Nuclear medicine uses the properties of radioactive and stable nuclides to diagnose morphological and/or physiological disorders and to provide therapy using unsealed radioactive sources.

One great advantage that nuclear imaging techniques have over other imaging modalities is that they represent function rather than morphology. Another is that, although ionizing radiation is used, adverse reactions are rare. A third advantage is that many of the techniques demonstrate exquisite sensitivity in detecting disease, but this is accompanied by the great disadvantage that radionuclide studies are seldom specific; that is, although certain notable exceptions occur, pathognomonic appearances are rare.

It is well known that radioactive substances can emit &agr;- or &bgr;-particles, or else &ggr;-rays. &agr;-Particles are emitted only by the naturally occurring radionuclides, such as radium, which are no longer used in medicine. &bgr;-Particles, because of their short path length before they are absorbed, are useful for therapy but useless for diagnosis as they cannot emerge from the body. This leaves only &ggr;-rays which, like X-rays, can pass out of the body and be detected externally.

The currently used unit of radioactivity is the becquerel (Bq), which is the Système International unit and equal to one disintegration per second. As this unit is a very small amount, the commonly used multiple is the megabequerel (MBq), where 1 MBq = 10&sup6; disintegrations per second. The older unit, the curie (Ci) = 3.7 × 10¹&sup0; disintegrations per second, so 1 millicurie (mCi) = 37 MBq.

The basic radiopharmaceutical is a &ggr;-emitting radionuclide attached to a compound which is specially tailored to target a particular organ system or physiological process. The quantities used are generally small and so the radiopharmaceuticals neither disturb the process under investigation nor provoke hypersensitivity reactions.

The most popular radionuclide used currently is called technetium-99m (where 99 is the mass number and m means ‘metastable’; it can also be written &sup9;&sup9;Tc&supm;). Like all metastable radionuclides, technetium-99m has the triple advantages of being a pure &ggr;-emitter, having a short half-life, and being readily available from a longer-lived generator containing its parent radionuclide. In fact it is now used so widely that &ggr;-ray detectors are specifically designed to function optimally at the &ggr;-ray energy of technetium-99m.

Technetium-99m originates from the generator as the pertechnetate, &sup9;&sup9;Tc&supm;O&sub4;, but it can, for example, be attached to human serum albumin macroaggregates which blockade the pulmonary capillaries and can be used to detect pulmonary emboli. It can be linked to phosphates or phosphonates which undergo chemiabsorption on to bone crystal in order to image bone. Reticuloendothelial uptake in liver, spleen, and bone marrow is effected by linking it to sulphur or tin colloid, which is phagocytosed. In the form of the pertechnetate it can be taken up by active ion transport into gastric mucosa, thyroid, and salivary glands. Hepatocyte and biliary-tract imaging uses the active cellular transport and excretion of &sup9;&sup9;Tc&supm;-labelled substituted carbamoyliminodiacetic acid. It can also be chelated with, say, dimercaptosuccinic acid, which will be taken up by the renal cortical cells to produce renal images. Chelated to diphenyltriaminepenta-acetic acid it can measure the glomerular filtration rate. The most exciting recent compound linked to technetium-99m is hydroxymethylpropylamine oxime (‘Ceretec’, ‘Exametazime’: Amersham), which crosses the blood–brain barrier and can be used, say, to investigate neuropsychiatric disorders such as Alzheimer's dementia and schizophrenia.

Other radionuclides are, of course, also used in nuclear medicine. Examples are thallium-201 (²&sup0;¹Tl) chloride for myocardial imaging, and gallium-67 (&sup6;&sup7;Ga) citrate for the detection of tumours and infective/inflammatory foci. Indium-111 (¹¹¹In) has been used to label white cells (to detect infection) or monoclonal antibodies (to detect targets as diverse as tumours or areas of endometriosis). Iodine-123 (¹²³I) has replaced one of the original radionuclides used in nuclear medicine, iodine-131 (¹³¹I), in the diagnosis of thyroid disorders.

All of the above radionuclides are called single-photon emitters as they emit single &ggr;-ray photons. However, there is also a class of so-called positron-emitting radionuclides, that emit two photons, a positive electron (or positron), and a conventional negative electron in coincidence at an angle of 180° to each other. These charged subatomic species cannot exist as such for any length of time and are annihilated by their opposite number, yielding two high-energy &ggr;-rays. Among the positron-emitting radionuclides are important biological atoms, such as oxygen-15 (¹&sup5;O), nitrogen-13 (¹³N), carbon-11 (¹¹C), and fluorine-18 (¹&sup8;F), which can be incorporated into a variety of tracers. Fluorine-18, for example, has been incorporated into fluorodeoxyglucose and has revolutionized neurophysiology and neuropathology.

Unfortunately, these potentially most useful radiopharmaceuticals are produced in cyclotrons and have extremely short half-lives (of the order of a few minutes). Therefore the necessary in-house facilities needed to produce these include the cyclotron, on-line radiochemical processing, purification, and sterilization, and a positron camera—an exorbitantly expensive facility.

It is the ability of radiopharmaceuticals to act as indicators of physiological processes which separates nuclear medicine from most other imaging processes, such as computed tomography and ultrasound, which are largely conveyors of morphological information. However, exceptions exist. The intravenous urogram, for example, is a functional as well as a morphological study.

The distribution within the body of single-photon-emitting radionuclides (&ggr;-emitters) is mapped by means of a scintillation, or gamma, camera (Fig. 1) 179. A modern camera can either accumulate flat (‘planar’) images or rotate around the body to reconstruct, via the computer, tomographic images in the transverse, sagittal, or coronal planes.

The basic component of such cameras is a thallium-activated sodium iodide crystal, which can convert &ggr;-rays into light photons. The latter are then converted into electrons and subsequently aggregated into amplified filtered pulses. The pulses, suitably distributed spatially, can then produce maps of the distribution of radioactivity on either X-ray or polaroid film, or on to light-sensitive paper. Otherwise, pulses may be stored in the memory banks of a computer for later manipulation.

In order to screen out unwanted &ggr;-rays, both from the patient and from other sources of radiation in the imaging suite or from background radiation, collimators are used. These usually consist of a lead slab penetrated by many small holes, and acting like a sieve, keeping out &ggr;-rays from any direction except those from the organ of interest. Further sorting of unwanted energies is done electronically.

