Diagnostic Imaging of Salivary Gland Pathology

The Metabolic Reboot

How To Improve Your Metabolism

Get Instant Access

Pradeep K. Jacob, MD, MBA


Introduction Imaging Modalities Computed Tomography (CT) CT Technique Advanced Computed Tomography Magnetic Resonance Imaging (MRI) MRI Technique Spin-Echo T1 Spin-Echo T2

Proton Density Images (PD)

Gradient Recalled Echo Imaging (GRE)

Short Tau Inversion Recovery (STIR)

Gadolinium (Gd) Contrast

Fluid Attenuation Inversion Recovery (FLAIR)

Diffusion Weighted Images (DWI)

MR Spectroscopy (MRS)

Dynamic Contrast Enhanced Magnetic Resonance Imaging

Other Magnetic Resonance Imaging Techniques Ultrasonography (US)

US Technique Radionuclide Imaging (RNI) Positron Emission Tomography (PET) Positron Emission Tomography/Computed Tomography (PET/CT) Diagnostic Imaging Anatomy Parotid Gland

Submandibular Gland (SMG) Sublingual Gland (SLG) Minor Salivary Glands Pathology of the Salivary Glands Vascular Lesions Lymphangioma (Cystic Hygroma) Hemangioma Acute Sialadenitis Chronic Sialadenitis HIV-Lymphoepithelial Lesions Mucous Escape Phenomena

Sialadenosis (Sialosis) Sialolithiasis Sjogren's Syndrome Sarcoidosis

Congenital Anomalies of the Salivary Glands

First Branchial Cleft Cyst Neoplasms—Salivary, Epithelial Benign Pleomorphic Adenoma Warthin's Tumor Oncocytoma Malignant Tumors Mucoepidermoid Carcinoma Adenoid Cystic Carcinoma Neoplasms—Non-salivary Benign Lipoma

Neurogenic Tumors Malignant Lymphoma Metastases Summary References


Anatomic and functional diagnostic imaging plays a central role in modern medicine. Virtually all specialties of medicine to varying degrees depend on diagnostic imaging for diagnosis, therapy, and follow-up of treatment. Because of the complexity of the anatomy, treatment of diseases of the head and neck, including those of the salivary glands, are particularly dependent on quality medical imaging and interpretation. Medical diagnostic imaging is divided primarily into two major categories, anatomic and functional. The anatomic imaging modalities include computed tomography (CT), magnetic resonance imaging (MRI), and ultrasonography (US). Although occasionally obtained, plain film radiography for the head and neck, including salivary gland disease, is mostly of historical interest. In a similar manner, the use of sialography has been significantly reduced, although both plain films and sialography are of some use in imaging sialoliths. Functional diagnostic imaging techniques include planar scintigraphy, single photon emission computed tomography (SPECT), positron emission tomography (PET), and magnetic resonance spectroscopy (MRS), all of which are promising technologies. Recently, the use of a combined anatomic and functional modality in the form of PET/CT has proved invaluable in head and neck imaging. Previously widely employed procedures including gallium radionu-clide imaging are less important today than in the past.

Imaging Modalities


Computed tomography has become indispensable in the diagnosis, treatment, and follow-up of diseases of the head and neck. The latest generation of multiple-row detector CT (MDCT) provides excellent soft tissue and osseous delineation. The rapid speed with which images can be obtained along with the high spatial resolution and tissue contrast make CT the imaging modality of choice in head and neck imaging. True volumetric data sets obtained from multidetector row scanners allow for excellent coronal, sagittal, or oblique reformation of images as well as a variety of 3-D renderings. This allows the radiologist and surgeon to characterize a lesion and assess involvement of adjacent structures or local spread from the orthogonal projections or three-dimensional rendering. The ability to manipulate images is critical when assessing pathology in complex anatomy, such as evaluation of parotid gland masses to determine deep lobe involvement, facial nerve involvement, or extension into the skull base. Images in the coronal plane are important in evaluating the sub-mandibular gland in relation to the floor of the mouth. Lymphadenopathy and its relationship to the carotid sheath and its contents and other structures are also well delineated. CT is also superior to MRI in demonstrating bone detail and calcifications. CT is also the fastest method of imaging head and neck anatomy. Other advantages include widespread availability of scanners, high-resolution images, and speed of image acquisition, which also reduces motion artifacts. Exposure to ionizing radiation and the administration of IV contrast are the only significant disadvantages to CT scanning.

