In recent years, magnetic resonance imaging has become widely accepted as a powerful imaging tool1. It relies on the nuclear magnetic resonance of hydrogen atoms found throughout the human body to produce a high-quality three-dimensional image of bodily structures using an MRI machine (see figure 5)2. MRI also gives excellent soft tissue contrast and, as a result, is highly useful in the diagnosis of conditions such as cancer, where it is the investigation of choice for patients with primary CNS tumours3. Figure 1 shows the appearance of a primary brain tumour (in this case, glioblastoma) in an image produced using MRI.
Figure 1 - MRI scan showing a large tumour in the pineal region of the brain.An MRI scan shows a large tumor in the pineal region of the brain
The white arrow indicates the location of the tumour, which can be seen in sharp contrast to the surrounding tissue. Taken from (4).
1.2 Physical Principles of MRI
To understand how an image is produced using MRI, some understanding of the underlying physical concepts is required. These concepts are quite complex, so will only be explored in superficial detail in this report. As mentioned previously, MRI relies on the magnetic properties of hydrogen atoms. The nucleus of a hydrogen atom is a single proton ("H+"; a spinning, positively-charged particle), which can be thought of as a bar magnet with north and south poles (see figure 2) since the particle itself has a small magnetic field1. Therefore, when a patient is placed inside an MRI machine, which produces a strong magnetic field using a superconducting magnet, the majority of hydrogen atoms within the patient align in the direction of the MRI machine's magnetic field1 and so their "net magnetisation vector", M0, is in the same direction as the strong magnetic field, B0 (see figure 3).
Figure 2 - A diagrammatic representation of a hydrogen atom nucleus.
"N" indicates the "north pole" and "S" indicates the "south pole". "H+" indicates the hydrogen atom nucleus itself (a single proton). The vertical arrow indicates the proton's spin axis. Adapted from (1).
Figure 3 - A diagrammatic representation of the alignment of hydrogen nuclei in a strong magnetic field.
Figure 3(a) shows the hydrogen nuclei in their resting state (i.e. outside a strong magnetic field). Figure 3(b) shows the hydrogen nuclei in the strong magnetic field generated by an MRI machine - the majority are aligned in the direction of the magnetic field, B0. Adapted from (1).
Whilst aligned in the strong magnetic field, the hydrogen atoms do not remain stationary - they "precess" around the direction of the field1. This is a type of movement in which the bottom of the hydrogen nuclei's spin axis describes the apex of a cone shape (see figure 4). The frequency of precession is known as the Larmor frequency1 and is proportional to the magnetic field strength5. It is calculated using the formula ω0 = γB0, where ω0 is the Larmor frequency, γ is the gyromagnetic ratio (a nuclei-specific constant; for hydrogen, γ = 42.6MHz/Tesla) and B0 is the strength of the applied magnetic field5.
Figure 4 - Nuclear precession.
The hydrogen atom nucleus precesses around the direction of the magnetic field (or, more accurately, the direction of the net magnetisation vector), M0. In this way, the bottom of the nuclei's spin axis describes the apex of a cone.
A radiofrequency (RF) coil is now used to produce a second magnetic field which is perpendicular to the strong magnetic field, B01. This field is known as the RF pulse and has the same frequency as the Larmor frequency, resulting in the nuclear magnetic resonance (NMR) of the precessing hydrogen nuclei1. This causes them to move into a high-energy state which is perpendicular to the direction of B0, meaning that their net magnetisation vector (M0) is now in the transverse plane1.
When the RF pulse is removed, the energy absorbed by the protons is dissipated into their surroundings - this is known as T1 relaxation. T2 relaxation is the de-phasing of the precession of the protons (as the RF pulse brought the precession of the protons into phase)1. Scans can be T1- or T2-weighted depending on the type of image required - a T1-weighted scan provides excellent anatomical definition, but is not very sensitive to the presence of pathology, whereas T2-weighted scans are quite sensitive to the presence of pathology1. T1 and T2 relaxation times differ significantly in tissues (often, T1 is longer8), and so the scan is weighted by adjusting the RF pulse repetition frequency to a value such that the pulses do not interfere with the relaxation which needs to be visualised8. Relaxation times can be affected by contrast agents, which are discussed later8.