The data can therefore be in either digital or analogue form, but the two are interconvertable. So-called ‘regions of interest’ can, for example, be drawn around areas of physiological interest (such as the renal cortex, the left ventricle, the oesophagus) and count-rate changes within these regions can be derived as activity–time curves. In this way, for example, renograms can be produced to demonstrate the passage of a radiopharmaceutical such as &sup9;&sup9;Tc&supm;-diphenyltriaminepenta-acetic acid through the kidney.

Positron detectors detect the pair of &ggr;-rays resulting from the positron–electron annihilation; in order to do this they are placed on either side of the source, usually in a ring formation, and only register a count if both detectors detect photons in coincidence. Image-control techniques are then used to reconstruct the cross-sectional image. Single-photon-emission tomography uses the rotating gamma-camera to detect photons from multiple angles (180° or 360°) around the patient and an associated computer to reconstruct multiple slices. Technetium–99m and iodine–123 are the common radionuclides used for single-photon-emission tomography, and the principal organs of interest are the brain and the myocardium.

Bone blood flow and bone turnover are the major factors determining the uptake of radiopharmaceuticals into bone. In the resting state, about one-third of the maximum potential blood flow through bone is excluded by sympathetic tone, and so processes that alter this tone also affect bone blood flow. Chemiabsorption on to recently deposited or exposed bone crystal is the principal method of radiopharmaceutical uptake.

Normal bone will therefore be active (‘warm’ rather than ‘hot’), with the larger bones, such as the pelvis, and the joints predominating. Diseased areas show up as photon-abundant or ‘hot’ areas due to increased blood flow, bone turnover, or loss of sympathetic tone. However, a purely lytic area, such as an osteoclastoma or an infarct, can be photon-deficient (‘cold’), and extraosseous uptake of bone-seeking radionuclides is well described, for example in paretic states, neuropathies, areas of necrosis, and infarction.

&sup9;&sup9;Tc&supm; complexed organic and inorganic phosphate compounds are the major bone-seeking tracers, to which hydroxyl groups are often added to improve crystal binding. The currently favoured compounds are the diphosphonates. The skeleton takes up about 60 per cent of the injected dose, the rest being excreted through the kidneys and, at the same time, providing renal scintigrams, which are crude assessments of renal function. Skeletal scintigraphy is usually done in three phases, detecting flow, equilibrium uptake, and the so-called late crystal phases.

As with most radionuclide studies, skeletal scintigraphy is exquisitely sensitive (10¹&sup4; times as sensitive as plain skeletal radiography), but non-specific. Its main use in modern nuclear medicine is the detection of metastases and, nowadays, requests for skeletal scintigraphy constitute nearly half of the routine workload of most general nuclear medicine departments (Fig. 2) 180.

Some 20 to 30 per cent of bone metastases are seen on skeletal scintigraphy, but not on radiographs. However, in about 5 per cent of cases or less, the reverse is true and it is possible that processes infiltrating the bone marrow, such as myeloma and leukaemia, will be missed on skeletal scintigraphy.

A variant of the appearance of metastatic disease in the skeleton is the so-called ‘superscan’, when the whole skeleton shows relatively uniform high uptake with little in the way of discrete foci. It might even be reported as normal until it is noted that no renal images are present—an indication of the fact that nearly all the dose has been taken up by rapidly metabolizing bone and very little is left to excrete through the kidneys (Fig. 3) 181. However, this pattern may also be seen in metabolic bone disease.

Benign bone tumours show varying degrees of uptake. That in an osteoid osteoma may be almost pathognomonic with a central, very hot nidus, surrounded by an oval of slightly lesser intensity (Fig. 4) 182. Most other bone neoplasia show varying degrees of uptake and are usually investigated by plain radiography, computed tomography (CT), or magnetic resonance imaging (MRI). However, multifocality can be demonstrated easily on scintigraphy.

As far as trauma is concerned, 80 per cent of all fractures show increased uptake by 24 h, and nearly all are visible by 3 days. Injury often removes the sympathetic drive, resulting in generally increased, but diffuse, uptake—a situation called the reflex sympathetic dystrophy syndrome. Failure to see increased flow at a fracture site a few weeks after injury suggests a lack of bone union. On skeletal radiography, rib fractures usually present as focal areas of increased uptake arranged in a vertical line, as opposed to metastases, which are randomly located. In many parts of the skeleton, such as wrist, ankle, face, and base of skull, fractures not seen on conventional radiology may be detected easily on scintigraphy. This includes stress fractures of the tibia.

Avascular necrosis of the femoral head or the scaphoid may be detected as areas of photon deficiency during both the flow and crystal phases. In the case of hip fractures, account must be taken of the fact that a photon-deficient head can be due to tamponade of the artery by surrounding oedema. To distinguish between necrosis and oedema, a bone-marrow scintigram using &sup9;&sup9;Tc&supm;-sulphur or &sup9;&sup9;Tc&supm;-tin colloid, which is phagocytosed by the marrow reticuloendothelial cells, will demonstrate whether the marrow has survived or not. The marrow will survive in generalized oedema, but not in avascular necrosis.

Skeletal scintigraphy can detect osteomyelitis both on the flow study, which demonstrates increased blood flow to the affected area, and on the delayed study due to bone turnover. Radiographic changes may not occur before 1 to 2 weeks. The technique is also useful in distinguishing between osteomyelitis and cellulitis, as the latter produces only increased flow, while osteomyelitis shows both increased flow and increased bone turnover. Bone injured in any way can produce misleading images as reactive bone shows a high affinity for radionuclides for up to 9 months after a fracture.

The problem is at its most frustrating in the case of a prosthetic hip replacement, which can become loose or infected, or both. The problems may be solved by using an infection-seeking agent such as gallium-67 citrate or ¹¹¹In-labelled white cells, but the problem may be extremely difficult to resolve.

Radionuclides, such as technetium-99m methylene diphosphonate (HMDP) can also be used to investigate metabolic bone disease, such as Paget's disease, renal osteodystrophy, primary hyperparathyroidism, osteoporosis, and a number of other rarer conditions. These can be detected, quantified, and followed-up easily, using serial scintigraphy or bone absorptiometry.

The skeletal scintigraphic agents, notably technetium-99m pyrophosphate, are often taken up in soft tissue. The best known is uptake in areas of myocardial or cerebral infarction or ischaemia, while this uptake is also quite common in necrotic tumours and a host of other situations, including scar tissue in the lungs of patients in chronic renal failure.

