Images in computed tomography (CT) are reconstructed mathematically from data derived from multiple X-ray projections through the region of interest in the patient . Originally data were acquired one slice at a time; after each exposure the patient would move a specific distance along the scanner table and the procedure would be repeated. Numerous technical advances have dramatically increased the speed and quality of the technique since its announcement in 1972 (for example improvements in detector sensitivity and efficiency, power of the X-ray beam, computational power and speed). Probably the most important innovation has been the development of slip ring technology, which allows the continuous rotation of the X-ray source as the patient moves through the scanner. This is the basis of spiral (helical) scanning, named after the helical pattern the X-ray beam traces around the patient. This has been followed by the development of CT scanners employing more than one bank of detectors (multislice or multidetector CT) . This technique has permitted substantial increase in speed and quality of scanning. Thin slices (1 mm or less) can be obtained over large areas and patients can be scanned in a matter of seconds (for example the entire chest during suspended inspiration). It also facilitates high quality reconstruction in any plane or as a three-
dimensional representation. These advantages can be further enhanced by increasing the number of detectors. Initial experience was with four banks of detectors (four slice scanners). Currently 16 and 32 slice scanners are commercially available; 64 and 128 slice scanners are being developed. The rapid acquisition of good quality images is especially useful in patients who are unwell and restless or with bulky tumors which cause them to swallow continuously.
The protocol for CT scanning will vary between centers, relating to local experience and equipment. The underlying principle for any head and neck tumor is to obtain high quality images of the primary tumor and adjacent areas of potential direct spread and to extend the scan to screen the area of predicted lymph node drainage. Conventionally CT protocols could be described in terms of slice thickness, which would correspond to the actual thickness of the slice that was exposed at the time of the investigation. So relatively thin slices would therefore be obtained following contrast through the primary tumor, say 3-5mm , while thicker slices of the order of 8-10 mm were recommended for the rest of the neck. Some authorities would also suggest a relatively thick slice scan (8-10mm) through the area of interest before contrast was administered. Axial scans are routinely obtained but coronal images should also be generated to demonstrate particular features such as contiguous extension into the mediastinum. When considering multislice scanners (which are likely to become universally available in the not too distant future) the principle of obtaining relatively thin slices through the area of interest is retained but is derived in a more virtual way. It is more useful to think in terms of the scanner acquiring a volume of data which can be displayed in a variety of ways. If the table on which the patient lies moves rapidly through the scanner as the X-ray tube rotates, then the effective slice thickness increases. Conversely if the table movement is slower, then the effective slice thickness falls. If a 16 slice scanner is considered; each row of detectors is around 0.75 mm thick. Therefore the thinnest effective slice thickness that can be acquired is 0.75 mm. This would only be acquired if the scanner is set so that the table on which the patient lies moves forward 12 mm for each rotation of the tube (this can be referred to as a pitch of 1). This would be associated with a considerable radiation dose to the patient but would produce high quality data through the area examined. In practice for head and neck work a larger effective slice thickness of 1.5mm is recommended . This still permits excellent quality reconstructions in any plane but involves half the radiation dose (obtained by having the scanner set so that the table advances 24 mm per tube rotation; pitch of 2). In routine clinical use axial reconstructions plus coronal and (for specific questions) sagittal reconstructions are sufficient. Iodinated material used for CT contrast will reduce subsequent 131I uptake by thyroid tissue for 4 to 8 weeks . Therefore the timing of the sequence of examinations should be thought out in advance or the alternative modality of MRI considered. In most circumstances where CT is performed for thyroid cancer it would appear logical to extend the examination to include the chest to search for pulmonary metastases.
The principles underlying magnetic resonance imaging (MRI) are relatively simple, although the applications can be extremely complex . The following summary merely attempts to give some idea of the underlying concepts. All nuclei carry a positive charge (by virtue of the protons present) and are known to spin. A spinning positive charge behaves like a small magnet. In biological tissue by far the most numerous and important nucleus is hydrogen, containing a single proton. Although functional MRI (largely experimental) may image different nuclei (notably phosphorus), clinical MRI is essentially concerned with producing images that reflect the distribution of hydrogen nuclei and to a certain extent their environment (what sort of chemical bonds and structures they are involved in). Normally the spinning protons are randomly aligned, as are the small magnetic fields they produce, and therefore there is no net magnetic effect. If, however, the patient is placed within a huge magnetic field (from 0.15 to 3.0 tesla or more in some experimental scanners) there will be a predominant alignment along the axis of the field. This is referred to as the magnetization of the sample. In order for an MRI signal to be produced the magnetization must be deflected from this parallel state. This can be achieved by the application of a second magnetic field perpendicular to the original field. This deflects the magnetization. As soon as the net magnetization is deflected from its initial state it starts to precess like a spinning top, due to the angular momentum of the spinning protons. In order to produce a deflection of the net magnetization the deflecting field is made to vary sinusoidally at the same frequency as the magnetization pre-cesses. This can be regarded as a resonance phenomenon. The frequency of sinusoidal variation is in the radiofrequency range and hence the deflecting field may be referred to as a radio-frequency field. The coil that produces the deflecting field is often referred to as the radiofrequency (RF) coil. When the deflecting pulse is switched off there is a tendency for the magnetization to return to the original axis. As it moves towards its original alignment it induces an electric current in the RF coil. This current is the basis for one of the major components of the resultant MR image and is often referred to as the free induction decay (FID). Three important parameters affect the FID, namely T1, T2, and spin (proton) density. Before the deflecting pulse is transmitted the spinning protons are in a low energy state. After the pulse they have been deflected to a high energy state. They will gradually lose energy as they return to their initial state, a change that shows an exponential time course and has a constant referred to as the T1 relaxation time. This may also be called the spin-lattice relaxation time, indicating it is related to energy transfer from excited protons to the environment. There is also a tendency for the numerous components of the magnetization (the spinning protons) to dephase with respect to each other. This dephas-ing shows an exponential pattern over time, with a time constant of T2. The T2 relaxation time is a property of the tissue under investigation and it generally increases with disease. Alternatively it may be called the spin-spin relaxation time, the name indicating it is related to interactions between the spinning protons. The FID is proportional to the spin density, that is, the FID is related to the number of protons involved in the MR process per unit volume.