CT Technique

The CT scanner contains a gantry, which holds an X-ray tube and a set of detectors. The X-ray tube is positioned opposite the detectors and is physically coupled. A "fan beam" of X-rays is produced and passes through the patient to the detectors as the tube and detector rotate around the patient. In the newer generation of scanners, the multiple rows of detectors are fixed around the gantry and only the tube rotates. A table carries the patient through the gantry. The detectors send signals, dependent on the degree of X-ray attenuation, to a computer, which uses this data to construct an image using complex algorithms.

For most CT studies (especially in the head and neck), intravenous contrast is administered. IV contrast is a solution consisting of organic compounds bonded with iodine molecules. Iodine is a dense atom with an atomic weight of 127, which is good at absorbing X-rays and is biocompatible. IV contrast readily attenuates the X-ray beam at concentrations optimal for vascular and soft tissue "enhancement," but short of causing attenuation-related artifacts. Streak artifacts, however, can occur if the concentration is too high, as seen occasionally at the thoracic inlet and supraclavicu-lar region from dense opacification of the subcla-vian vein during rapid bolus injection of IV contrast.

CT of the neck should be performed with intravenous contrast whenever possible to optimize delineation of masses and inflammatory or infectious changes in the tissues and to enhance vascular structures. Imaging is obtained from the level of the orbits through the aortic arch in the axial plane with breath hold. The images are reconstructed using a computer algorithm to optimize soft tissue delineation, and displayed in soft tissue window and level settings (Figures 2.1 and 2.2). In a similar manner images are reconstructed using a computer algorithm to optimize bone details as more sharp and defined (Figure 2.3). The lung apex is often imaged in a complete neck evaluation and displayed using lung window settings

Neck With Ecs Lymph Node
Figure 2.1. Axial CT of the neck in soft tissue window without contrast demonstrating poor definition between soft tissue structures. The blood vessels are unopacified and cannot be easily distinguished from lymph nodes. Note the sialolith (arrow) in the hilum of the left submandibular gland.

Figure 2.3. Axial CT of the skull base reconstructed in a sharp algorithm and in bone window and level display demonstrating sharp bone detail. Note the sharply defined normal right stylomastoid foramen (arrow).

Stylomastoid Foramen Mri

Figure 2.2. Axial CT of the neck in soft tissue window with IV contrast demonstrates improved visualization of structures with enhancement of tissues and vasculature. Note the small lipoma (arrow) anterior to the left submandibular gland, which distorts the anterior aspect of the gland with slight posterior displacement.

Figure 2.2. Axial CT of the neck in soft tissue window with IV contrast demonstrates improved visualization of structures with enhancement of tissues and vasculature. Note the small lipoma (arrow) anterior to the left submandibular gland, which distorts the anterior aspect of the gland with slight posterior displacement.

(Figure 2.4a). Dedicated CT scans of the chest are beneficial in the postoperative evaluation of patients with salivary gland malignancies, as lung nodules can be observed, possibly indicative of metastatic disease (Figure 2.4b). Multiplanar reformatted images of the neck are obtained typically in the coronal and sagittal planes (Figures 2.5 and 2.6), although they may be obtained in virtually any plane desired or in a 3-D rendering.

The Hounsfield unit (H) (named for Godfrey Hounsfield, inventor of the CT scanner) is the unit of density measurement for CT. These units are assigned based on the degree of attenuation of the X-ray beam by tissue in a given voxel (volume element) and are assigned relative to water (0 H) (Table 2.1). The scale ranges from -1024 H for air, to +4000 H for very dense bone. The images are created based on a grayscale from black (-1024 H) to white (+4000 H) and shades of gray. Despite the wide range of units, the majority of tissues in the human body are between -100 and +100 H. Soft tissues and parenchymal organs are in a range between 20 and 80 H, whereas fat is approximately -100 H. Simple fluid is 0 H, but proteinaceous fluid can be upward of 25 H. Unclotted and clotted

Figure 2.4. Axial CT of the neck at the thoracic inlet in lung windows demonstrating lung parenchyma (a). Axial image of dedicated CT of chest demonstrating cannon ball lesions in a patient previously treated for adenoid cystic carcinoma of the palate (b). These lesions are representative of diffuse metastatic disease of the lungs, but not pathognomonic of adenoid cystic carcinoma.