Skeletal scintigraphy can also be used for the detection and follow-up of joint disease, such as rheumatoid arthritis, where there is a strong correlation with clinical symptoms.

Reasonable renal images may be obtained using &sup9;&sup9;Tc&supm;-dimercaptosuccinic acid, and this should be used in patients allergic to iodine-containing contrast media and so unable to undergo intravenous urography. All renal masses show up as photon-deficient areas. A renal column of Bertin, which may appear as a mass on urography, shows photon-abundance on scintigraphy. However, the principal use of the technique is for the detection of renal scarring, especially in children, which is notoriously difficult or impossible to detect with intravenous contrast. By measuring the uptake of labelled dimercaptosuccinate by each kidney, an assessment of relative renal function can be made, and renal scintigraphy has also been found to be sensitive in the detection of renal trauma.

A second useful tracer technique for investigating the kidneys is renography. This is usually performed with &sup9;&sup9;Tc&supm;-diphenyltriaminepenta-acetic acid or &sup9;&sup9;Tc&supm;-mercaptoacetyltriglycine. The normal renogram is a computer-generated activity–time curve depicting the passage of radionuclide through the kidney, and is obtained by defining a ‘region of interest’ over the kidneys. The normal renogram is shown in Fig. 5 183. Its analysis is complex, utilizing so-called deconvolutional analysis and deriving transit times, but it seems to be just as useful it if is interpreted simply. The rising part of the curve (Phase I, or ‘vascular phase’) is a function of radionuclide reaching the kidney; the peak (Phase II, or ‘handling phase’) is related to the efficiency of parenchymal function, while the descending part of the curve (Phase III, or ‘excretory phase’) shows the rate of washout of tracer from the renal area. Much of the activity ‘seen’ by the gamma-camera will be background activity, and ideally this should be subtracted from the renal curves.

In parenchymal failure there is a normal vascular phase, but the handling peak is of poor count rate and broadened. One of the most difficult diagnoses to make without invasive contrast angiography is that of renovascular hypertension. However, renography following the administration of angiotensin-converting enzyme inhibitors is now reported to increase the accuracy of the test. An example of its use is given in Fig. 6 184.

In obstructive situations one might expect a renogram similar to that shown in Fig. 7(a) 185, where the vascular and handling phases are normal, but the excretory phase fails to descend normally and may even continue to rise. However, a similar curve is obtained when there is hydronephrosis without current obstruction. In this instance the administered tracer simply pours into the dilated pelvicaliceal system. The curve will eventually descend, but not within the 30 min or so normally assigned to a renogram study. There is, however, a very simple way of distinguishing between this ‘floppy-bag’ scenario and true obstruction, and that is to perform a diuretic-provocation renogram. There will be prompt clearance of tracer from the pelvis in the cases of non-obstructive hydronephrosis, as shown in Fig. 7(b) 186.

Radionuclide methods for detecting vesicoureteric reflux can be direct or indirect and are excellent for follow-up studies as the radiation dose to the patient is much lower than for voiding cystourethrography. The direct test is similar to its radiographic counterpart in that technetium-99m is instilled into the bladder via a catheter and images are recorded as the patient voids (Fig. 8(a)) 187. The indirect study can be done as part of a conventional renogram, with the patient being asked to void near the end of the excretory phase. If reflux is present, the curve will show a sharp rise, as shown in Fig. 8(b) 188.

Serial renograms can be very useful in monitoring renal grafts after transplantation. Apart from being able sometimes to detect early rejection 1 to 2 days before biochemical abnormalities become obvious, the method is useful for detecting acute tubular necrosis (where the renogram steadily rises), renal vessels, ureteral obstruction, thrombosis, and extravasation. &sup9;&sup9;Tc&supm;-sulphur colloid is also useful in identifying graft rejection, as macrophages accumulating in the failing transplant take up this agent to the same extent as the liver and spleen.

Kidneys with active pyelonephritis take up gallium-67 citrate for much longer (about 72 h) than normal kidneys (24 h).

Any of the above studies may be coupled with measurements of total glomerular filtration rate and effective renal plasma flow using &sup5;¹Cr-ethylenediaminetetraacetic acid clearance and iodine-123 orthohippurate or &sup9;&sup9;Tc&supm;-mercaptoacetyltriglycine, respectively.

Radionuclide methods can also be used to investigate acute testicular pain, which can be due to torsion or acute epididymitis. In the former, a cut-off in the flow of the testicular artery is accompanied by decreased vascularity of the testes and increased activity of the surrounding scrotal structures. Epididymitis shows only a low-intensity, semicircular focus or even a small focal area.

In cases of male impotence involving inerectibility, it is possible to measure penile blood flow and venous leakage quantitatively using &sup9;&sup9;Tc&supm;-labelled red cells.

By asking the patient to insert a small syringe containing a low dose of technetium–99m pertechnetate into her vagina like a tampon and imaging her pelvis an hour or so after she has pressed the plunger and removed it, tubal patency can be tested. The tracer is seen in the area of the ovaries if the fallopian tubes are patent.

Pulmonary perfusion imaging for the detection of pulmonary embolism is based on the principle of capillary blockage, in that labelled particles of human serum albumin, about 20 to 50 &mgr;m in diameter, are injected intravenously. These are mixed in the blood after passage through the right side of the heart and are then trapped in the pulmonary capillary bed with a distribution reflecting pulmonary arterial blood flow. The most popular radiopharmaceutical for this purpose is &sup9;&sup9;Tc&supm;-macroaggregated human serum albumin. The albumin particles are broken down by alveolar macrophages after about 3 to 12 h.

Where there is decreased pulmonary blood flow, for example distal to the site of an embolus, there will be a region of photon-deficiency (Fig. 9) 189. However, many other pathological processes also decrease arterial perfusion, but as they invariably also affect regional ventilation, perfusion studies cannot be interpreted correctly without a concomitant ventilation scintigram.

Ventilation scintigraphy can be documented using inert gases such as xenon-133 (¹³³Xe) or krypton-81m (&sup8;¹Kr&supm;), or as an aerosol such as &sup9;&sup9;Tc&supm;-diphenyltriaminepenta-acetic acid delivered through a nebulizer. A more recent agent is technetium-99m pseudogas, which is an ultrafine, mainly monodispersed aerosol. It is generated when a spray of technetium-99m pertechnetate solution is ethanol is burnt. Krypton-81 gas is the safest and most popular agent, and it simply inhaled in oxygen by the patient standing in front of, or lying under, the gamma-camera. It is derived from a rubidium-81 (&sup8;¹Rb) generator.