As the protons relax, the net magnetisation vector returns to its original direction and induces a current in RF receiver coils built into the MRI machine1. This current is known as the MR signal and is the basis for the production of an image, as complex computer analysis can determine the location of the hydrogen nuclei which produced the signal. The strength of the signal depends on several factors - one of these factors is the proton density of the tissue being imaged1, which allows tumours to be identified since their density differs from the surrounding normal tissue. For example, astrocytoma - a type of brain cancer - is hypodense compared to normal brain tissue and so will produce a "hyposignal"; a weak MR signal due to its low tissue density. Depending on the type of scan (i.e. T1- or T2-weighted), this signal will appear either darker or brighter than the surrounding tissue on the image7, allowing the tumour to be identified.
By altering the strength of the magnetic field at certain points in the scanner, specific areas of the body can be imaged in isolation if necessary. This is done using smaller magnets (or so-called "gradient coils" in addition to the main superconducting magnet to create a "gradient field". This will result in different Larmor frequencies at different points in the scanner, and so a specific area can be imaged in isolation by applying an RF pulse with a frequency equal to that area's Larmor frequency5.
1.3 Advantages and Disadvantages of MRI
The main advantage of MRI is that, as mentioned previously, it provides excellent soft tissue contrast1. Other advantages are the lack of artefacts due to bones1, the ability of the MRI scanner to obtain images in any plane1 (and the fact that images can be obtained in the supine position, making it highly useful for imaging of the spine and this, combined with its excellent soft tissue contrast, makes MRI highly useful for the imaging of spinal disorders). Additionally, MRI does not utilise ionising radiation, which results in less risk to patients1 and means it is an ideal imaging method for pregnant patients and young children6.
The main disadvantage of MRI is the cost - the cost of equipment and maintenance is very high1, and considering that a scan can take up to 90 minutes6 patient care must be weighed carefully against these costs. MRI also produces a variety of artefacts in soft tissue1, can not be used where a patient has a pacemaker6 or magnetic metals inside their body (due to the strong magnetic field generated by the scanner)1, can not produce fine bone detail1 and is less sensitive than other imaging methods at detecting certain substances (e.g. MRI is less sensitive at detecting calcification and haemorrhage than CT imaging1).
1.4 Recent developments
Contrast Agents - contrast agents are chemicals which increase the MR signal, producing a clearer image1. They do this by shortening the relaxation time (either T1 or T2, depending on the agent) and increasing the rate of MR signal transmission8. Traditionally, gadolinium chelate contrast agents (which are extracellular agents) have been the only choice available9, but new contrast agents are in development. This is expanding the applications of MRI, as more systems are able to be imaged if the correct contrast agents are available9.
Magnetic Resonance Angiography (MRA) - flowing blood can be shown as signal void (black) or increased signal (white). Using this information, a computer can construct a three-dimensional image of the blood vessels, which is useful in identifying and diagnosing ischaemia1.
Echoplanar Imaging - an ultrafast MRI technique allowing the visualisation of physiological events in real time, for example through BOLD (blood oxygen-level-dependent imaging), diffusion weighted imaging (DWI), perfusion-weighted imaging (DWI) or MR spectroscopy (where certain molecules are identified in a certain tissue; has uses in imaging of dementia and the progression of diseases)1.
Figure 5 - A diagrammatic representation of an MRI scanner, showing the patient inside the machine and inside the strong magnetic field created by the superconducting main magnet. Adapted from (1).
2.0 Positron Emission Tomography (PET)
2.1 Introduction
Positron emission tomography (PET) is a relatively new technology which relies on positron-emitting nuclides. Such nuclides were discovered in the 1940s and PET imaging devices were developed in the 1970s, but PET has only recently seen use in clinical applications, having been used mainly in research since its development10.
2.2 Physical Principles of PET imaging
As with MRI, it is useful to understand the physical principles of PET imaging before exploring its relevance in the clinical setting. Unlike MRI, PET imaging is used purely to assess bodily function - it cannot image structures, and so is often combined with CT or MRI imaging so that bodily functions can be "mapped" to a particular region of the body10.