Figure 2.4. Axial CT of the neck at the thoracic inlet in lung windows demonstrating lung parenchyma (a). Axial image of dedicated CT of chest demonstrating cannon ball lesions in a patient previously treated for adenoid cystic carcinoma of the palate (b). These lesions are representative of diffuse metastatic disease of the lungs, but not pathognomonic of adenoid cystic carcinoma.

blood varies depending on the hemoglobin concentration and hematocrit, but average measurements are 50 H and 80 H, respectively. CT images are displayed using a combination of "window widths" (WW, range of CT numbers from black to white), and "window levels" (WL, position of the

Figure 2.5. Coronal CT reformation of the neck in soft tissue window at the level of the submandibular glands. Orthogonal images with MDCT offer very good soft tissue detail in virtually any plane of interest in order to assess anatomic and pathologic relationships.
Figure 2.6. Sagittal CT reformation of the neck in soft tissue window at the level of the parotid gland. Note the accessory parotid gland (black arrow) sitting atop the parotid (Stenson's) duct (thin white arrow). Also note the retromandibular vein (large white arrow) and external auditory canal.
Table 2.1. CT density in Hounsfield units (H).

Tissue or Structure

Hounsfield Unit (H)

Water or CSF




Soft tissue, muscle (a)


Unclotted blood (b)


Clotted blood (b)


Parotid gland (c)

-10 to +30

Submandibular gland (c)


Sublingual gland (d)










Grey matter


White matter


(a) Depends on degree of fat deposition.

(b) Depends on the hemoglobin concentration and hematocrit.

(c) Depends on age and fat deposition.

(d) Very limited evaluation secondary to partial volume effect. CSF = cerebrospinal fluid.

(a) Depends on degree of fat deposition.

(b) Depends on the hemoglobin concentration and hematocrit.

(c) Depends on age and fat deposition.

(d) Very limited evaluation secondary to partial volume effect. CSF = cerebrospinal fluid.

window on the scale), which are based on the attenuation characteristics of tissues. Typically, head and neck images are interpreted using "soft tissue windows" (WW 500 H, WL 30 H), "bone windows" (WW 2000, WL 500), or "lung windows" (WW 1500, WL -500). "Soft tissue windows" demonstrate the slight density differences of soft tissues, whereas "bone windows" demonstrate cortical and medullary features of bones with sharp detail. "Lung windows" demonstrate the sharp interface of air and the fine soft tissue components of lung parenchyma.

Although the density of the salivary glands is variable, the parotid glands tend to be slightly lower in density relative to muscle, secondary to a higher fat content, and become progressively more fat replaced over time. The CT density of parotid glands varies from -10 to +30 H. The sub-mandibular glands are denser than parotid glands and are equivalent in density to muscle. The sub-mandibular glands vary in density from +30 to +60 H.

CT angiography (CTA) is a powerful method that allows visualization of arterial vasculature, demonstrating the vascular anatomy of arteries and veins. CTA can be critical in preoperative evaluation to determine the degree of vascularity

Parotid Gland Vasculature
Figure 2.7. CT angiogram of the neck at the level of the parotid gland demonstrating the retromandibular vein and adjacent external carotid artery (large white arrow). Note the right cervical lymphangioma (thin white arrow) associated with the tail of the right parotid gland.

of lesions and to plan an appropriate surgical approach to minimize blood loss or perform preoperative embolization. CTA is obtained with fast image acquisition over a defined region of interest while administering a rapid IV contrast bolus timed to arrive in the region of interest during image acquisition. CTA images may be rendered in 3-D data sets and rotated in any plane (Figure 2.7). CTA is not only useful for preoperative planning; it can also be quite useful in diagnosis of salivary gland vascular pathology such as aneurysms or arteriovenous fistulae (AVFs) (Wong, Ahuja, and King et al. 2004).

CT scanning, as with all imaging modalities, is prone to artifacts. Artifacts can be caused by motion, very dense or metallic implants (dental amalgam), and volume averaging. Motion artifact is common and may result from breathing, swallowing, coughing, or sneezing during the image acquisition or from an unaware or uncooperative patient. Metallic implants cause complete attenuation of X-rays in the beam and result in focal loss of data and bright and dark streaks in the image. Because the image is created from a three-dimensional section of tissue averaged to form a two-dimensional image, the partial volume or volume averaging artifact results from partial inclusion of structures in adjacent images. Finally, the beam hardening artifact is produced by attenuation of low energy X-rays, by dense objects, from the energy spectrum of the X-ray beam, resulting in a residual average high energy beam (or hard X-rays), which results in loss of data and dark lines on the image. This phenomenon is often seen in the posterior fossa of head CT scans caused by the very dense petrous bones. A multidetector row CT scanner can help reduce metallic artifacts using advanced algorithms, and can reduce motion artifacts secondary to faster scanning speeds.