The sensitivity of pulmonary perfusion imaging approaches 100 per cent in detecting emboli in a given patient but, when compared with pulmonary angiography, both tests miss some emboli and the specificity is low. The following is a simplistic guide to interpretation using perfusion and ventilation scintigraphy and the contemporaneous chest radiograph.

1.Normal: if pulmonary perfusion is normal, then it can be assumed that both ventilation scintigraphy and the chest radiograph will be normal.

2.Unmatched perfusion defect(s): here abnormal perfusion with a photon-deficient defect occurs with normal ventilation and a chest radiograph which is either normal or shows oligaemia and vessel cut-off in the affected area. This is diagnostic of early reversible pulmonary embolus.

3.Matched perfusion defect(s), where perfusion defect(s) correspond exactly with the ventilation defect(s). In this instance the patient will either have known obstructive airways disease (e.g. emphysema or asthma), the hypoxia causing vessel close-down and so a perfusion defect, or else he will have opacities on the chest radiograph corresponding to the matched defect(s). The diagnosis is then that of the opacities, such as neoplasia, collapse, infection, oedema, effusion, fibrosis, haemorrhage and, of course, established infarction. Therefore a matched defect may mean that a patient has had an embolus, albeit a little late for early anticoagulation.

The above schema is, in fact, very difficult to operate in practice, due to on-going controversy as to what constitutes a significant perfusion defect. The result is that opinions are usually given as probabilities. In the case of normal perfusion, there need be no further evaluation for pulmonary embolism. Where there is clinical suspicion of emboli, but with a low probability of embolism on scintigraphy, no further evaluation is needed and also no treatment unless there is radiographic evidence of deep venous thromboses. If the probability of emboli is high, anticoagulation is imperative, but angiography is unnecessary. In the intermediate probability range,management depends on whether the clinical suspicion of pulmonary embolism is high or low. If it is high, either venography or angiography is necessary, and anticoagulation is necessary if either are positive. However, if the clinical suspicion is low, further angiography or venography is unnecessary and the appearances are usually due to a different disease.

Liver and biliary system

The liver–spleen scintigram, using &sup9;&sup9;Tc&supm;-sulphur or &sup9;&sup9;Tc&supm;-tin colloid, was once one of the most frequently used procedures in nuclear medicine. Its decline has not been due to a lack of accuracy, but to the ability of ultrasound and cross-sectional imaging techniques, such as CT and MRI, to characterize lesions and their surrounds much more effectively.

However, one existing technique worth describing is that of hepatic arterial perfusion scintigraphy, which is used for the mapping of hepatic metastases and involves the delivery of &sup9;&sup9;Tc&supm;-macroaggregated albumin (as described above for use in pulmonary perfusion scintigraphy) through a catheter inserted into the common hepatic artery, just distal to the origin of the gastroduodenal artery. The technique is derived from that used to deliver chemotherapeutic agents. In this way the hepatic arterial perfusion scintigraphy technique ‘locks in’ the pattern of chemotherapy delivery for later imaging. This ensures chemotherapy delivery to the entire liver tumour burden and identifies inadvertent delivery to the gut or to the systemic circulation via arteriovenous shunting.

The use of &sup9;&sup9;Tc&supm;-labelled red blood cells has become valuable for the characterization of known liver masses. The method relies on the fact that most cavernous haemangiomas have decreased perfusion, but increased blood-pool activity. This state of affairs is mimicked only by the rare angiosarcoma. The combined use of this radiopharmaceutical and single-photon-emission tomography can detect haemangiomas as small as 1 cm diameter. Advances in single-photon-emission tomography technology, such as high-resolution and dynamic single-photon-emission tomography, have improved this figure to less than 1 cm.

The standard liver-spleen scintigram, which relies on the uptake of &sup9;&sup9;Tc&supm;-labelled colloid by the reticuloendothelial cells of the spleen and their counterpart, the Kupffer cells in the liver, still serves as a quick and easy procedure for the establishment of hepatosplenomegaly and is highly sensitive in assessing diffuse liver disease. The patterns in alcoholic liver disease (small liver, left-lobe hypertrophy, shift of colloid uptake to an enlarged spleen) and in isolated hepatic venous obstruction (Budd–Chiari syndrome with its ‘hot’ caudate lobe) are particularly well recognized.

In general, the combined use of &sup9;&sup9;Tc&supm;-sulphur colloid imaging, gallium-67 citrate imaging, hepatobiliary imaging (see below) and radiolabelled red cell imaging allow the histological characterization of mass lesions, not yet realized by CT and MRI. Such a combination of methods is very useful in the characterization of focal nodular hyperplasia, hepatic adenoma, focal fatty replacement, cavernous haemangioma, hepatocellular carcinoma, and macroregenerating nodules.

The introduction of &sup9;&sup9;Tc&supm;-labelled substituted carbamoyliminodiacetic acid compounds has permitted a non-invasive method of investigating hepatobiliary disorders. These so-called bifunctional agents are extracted from the blood by the hepatocyts and are concentrated in bile. In this way one obtains an estimate of hepatocyte function and an image of the biliary excretion pathway. A great advantage of the method is that it can be used when bilirubin levels are as high as 30 mg per cent.

The technique should be the first choice in the diagnosis of acute cholecystitis, where the cystic duct is obstructed. A normal study demonstrates, in temporal sequence over about an hour, the liver, common bile duct, gallbladder, duodenum, and jejenum. If the gallbladder is not seen within 30 min of the intravenous injection of the &sup9;&sup9;Tc&supm;-labelled compound, acute cholecystitis is present (always provided the patient has a gallbladder!) (Fig. 10) 190. Concomitant ultrasound is useful when thickening of the gallbladder wall, a pericholecystic collection, or a stone in the cystic duct can be found. A similar technique can be used in jaundiced neonates to distinguish between biliary atresia and neonatal hepatitis. In general, however, jaundice is always investigated by ultrasound in the first instance.

The role of the technique in chronic cholecystitis is limited, but in the postoperative patient it is useful in assessing biliary leaks, functional patency of the biliary system and for detecting cystic-duct remnants. It is also the only reliable way of demonstrating enterogastric reflux quantitatively under physiological conditions.