As mentioned previously, PET relies on the use of positron-emitting nuclides - these nuclides are injected into the body and their location is monitored using a PET scanner to investigate their uptake by organs and tissues10. Positrons are the antiparticles to electrons; therefore, when a positron meets an electron in the body, annihilation occurs and two gamma photons are produced11. This can be represented as a particle equation:
e+ + e- ïƒ Î³a + γb
Where e+ is the positron, e- is the electron and γa and γb are the two gamma photons.
The two gamma photons each have 511keV of energy and travel in opposite directions10. The recognition of the photons by detectors on each side of the body allows the location of the annihilation to be determined and so this is how the nuclide is monitored after being injected into the body10. Most commonly, these detectors take the form of a full or partial ring around the patient (see figure 6), so as to ensure that all or most of the gamma photons are detected10.
Figure 6 - Diagrammatic representation of a patient inside a PET scanner.
The patient is seen lying supine from below. The positron emitter has caused the annihilation of an electron and a positron, resulting in the production of two gamma photons, γa and γb. These are detected by the ring-shaped detector, and the time interval between their detection is used to determine the position of the annihilation in the patient's body (indicated by a yellow star on the diagram). Adapted from a description in (10).
The main nuclide used in clinical settings is 18F-tagged deoxyglucose, known as 18FDG. Malignant tumours tend to show an increased rate of glucose metabolism, likely due to a large number of glucose transporters and a reduced rate of glycolysis and a reduced rate of dephosphorylation of glucose-6-phosphate (G6P; phosphorylated glucose which can later be converted to pyruvate via glycolysis)10. This means that FDG is readily taken up by cells in the body. FDG undergoes glycolysis in tumours, but the dephosphorylation of FDG-6-phosphate is slow, so FDG accumulates in the cell10. The level of FDG in the cell peaks and plateaus when the rate of FDG uptake is equal to the rate of dephosphorylation of FDG-6-phosphate; in studies, this generally occurs 45-60 minutes after injection, so imaging is carried out after this time. The imaging will then indicate the uptake of the FDG - areas with high or low uptake compared to the mean uptake in the rest of the body are investigated10, as low uptake may indicate dying tissue and high uptake may indicate a tumour for the reasons given above (see figure 7).
Figure 7 - A combined PET/MRI scan of a tumour in a lab mouse.Image: scan of brain with colored spots
The colours represent the concentration of the nuclide in that area - red indicates that a relatively large amount of the nuclide has been taken up, whilst blue and yellow indicate a relatively low amount has been taken up. A dark area indicates that no nuclide was taken up in that area. In this image, the white arrow points to a "hole" in the colours; this likely indicates dying tissue. The red area indicates a tumour. Taken from (12).
2.3 Applications of PET Imaging
In the past, PET imaging was mainly used to investigate brain metabolism due to the small size (and therefore, low cost) of the scanner required10. However, PET now has a wide range of applications, particularly in the diagnosis of cancers.
Lung cancer - PET is useful in staging non-small cell lung cancer, where surgical excision is an option and so careful assessment of the tumour is required. In small cell lung cancer, systemic chemotherapy is the first line treatment and so PET imaging is less useful. PET is also good for discriminating between benign and malignant tumours, but may produce false negatives when the tumour size is less than 1cm (since analysis of the image produced is more difficult with decreasing tumour size)10.
Colorectal cancer - here, PET imaging is useful for monitoring patients for recurrence of relapsed cancer, or for locating distant metastases10.
Lymphoma - most lymphomas show high uptake of FDG and so PET imaging is useful for its diagnosis. Lymphoma patients with the greatest uptake of FDG at presentation often have the worst prognosis10.
Head/neck tumours - again, head and neck tumours show high uptake of FDG. In this case, PET imaging is particularly useful in locating primary occult (hidden) tumours in patients presenting with metastatic tumours in the neck region10.
Breast cancer - can be useful in the diagnosis of breast cancer, but as mentioned previously may give false positives if the tumour size is less than 1cm. Again, PET is useful for locating distant metastases in breast cancer patients.
Other applications - PET imaging is useful for detecting metastasis in melanoma, and in patients with refractory epilepsy metabolic brain imaging may be useful in determining the region of the brain causing seizures so that it can be surgically excised (since, during a seizure, the tissue causing the seizure will have increased metabolic activity and therefore increased FDG uptake)10.