Advanced Computed Tomography

Newer CT techniques including CT perfusion and dynamic contrast enhanced multi-slice CT have been studied. Dynamic multi-slice contrast enhanced CT is obtained while scanning over a region of interest and simultaneously administering IV contrast. The characteristics of tissues can then be studied as the contrast bolus arrives at the lesion and "washes in" to the tumor, reaches a peak presence within the mass, and then decreases over time, that is, "washes out." This technique has demonstrated differences in various histologic types of tumors, for example, with early enhancement in Warthin's tumor with a time to peak at 30 seconds and subsequent fast washout. The malignant tumors show a time to peak at 90 seconds. The pleomorphic adenomas demonstrate a continued rise in enhancement in all four phases (Yerli, Aydin, and Coskum et al. 2007).

CT perfusion attempts to study physiologic parameters of blood volume, blood flow, mean transit time, and capillary permeability surface product. Statistically significant differences between malignant and benign tumors have been demonstrated with the mean transit time measurement. A rapid mean transit time of less then 3.5 seconds is seen with most malignant tumors, but with benign tumors or normal tissue the mean transit time is significantly longer (Rumboldt, Al-Okkaili, and Deveikis 2005).


Magnetic resonance imaging represents imaging technology with great promise in characterizing salivary gland pathology. The higher tissue contrast of MRI, when compared to CT, enables subtle differences in soft tissues to be demonstrated. Gadolinium contrast enhanced MRI further and accen tuates the soft tissue contrast. Subtle pathologic states such as perineural spread of disease are better delineated when compared with CT. This along with excellent resolution and exquisite details make MRI a very powerful technique in head and neck imaging, particularly at the skull base. However, its susceptibility to motion artifacts and long imaging time as well as contraindication due to claustrophobia, pacemakers, aneurysm clips, and deep brain and vagal nerve stimulators limit its usefulness in the general population as a routine initial diagnostic and follow-up imaging modality. Many of the safety considerations are well defined and detailed on the popular Web site www.mrisafety.com.

MRI Technique

Although the physics and instrumentation of MRI are beyond the scope of this text, a fundamental understanding of the variety of different imaging sequences and techniques should be understood by clinicians in order to facilitate reciprocal communication of the clinical problem and understanding of imaging reports.

In contrast to CT, which is based on the use of ionizing radiation, MRI utilizes a high magnetic field and pulsed radiofrequency waves in order to create an image or obtain spectroscopic data. MRI is based on the proton (hydrogen ion) distribution throughout the body. The basic concept is that protons are normally oriented in a random state. However, once placed in the imaging magnet, a high magnetic field, a large proportion of protons align with the magnetic field. The protons remain aligned and precess (spin) in the magnetic field until an external force acts upon them and forces them out of alignment. This force is an applied radiofrequency pulse, used for a specified time and specified frequency by an antenna called a transmit coil. As the protons return to the aligned state, they give off energy in the form of their own radiofrequency pulse, determined by their local chemical state and tissue structure. The radiofrequency pulse given off is captured by an antenna, called a receive coil. The energy of the pulse and location is recorded and the process repeated multiple times and averaged, as the signal is weak. The recorded signal is used to form the image. Several different types of applied pulse sequences of radio waves result in different types of images.

The impact of MRI is in the soft tissue contrast that can be obtained, non-invasively. The relaxation times of tissues can be manipulated to bring out soft tissue detail. The routine sequences used in clinical scanning are spin-echo (SE), gradient echo (GRE), and echo-planar (EPI). Typical pulse sequences for head and neck and brain imaging include spin-echo T1, spin-echo T2, proton density (PD), FLAIR, dwi, post-contrast T1, and STIR. A variant of the spin-echo, the fast spin-echo sequence (FSE), allows for a more rapid acquisition of spin-echo images. Any one of these can be obtained in the three standard orientations of axial, coronal, and sagittal planes. Oblique planes may be obtained in special circumstances.