As stated above, &sup9;&sup9;Tc&supm;-labelled sulphur, tin, or antimony trisulphide colloid, which is trapped by the reticuloendothelial cells, can be used to image the liver, spleen, bone marrow, and lymphatic system.

The spleen may be imaged without concomitant hepatic activity by injecting autologous red cells, denatured by keeping them in a water-bath at 40°C for about 1 h and labelled with technetium-99m. The technique is usually reserved for searching for splenuniculi.

The distribution of bone marrow (or at least its reticuloendothelial elements, which usually coincide with the haematopoietic elements) varies with age as the functioning marrow withdraws from the extremities into regions said to be ‘covered by an Edwardian bathing costume’. Any deviation from this pattern, say in myelofibrosis (marked decrease in marrow elements) or in the chronic haemolytic anaemias (extension of the marrow space), can be detected easily. The method is also useful for monitoring bone-marrow transplants and for deciding on optimal sites for marrow aspiration.

Injections of &sup9;&sup9;Tc&supm;-antimony trisulphide between the webs of the toes will enable the extent and integrity of the lower limb lymphatics to be evaluated. This technique may therefore be used to assess patients with Milroy's disease. Similar techniques can be used to evaluate lymphoedema in the arms and to plot the position of the internal mammary chain prior to chest wall irradiation for appropriate medially situated breast cancers.

The liver and hepatobiliary systems have already been discussed. The salivary glands can be imaged following intravenous injection of technetium-99m as the pertechnetate, when the normal parotid and submaxillary glands become visible. Sublingual glands are too small to be seen. The activity can be discharged with a sialogogue. There is no uptake from, or discharge into, the four glands in true xerostomia or Sjögren's syndrome, whereas these are normal in cases of psychological origin. The method may also be helpful in the diagnosis of Warthin's tumour (papillary cystadenoma lymphomatosa) where there is no uptake or excretion of tracer.

Structural disease of the oesophagus is correctly investigated using endoscopy and barium meal, but the investigation of oesophageal function may be performed using radionuclides. The principal indications are either gastro-oesophageal pain, dysphagia, or one of the neuromuscular disorders. The two principal radionuclide tests for the investigation of these conditions are the measurement of oesophageal transit time and the physiological oesophageal reflux test. Reflux in babies can be determined by adding &sup9;&sup9;Tc&supm;-sulphur colloid or &sup9;&sup9;Tc&supm;-diphenyltriaminepenta-acetic acid and by taking late images over the chest to look for aspiration of tracer.

Abnormalities of gastric emptying are common after surgery for peptic ulceration, and include dumping syndrome, diarrhoea and gastric stasis. Early dumping is due to rapid gastric emptying associated with a fall in plasma volume. Late dumping is due to reactive hypoglycaemia resulting from rapid emptying. Gastric emptying-studies using liquid and solid meals labelled with &sup9;&sup9;Tc&supm;-, ¹¹¹In-, or ¹¹³I&supm;-diphenyltriaminepenta-acetic acid can assess such problems. They are not indicated in all postoperative patients, but dumping or stasis must be confirmed before any further surgery is contemplated.

In acute gastrointestinal haemorrhage, the site of bleeding is usually localized by angiogram studies. However, radionuclide angiography is more sensitive for the detection of such bleeding, and can be a valuable guide to selective abdominal arteriography. The two tracers commonly used for this purpose are &sup9;&sup9;Tc&supm;-labelled sulphur colloid or red cells. Labelled red cells are better than colloid for detecting intermittent bleeding or bleeding near the liver and spleen. They can also be used for serial imaging over 24 h as the labelled cells continue to circulate. One disadvantage of the technique is that target:non-target ratios are low due to the high circulating background of labelled cells (Fig. 11) 191. Another is that blood leaking into the bowel may be displaced both distally and proximally and this may result in the false identification of the bleeding site. Finally, free pertechnetate can concentrate in gastric or colonic secretions. Radionuclide angiography using this method is reported to detect bleeding rates of 0.1 ml/min, which compares favourably with a rate of only 0.5 ml/min for contrast angiography.

The rationale for the use of radiocolloid for the detection of bleeding is that intravascular colloid is cleared rapidly by the hepatic and splenic phagocytes, leaving visible any colloid extravasating from a bleeding site. This method is useful in detecting active lower gastrointestinal tract bleeding and in monitoring the efficacy of therapy. The use of colloid yields a higher target:non-target ratio than that of labelled red cells as the background is rapidly cleared of activity. A disadvantage is that bleeding sites near the liver or spleen are difficult to detect.

Rectal bleeding in children may be due to bleeding from ectopic gastric mucosa in a Meckel's diverticulum. Gastric mucosa will take up pertechnetate with an accuracy of between 85 and 95 per cent in children.

It is well known that the thyroid traps inorganic plasma iodide, which is later organified to form the thyroid hormones tri-iodothyronine (T&sub3;) and thyroxine (T&sub4;). Iodine uptake studies using iodine-131 as sodium iodide were among the very first procedures in nuclear medicine to be carried out in the clinical situation, shortly after this fission product of uranium-235 became available after the Second World War. Today, the radionuclide of choice is the much safer iodine-123, with iodine-131 used mainly for therapy of toxic nodules, as a result of its high output of &bgr;-particles.

A number of functional thyroid studies can be performed using iodine-123; namely, uptake tests to distinguish between hypo-, eu-, and hyperthyroidism; thyroid stimulating hormone (TSH) or thyrotropin releasing hormone (TRH) stimulation and repression tests to distinguish between disorders of the thyroid–pituitary–hypothalamic axis and the gland itself; and the perchlorate washout test to evaluate whether organification of iodide is normal.

Functional morphology can be assessed using iodine-123 or technetium-99m as pertechnetate (which is trapped, but obviously not organified by the thyroid). Ectopic thyroid tissue, which may occur in the neck anywhere from the back of the tongue to behind the sternum, may be mapped. Thyroglossal cysts are usually detected clinically, but scintigraphy is employed to ensure that such clinically detected masses do not contain thyroid tissue. Even the normal thyroid image may show anatomical deviations, with one lobe larger than the other or, rarely, just a solitary dominant lobe. Pyramidal lobes are also seen from time to time. One should also never underestimate the amount of regrowth of thyroid tissue that can take place after partial, or even so-called complete, thyroidectomy.

