the realignment process and forms the basis for differentiating tissue types.
While realigning, weak radio pulses are emitted by the different tissue types, which are received by an antenna (coil) and translated to variable pixel brightness (intensity). To locate where the signal originated from, the patient must remain perfectly immobile – hence the need of general anaesthesia.
Because of this fundamental difference with other imaging modalities, with signals originating from within, and depending on the composition of the tissue, MRI has the best contrast resolution (differentiating different tissue types).
However, spatial resolution – for example, the ability to differentiate two closely spaced objects – is not as good as other imaging modalities, as accurate localisation of different structures relies on amplification of very weak radio signals that are prone to suffer from distortion, amplification of background noise and interference from outside radio signals.
A Faraday cage around the magnet blocks outside radio signals, but, to receive sufficient signal from interrogated tissues, thick slices (3mm to 5mm) need to be obtained through the body, further reducing spatial resolution.
Coils/antennas come in different shapes and sizes to closely fit the anatomy of the patient, with better signal-to-noise ratios resulting from tighter fittings. Unfortunately, dedicated veterinary coils are often not available, leaving us with no other option than trying the best possible match.
Differences among tissues can further be exploited using different sequences of initiating/tilting radio pulses. Routinely used sequences are spin-echo T2, spin-echo T1, gradient-recalled echo (GRE), inversion recovery (STIR) and fluid attenuating sequence (FLAIR).
Examples of different signal intensities of different bodily tissues are displayed in Table 1. Besides providing good tissue contrast, this table also illustrates, by combining multiple sequences, how we can identify different tissue types/composition.

* Signal intensity varies and changes over time, relating to changes in molecular composition (oxyhaemoglobin, deoxyhaemoglobin, methaemoglobin, haemosiderin).
This excellent capability in differentiating tissues practically means MRI is used to detect subtle alterations in soft tissue composition, such as for:
- the central nervous system (for example, grey matter versus white matter, brain versus cerebrospinal fluid, or disc material versus spinal cord)
- musculoskeletal structures (cancellous bone, cortical bone, tendon/muscle and joint fluid interfaces)
- to a lesser extent, for abdominal organs (motion is a major limitation here).
Because of the different tissue types in or around joints, or because of the abnormal signal intensity of neoplastic or infectious lesions, MRI can still be used for diagnosing orthopaedic conditions, despite general belief MRI is not good for assessing bony structures.
However, where fine detail (high spatial resolution) is needed for picking up small structures, such as fragmented coronoid process and nasal turbinates, CT will be preferred.
Also, where no protons are present, such as in the majority of the lungs, or inside intact cortical bone, no or almost no signal/information will be obtained from those structures. However, CT will again perform brilliantly thanks to the strong natural contrast in lungs – air versus blood vessels for the lungs, highly attenuating cortical bone versus much less attenuating cancellous bone and soft tissues.
Hopefully it is clear from the examples CT and MRI complement each other perfectly – and both pieces of equipment are needed in a state-of-the-art imaging department.
Acknowledgement
The author thanks Matt Matiasovic for his careful revision of the manuscript.