Utilization of these parameters to form an MR image becomes quite complex since the above account considered only a single deflecting pulse. The production of images usually involves a sequence of pulses. Varying the strength and timing of the transmitted pulses, the timing of reception of the induced signals (time to echo, TE) and the rate of repetition of the whole sequence (time to repetition, TR) determine the characteristics of the resultant image. Generally sequences with short TEs and TRs are T1 weighted (T1W), sequences with long TEs and TRs are T2 weighted (T2W). Various designer sequences have evolved including fat suppression sequences, which are particularly useful to induce lymph nodes and other areas of pathology to stand out against background fat.
An intravenous contrast agent has been evolved for use in MRI. This is based on gadolinium which is chelated (Gd-DTPA) to make it safe to administer . Gadolinium is a paramagnetic material that shortens the T1 and T2 relaxation times.At the normal doses used (0.1-0.2mm per kg body weight) only T1 shortening is observed. Areas that take up Gd-DTPA therefore appear of increased signal intensity (brighter) on T1W scans. Tissue uptake of gadolinium is related to perfusion (avascular structures will not significantly increase their signal) and the integrity of the small vessels. Tissues that enhance include intensely vascular tissues such as nasal mucosa and diseased tissues such as areas of tumor and inflammation.
MRI and CT are powerful imaging tools and the choice of which modality to use is not always straightforward. MRI shows exquisite soft-tissue contrast  and will therefore better demonstrate tumor invasion of structures such as strap muscle, larynx, and esophagus. It does not involve ionizing radiation and this may be important, especially in younger patients. Traditionally it had the disadvantage of lower resolution and higher susceptibility to motion-induced degradation of images. However, although these problems have not been entirely overcome, each generation of scanner continues to improve image resolution and reduce scanning time. MRI offers imaging in any plane, which was an advantage over CT, but with the advent of multislice CT scanners this is equally offered by CT. MRI scanners are generally more claustrophobic than CT scanners and some individuals are unable to tolerate the procedure. Fine bone detail is not seen on MRI, although this may not matter in most situations and MRI is very sensitive to tumour infiltration of marrow. However, areas of calcification are much better seen on CT, which may be helpful in assessing thyroid carcinoma (although since the diagnosis is still histological this may not matter that much). Some foreign bodies are potentially dangerous in the powerful magnetic field of the MR scanner [9,10]. These include pacemakers, spinal dural electrodes, some shrapnel, heart valves, intradural clips and cochlear implants, and some intraorbital foreign bodies . The biggest advantage that CT has currently is the ease with which the chest can be assessed at the same time as scanning the neck. However, given the constant and rapid advance of these fields of technology this is likely to be a fleeting situation.
MRI scanning protocols vary considerably between centers, partly related to the availability of scanning time, the preferences of the radiologist and the capabilities of the MRI scanner (which may vary in field strength and software). As with CT, high quality images of the primary tumor and its potential sites of direct invasion should be obtained and the area of predicted lymph node drainage should be included . Scanning should be performed in multiple planes and in some centers will include sagittal as well as coronal and axial images. A T1W sequence is routine in most institutions, often repeated in two planes after intravenous contrast. A sequence in which signal from fat is largely absent is useful; either a STIR sequence or a post contrast fat suppression sequence is useful to facilitate demonstration of abnormal lymph nodes.
MRI scanners continue to improve in terms of speed and image quality. Scans used for radio therapy planning are generally in the 0.2 to 1.0 T (tesla) range and based on resistive magnets. Most diagnostic scanners are based on a super-conducting magnetic field of 1.0 to 1.5 T. Larger field strengths, generally 3 T (but up to 9T or more in some research situations), may be utilized. There are potential safety problems (nerve stimulation, for example, with very high field strengths) and a current lack of evidence to suggest routine imaging will be of better quality (brain imaging possibly excepted). However, scanning at this strength does facilitate the performance of magnetic resonance spectroscopy. This technique has largely been a research tool in neurological conditions but is also being used to assess tumor metabolism. Relatively low field spectroscopy (1.5 T) is also becoming technically feasible. Potentially this technique may detect malignancy in difficult situations (low volume and/or within previously treated sites) and offer a potentially extremely useful tool in oncology, especially in the detection of recurrent tumor and the effects of treatment .
The normal thyroid gland has homogeneous high density on CT due to the high iodine content. On MRI it appears as an intermediate signal on T1 weighted sequences and intermediate to low on T2 weighted sequences. It shows intense diffuse enhancement with intravenous contrast on both CT and MRI.
Was this article helpful?