Abnormal thyroid images can demonstrate focal or diffuse tracer uptake, and such uptake can, in turn, be photon-abundant (‘hot’) or photon-deficient (‘cold’). The division of thyroid nodules into hot or cold is important, as 15 to 20 per cent of cold nodules may be malignant, whereas hot nodules are nearly always benign and usually toxic. Unfortunately, there is no way of distinguishing further between the various causes of cold nodules (benign adenomas, carcinomas, metastases, cysts, haematomas, focal thyroiditis, abscesses, or combinations of these) using radionuclide methods, but ultrasound can, of course, distinguish between cystic and solid lesions, and the advent of fine-needle aspiration has, in any case, enabled a preoperative diagnosis of thyroid nodules to be made. The rarer medullary carcinoma of the thyroid, which produces hypercalcitonaemia, may be detected with technetium-99m in its pentavalent state coupled to dimercaptosuccinic acid.

Toxic goitres include the hyperplastic, hyperfunctioning gland of Graves' disease. As opposed to this diffuse variety of goitre, one gets the uninodular or multinodular toxic goitre, in which the overactive areas suppress normal tissue—the so-called ‘hot nodule’. While acute suppurative thyroiditis is rare, subacute thyroiditis is a painful, but self-limiting problem, which is probably of viral origin. It produces generalized decreased uptake. Hashimoto's thyroiditis is, by contrast, chronic and painless, and is probably due to lymphocytic infiltration causing an organification defect. Early images in this condition show an enlarged gland with normal or enhanced trapping, but later images show poor, uneven uptake.

In cases of hypothyroidism one may not even have sufficient counts to produce an image, but such a ‘non-image’ can provide diagnostic information which is just as useful as in any other, more dramatic, imaging situation. Euthyroid goitres simply show enlargement with normal uptake.

It is estimated that a surgeon seeking the source of excess parathormone will find the source on exploration most of the time, but is less than 75 per cent successful on re-exploration, and in this instance nearly all available imaging techniques have to be employed.

The thallium-201/technetium-99m pertechnetate subtraction technique is arguably the most useful imaging test. Thallium–201 is a potassium analogue and, besides its well-known property of being taken up in the mycardium, is also localized in parathyroid adenomas and a variety of other neoplasias. Normal parathyroid glands are not visible. The problem is that thallium-201 chloride is also taken up by a normal thyroid and so one has to subtract electronically a thyroid image taken after an injection of pertechnetate from one taken previously after an injection of thallium-201 in the same position. The technique is more complex than this simple description would lead one to believe, but it is claimed that about 80 per cent of parathyroid adenomas can be localized by this method. It is not useful in secondary hyperparathyroidism.

Neoplasia derived from the neural crest, so called apudomas, may be imaged using the noradrenaline and guanethidine analogue. ¹³¹I- or ¹²³I-labelled m-iodobenzylguanidine, which is taken up at sites of catecholamine storage granules. This has proved to be useful in localizing benign and malignant phaeochromocytomas, paragangliomas, neuroblastomas, carcinoid and medullary carcinomas, and has also been used therapeutically in these conditions.

When used with appropriate suppressive manoeuvres, such as the administration of dexamethasone, uptake of the cholesterol analogue, ¹³¹I- or &sup7;&sup5;Se-labelled norcholesterol, can supply evidence of adrenocortical hyperfunction. Adrenal cortical imaging is particularly appropriate to the localization of resectable neoplasia, or in the identification of a suspected adrenal remnant following surgery. Symmetrically increased uptake denotes adrenocorticotropic hormone (ACTH)-dependent hyperplasia, while asymmetrical uptake indicates ACTH-independent hypercorticalism. A unilateral image usually represents a neoplasm, commonly an adenoma. Adrenal carcinoma may destroy normal adrenal tissue with the resultant absence of an image.

The method is particularly useful in distinguishing between ACTH-induced adrenal hyperplasia and primary adrenal tumours that produce hypercorticalism. Results are only slightly less reliable in detecting Conn's tumours, producing hyperaldosteronism. The technique can also be used to distinguish between adrenal and ovarian causes of hirsutism.

Modern techniques in nuclear medicine provide valuable information about cardiac structure, pathology, perfusion, and function. Four principal radionuclide techniques are available:

(1)radionuclide ventriculoscintigraphy and the evaluation of ventricular function;

(2)radionuclide or ‘first-pass’ angioscintigraphy;

(3)myocardial perfusion scintigraphy;

(4)infarct-avid scintigraphy.

The cardiac blood pool may be imaged using &sup9;&sup9;Tc&supm;-labelled red cells. Images of the ventricles (usually the left) in systole and diastole are obtained by gating the gamma-camera to the patient's electrocardiographical signal (notably the R wave) and summing counts obtained over many cardiac cycles.

The most important functional parameter of the left (or right) ventricle that can be calculated is the ejection fraction which is identified as: Equation 21

In spite of the slight errors inherent in the method, the left-ventricular ejection fraction is a sensitive indicator of dysfunction. The value is normally greater than 50 per cent. Patients who have had recent infarction show ejection fraction decreases that correlate well with the volume of infarcted tissue, and so the parameter can serve as a guide to prognosis. It is, similarly, an excellent indicator of therapeutic effectiveness and can be assessed at rest or during exercise.

The method can also be used to generate contours of the ventricular blood pool in end-diastole and end-systole, which can be manipulated to assess wall-motion abnormalities such as hypokinesia, akinesia, and dyskinesia. This can also be accomplished by producing phase and amplitude maps from Fourier analysis of the activity–time curves of the summed counts over several cardiac cycles. Right-ventricular scintigraphy is also possible, but less accurate as there is no one view of the heart (compared with the left anterior oblique view of the left ventricle) on which the right ventricle is completely circumscribed.

Estimates of regurgitant fractions, in patients with aortic and mitral valve disease, and ventricular volumes can also be derived from radionuclide ventriculography and the method can be useful in the timing of valve-replacement surgery.

This is also known as ‘first-pass’ angioscintigraphy and documents an injection of technetium-99m as pertechnetate or labelled autologous red cells as it passes through the superior vena cava, the right atrium and ventricle, the lungs, the left atrium and ventricle and, finally, the aorta. There must be no chamber dilatation, delay in transit, or recirculation (Fig. 12) 192. The method can be used to detect any gross anatomical aberration in patients with congenital heart disease, such as right-to-left shunts. By computer analysis of activity-time curves over the lungs, left-to-right shunts can actually be quantified down to shunts as small as 10 per cent. In this way, the closing of shunts can be detected and the degree of success of corrective surgery can be observed.