3.0 Patient Outcomes
Having already explored the techniques involved in obtaining diagnostically useful images using both MRI and PET imaging, I will now explore how these techniques have affected patient outcomes in a clinical setting. For MRI, I will look at how patient outcomes have improved in breast cancer diagnosis and for PET, I will look at how patient outcomes have improved for cancer in general.
3.1 Effects of MRI on Breast Cancer Patient Outcomes
It is a well-known fact that women diagnosed with unilateral breast cancer have an increased risk of developing cancer in the contralateral breast13. As such, clinical breast examination (CBE) and mammography have long been used in women diagnosed with unilateral breast cancer to improve the detection rates of contralateral cancers13. However, both clinical breast examination and mammography produce false negatives - in the 1990s, mammography alone was shown to produce a 1-3% increase in the number of contralateral breast cancers detected, but up to 10% of patients later went on to develop a contralateral breast cancer13. As a result, these patients required a second round of cancer therapy which would not have been required had the contralateral cancer been discovered at the time of diagnosis13.
Preliminary studies suggested that MRI could have a higher detection rate of previously occult breast cancers than mammography and CBE combined13. A 2008 study aimed to prove this link. A cohort of 969 female participants were selected for MRI examination - to be eligible for the study, they had to have been diagnosed with unilateral breast cancer within the past 60 days, and had to have received normal CBE and mammographic findings for the contralateral breast within the past 90 days.
Contrast-enhanced MRI was performed, and the images were assessed for the presence of breast cancer using the BI-RADS (Breast Imaging Reporting and Data System) classification system, which uses a 0-5 scale to classify breast cancer - 0 being "needs additional assessment", 1 being "negative", 2 being "benign", 3 being "probably benign", 4 being "suspicious abnormality" and 5 being "highly suggestive of malignancy". For scores of 0 or 3, further imaging was necessary to determine the final BI-RADS score. A definitive diagnosis of breast cancer was based on histological examination of a breast biopsy, and included both ductal carcinoma and ductal carcinoma in situ. Additionally, the cancer status of participants was monitored for 365 days after imaging; if patients were diagnosed with contralateral breast cancer during this time period they were considered as being diagnosed with breast cancer for the purposes of the study13.
Of the 969 patients, 33 were diagnosed with contralateral breast tumours within 365 days of entering the study13. Of these 33, 30 were diagnosed as a result of a positive breast MRI screening13. 18 of these cancers were invasive carcinomas and 12 were ductal carcinoma in situ13. Of the three patients not diagnosed using MRI, one was diagnosed via a mastectomy sample (previously assessed as being BI-RADS 3) and two were diagnosed via prophylactic mastectomy (previously assessed as being BI-RADS 1)13. The tumours found in these patients were 1, 3 and 4mm in diameter13.
As a result of these findings it was determined that, in this study, MRI provided an extra diagnostic yield of 3.1%13. It had an estimated sensitivity (i.e. the chance that it will correctly identify people with breast cancer) of 91% and an estimated specificity (i.e. the chance that it will correctly identify people without breast cancer) of 88%13. It had a negative predictive value (i.e. the proportion of patients correctly diagnosed) of 99%, with the risk of participants developing breast cancer one year after receiving a negative MRI screening result estimated at 0.3%13.
It is also important to note that none of the cancers discovered as a result of the MRI investigations had metastatized to local lymph nodes, which may not have been the case had the cancers been left to progress untreated13. The discovery of the cancers in this study also allowed the participants to receive treatment for this cancer and their existing cancer at the same time, meaning patients did not need multiple rounds of cancer therapy13.
This study clearly shows that MRI can improve detection of contralateral breast cancers in conjunction with mammography and CBE. The increased detection came with a false positive rate of 10.9% and a risk of detecting benign disease on biopsy of 9.4%13, but nonetheless, the results demonstrate that MRI is an effective tool in detecting breast cancers and improving patient outcomes (although currently, the cost of MRI is too high for this type of screening to be widely used in unilateral breast cancer patients13).
3.2 Effects of PET on Cancer Patient Outcomes