First-pass scintigraphy can also be used for screening large, obstructed vascular lesions in the superior vena cava or the aorta, and it is occasionally helpful in the detection of peripheral vascular obstruction.

Myocardial cells treat thallium-201 chloride in the same way as they do potassium and so regional myocardial uptake of thallium-201 reflects intact myocardial perfusion to viable myocardium. Regions of ischaemia, acute infarction, or scarring appear as focal decreased uptake, and to distinguish among these it is essential to derive images after stress and then later at rest. An ischaemic area will ‘fill in’, whereas a scar will not.

Myocardial perfusion imaging can be used to differentiate ischaemic from idiopathic cardiomyopathy, to evaluate collateral coronary vessels or angioplasty, and to assess coronary bypass grafts.

There has been much interest recently in the replacement of thallium-201 by the more favourable &sup9;&sup9;Tc&supm;-labelled isonitriles, such as &sup9;&sup9;Tc&supm;-sestamibi.

This method, which can be used to detect both myocardial and cerebral infarcts between 12 h and 10 days after the event, depends on the property of ischaemic, infarcted, or necrotic tissue to take up certain bone-seeking tracers, such as technetium-99m pyrophosphate. A more recent advance embracing the same principle is the use of ¹¹¹In-labelled Fab fragments of antimyosin, which will bind to areas of myosin released in acute infarction.

While contrast venography provides the gold standard for detecting deep-vein thrombosis in the lower limb, it is not without its contraindications. ¹²&sup5;I-fibrinogen could detect clots in the calf only on direct counting and could not be imaged. &sup9;&sup9;Tc&supm;-fibrinogen could produce a scintigram but has been withdrawn from the market, probably because of the risk of HIV contamination. More recently, ¹¹¹In-platelets have been used for the detection of emboli, thrombi, and atherosclerotic lesions, with varying degrees of success.

Magnetic resonance imaging is now unquestionably the modality of choice in delineating cerebral lesions. Conventional cerebral scintigraphy using tracers such as technetium-99m, as pertechnetate or glucoheptonate, which only pass through the blood–brain barrier when a lesion is present, is rare nowadays but may still be useful in detecting early cerebral infections and in the differential diagnosis of cerebral infarction.

Cerebral radionuclide angiography may, however, still be a very useful rapid and non-invasive way of investigating the cerebral circulation. Any vascular lesion, such as an arteriovenous malformation, meningioma, or glioma, will be seen as a focal area of increased activity, while photodeficient areas are seen in cerebrovascular accidents, subdural haematomas, and in massive carotid occlusions. A characteristic sequence in patients who have recently had strokes is the so-called ‘flip-flop’ phenomenon, where there is diminished perfusion on the side of the brain involved during the arterial phase, but which changes to increased activity during the later venous phase (Fig. 13) 193. Jugular reflux, which occurs in the superior mediastinal syndrome, also produces a characteristic diagnostic picture.

Unfortunately, this very important branch of nuclear medicine has, as yet, little part to play in neurosurgery, but it is of undoubted importance in the investigation of neuropsychiatric disorders, such as the dementias, schizophrenias, the epilepsies, and even the depressive illnesses—none of which has hitherto been amenable to radiological diagnosis. Of all of these conditions, the epilepsies are arguably of the greatest interest to the neurosurgeon because it is possible to resect the sources of seizure disorders if they can be identified anatomically (Fig. 14) 194. Epileptogenic foci show areas of decreased perfusion and metabolism interictally, but become areas of hyperperfusion during a seizure.

The principal agents used for these studies are ¹&sup8;F-deoxyglucose and oxygen-15 in positron-emission tomography studies and &sup9;&sup9;Tc&supm;-hydroxymethylpropylamine oxime in single-photon-emission tomography studies. The last agent can also be used as an aid in the diagnosis of cerebral death.

Any isotope can be used in the evaluation of ventriculovenous shunt patency and ¹¹¹In-diphenyltriaminepenta-acetic acid cistinography can assist in the diagnosis of normal-pressure hydrocephalus.

In the investigations for epiphora, patency of the nasolacrimal ducts may be investigated under physiological conditions by instilling a drop of normal saline containing a few bequerels of technetium-99m as pertechnetate into each conjunctival sac, and documenting the progress of the tracer through the canals into the nose (Fig. 15) 195.

Many of the methods used for detecting benign and primary and secondary malignant tumours have already been discussed, but it is also necessary to describe radionuclide methods specifically designed to image tumours in general. These include mainly gallium-67 citrate scintigraphy and the use of labelled monoclonal antibodies against tumour-associated antigens.

Gallium-67 citrate was initially a bone-seeking agent and it was being used for this purpose in a patient with Hodgkin's disease when its propensity to seek out lymphomatous nodes was realized. It is, however, also taken up by a number of normal tissues, such as lymphoid tissue (notably Waldeyer's ring), lacrimal glands, liver, spleen, breasts, genitalia, and faeces, making the observation of any abnormal gallium-67 uptake difficult.

Nevertheless it is useful for detecting the extent of intrathoracic lymphoma, with a sensitivity of 75 to 80 per cent and it can detect primary bronchial carcinoma with a sensitivity of 80 to 85 per cent. However, it was soon found to be an infection/inflammation-seeking agent as well, and so it lacks specificity. Also, it does not concentrate as well in adenocarcinomas as in other types. If primary lung cancer concentrates gallium-67 citrate, detection of its spread to mediastinal nodes has a sensitivity approaching 100 per cent, compared to mediastinoscopy. However, specificity is much lower, at 53 to 60 per cent.

With the advent of polyclonal antibodies against tumour- associated antigens, followed closely by the now ubiquitous monoclonal antibodies, hopes of finding tumour-specific radiopharmaceuticals were rekindled, as were hopes of using them at high activities for therapy. Antibodies can be labelled with specific radionuclides (iodine-123), indium-111, and even technetium-99m as pertechnetate) and provide at least the potential for the specific delivery of radionuclides to a particular cancer for detection and treatment.

Polyclonal antibodies are mixtures of antibodies having affinity for the tumour as well as, unfortunately, cross-reactive affinities towards other tissues. One of the first and best of these to be tried was carcinoembryonic antigen. Monoclonal antibodies, produced by hybridoma technology are, on the other hand, directed against relatively specific antigenic determinants in the neoplastic cell, and so are less eager to react with normal cells. However, this lack of cross-reactivity is not absolute, and much effort is being expended in producing antibody fragments, such as the F(ab)&sub2; fragments, which have little or no cross-reactivity with normal tissue.

Preliminary results in cancers with well-characterized antibodies to specific tumour-associated antigens have been encouraging, for example in malignant melanoma (Fig. 16) 196, ovarian cancer, and cutaneous T-cell lymphoma. In these instances success is probably due to the fact that the antigen–antibody reaction probably occurs on the cell surface.

Therapeutic studies show some promise, but at the present time both diagnostic and therapeutic applications have not lived up to their initial theoretical promise. Exciting new developments are taking place in tumour receptor imaging. For example, radioligands which bind oestrogen and somatostatin receptors in breast carcinoma have been used to image these tumours and their secondaries and to monitor, for example, the efficacy of tamoxifen, which can be seen to make oestrogen-positive tumours (about 65 per cent of the total) disappear.

Intra-abdominal infection and inflammation are major surgical problems, to which all imaging modalities (but particularly ultrasound) can contribute. As far as radionuclide imaging is concerned, the best-known agents for seeking out inflammatory or infective foci have been gallium-67 citrate and &sup9;&sup9;Tc&supm;-labelled colloids, but these have been superseded by radiolabelled white cells (either pure polymorphonuclear leucocytes or mixed white cells) labelled in vitro with indium-11 chelates or &sup9;&sup9;Tc&supm;-hydroxymethylpropylamine oxime. More recently, microcolloids, monoclonal antibodies, and immunoglobulins have been introduced.

Gallium-67 citrate has been used for some time and has a particular reputation for seeking chronic infection. It is simple to use and requires no special preparation, but—as stated above—it targets so many normal structures that there are few areas where the target:non-target ratio is high enough to engender analytical confidence. It is also difficult to image technically as it has three so-called photopeaks which make gamma-camera tuning difficult. This makes its sensitivity for assessing intra-abdominal sepsis rather low (about 60 per cent). Its specificity is also low because it is a tumour seeker, and sequential studies over 3 days have to be carried out to enhance accuracy.

On the other hand, it is now a simple matter to label leucocytes with either indium-111 oxime, indium-111 tropolone, or technetium-99m hydroxymethylpropylamine oxime (the same substance as is used for cerebral perfusion and tumour imaging). Indium is transported by blood constituents as if it was iron, and so it is effectively trapped by iron-binding proteins in the plasma and on the surface of all blood cells. Therefore, in order to label white cells, the plasma has to be removed first (from about 60 ml of whole blood), followed by the red cells and platelets. Finally, if it has been decided to label only granulocytes, these have to be separated off from the other white cells. The appropriate white cells are incubated with indium-11 oxime or tropolone and re-suspended in platelet-poor plasma for re-injection into the patient. It would appear that pure granulocytes are not significantly superior for the detection of acute soft-tissue infections so, as their preparation requires an extra stage, it seems just as effective to label mixed white cells.

There is, as expected, normal leucocyte uptake in the haematopoietic areas of the bone marrow, liver, and spleen, but no renal or bowel activity as in the case of gallium-67 citrate. This method therefore provides a larger clear background in which to identify sepsis. Focal areas close to the liver and spleen may be imaged by using &sup9;&sup9;Tc&supm;-sulphur colloid to outline the liver and spleen and then subtracting these areas electronically from the overall image.

When the inflammatory response due to intra-abdominal sepsis is marked, early imaging at 4 h after re-injection of the labelled leucocytes can be obtained (Fig. 17) 197. However, with less acute responses, uptake is slower and may be seen only at 24 h. Delayed imaging can distinguish between abscess and exudate in Crohn's disease and ulcerative colitis. The labelled exudate moves distally, whereas the activity over an abscess obviously remains static. Leucocyte abundance in the bowel is always pathological.

The sensitivity of indium-111 tropolone-labelled mixed white cells is 100 per cent in detecting acute soft-tissue infections, but drops to 85 per cent in mixed infections. The specificity is about 92 per cent, with false-positive results due to uptake in inflammatory conditions such as acute pancreatitis and inflammatory bowel disease, necrotic tumours, dissolving haematomas, and in patients on non-steroidal anti-inflammatory agents. Swallowed saliva may also produce misleading appearances. The more recent techniques using white cells labelled with &sup9;&sup9;Tc&supm;-hydroxymethylpropylamine oxime and &sup9;&sup9;Tc&supm;-human immunoglobulin have yet to be widely evaluated.

When intra-abdominal infection is suspected and localizing signs are present, or if scintigraphy is negative or inconclusive, then ultrasound or computed tomography must be used. If not, leucocyte scintigraphy should otherwise be the investigation of choice, with a negative study virtually excluding infection. However, a positive study is seldom sufficient to allow surgical drainage without further help from ultrasound or computed tomography.

The non-imaging branch of nuclear medicine is often referred to as nuclear pathology. Although not strictly the province of the surgeon, some of the tests under this heading could be useful in clinical or research situations.

Paramount among these tests is the ability to estimate accurately almost infinitesimal amounts of certain hormones, enzymes, and drugs in the blood using competitive binding assays, especially those using so-called radioimmunoassay techniques involving radioactive tracers, particularly iodine–125.

The assessment of a number of physiological parameters involves the use of radionuclides. These include measurements of body composition (especially blood and plasma volumes), circulation and blood flow, tumour turnover, and metabolism.

The universally adopted policy in radiation protection is to keep radiation doses as low as reasonably achievable (the ALARA principle), and so it is important that the clinician considers that the potential gain for the patient in using a radioactive test will exceed potential risks. Non-essential, repetitive examinations and the use of radionuclides with large &bgr;-irradiation components must be discouraged. All ionizing radiation investigations on pregnant women are to be avoided unless the clinical benefit far outweighs the risk. In addition, it is essential to use sensitive detection equipment, good handling techniques, and highly trained personnel. If technetium–99m is used as the basis of most studies in nuclear medicine, absorbed doses will be of the same order as those incurred in conventional radiography and often far less.

Adazraki NP, Mishkin FS. eds. Fundamentals of nuclear medicine. 2nd edn. New York: The Society of Nuclear Medicine, 1988.

Maisey MN, Britton KE, Gilday DL, eds. Clinical nuclear medicine. London: Chapman and Hall Medical, 1991.

Mistry R. Manual of nuclear medicine procedures. London: Chapman and Hall, 1988. (A detailed how-to-do-it recipe book.)