Spin-Echo T1

On T1 weighted images a short repetition time (tr) and short echo time (te) are applied, resulting in an image commonly used for anatomic depiction. Water signal is very low and is displayed as dark gray to black pixels on the grayscale. Fat is very bright, allowing tissue planes to be delineated. Fast flowing blood is devoid of signal and is therefore very black. Muscle tissue is an intermediate gray. Bone that has few free protons is also largely devoid of signal. Bone marrow, however, will vary depending on the relative percentage of red versus yellow marrow. Red marrow will have a signal similar to but slightly lower than muscle, whereas yellow marrow (fat replaced) will be bright. In the brain, cerebrospinal fluid (CSF) is dark, and flowing blood is black. Grey matter is dark relative to white matter (contains fatty myelin), but both are higher than CSF but less than fat. Cysts (simple) are dark in signal unless they are complicated by hemorrhage or infection or have elevated protein concentration, which results in an increased signal and slightly brighter display (Figure 2.8, Table 2.2).

Spin-Echo T2

The T2 images are obtained with a long tr and te. The T2 image is sensitive to the presence of water in tissues and depicts edema as a very bright signal. Therefore, CSF or fluid-containing structures such as cysts are very bright. Complicated cysts can vary in T2 images. If hemorrhagic, they can have a heterogenous or even uniformly dark signal caused by a susceptibility artifact. These

Parotid Gland Duct Dilatation
Figure 2.8. Axial MRI T1 weighted image at level of the skull base and brainstem without contrast demonstrating high signal in the subcutaneous fat, intermediate signal of the brain, and low signal of the CSF and mucosa. Note dilated right parotid duct (arrow).
What Intermediate Signal Mri
Figure 2.9. Axial MRI FSE T2 weighted image demonstrating the high signal of CSF and subcutaneous fat, intermediate signal of the brain and mucosa, and the low signal in the arteries.

artifacts can be caused by metals, melanin, forms of calcium, and the iron in hemoglobin. Increased tissue water from edema stands out as bright relative to the isointense soft tissue. The fast spin-echo T2 is a common sequence that is many times faster than the conventional spin-echo T2 but does alter the image. Fat stays brighter on the fast spin-echo (FSE) sequence relative to the conventional (Figure 2.9, Table 2.2).

Table 2.2. Tissue characteristics on T1 and T2 MRI.*

Proton Density Images (PD)

Proton density images are obtained with a long tr but short te, resulting in an image with less tissue contrast but high signal to noise ratio. These are uncommonly used in the head and neck.

Gradient Recalled Echo Imaging (GRE)

Gradient recalled echo imaging is the second most common type of imaging sequence after the spin-

Increased signal

Intermediate signal

Decreased signal

Calcium (a)

Proteinaceous fluid (high) (b) Slow-flowing blood Melanin

Hyperacute hemorrhage (#) (oxyhemoglobin)

Subacute hemorrhage (intracellular and extracellular methemoglobin)

Gadolinium contrast



Hyperacute hemorrhage (oxyhemoglobin) Acute hemorrhage (deoxyhemoglobin) Calcium (a) Grey matter

White matter (brighter than grey matter) Soft tissue (muscle) Proteinaceous fluid (b)

Water (CSF) or edema Fast-flowing blood Calcium (a) Soft tissue

Acute hemorrhage (deoxyhemoglobin) Chronic hemorrhage (hemosiderin) Calcification Air

Simple cyst (low protein)

Water (CSF) or edema Proteinaceous fluid Hyperacute hemorrhage (oxyhemoglobin

Subacute hemorrhage (extracellular methemoglobin) Slow-flowing blood Fat (FSE T2 scans)

Grey matter (brighter than white matter) White matter Proteinaceous fluid (b) Calcium (a)

Calcium (a) Melanin Hemosiderin Flowing blood Hemorrhagic cyst Iron deposition

Acute hemorrhage (deoxyhemoglobin) Early subacute hemorrhage (intracellular methemoglobin)

Chronic hemorrhage (hemosiderin) Air

Fast flow

Fat (conventional or non-FSE T2 scan)

(*) MRI signal on T1 and T2 predominantly from intracranial exam at 1.5T (Tesla).

(#) MRI signal of intracranial hemorrhage is quite complex and dependent on multiple factors with degrees of variability.

(a) Signal from calcium deposition is complex. Calcium concentration of under 30% by weight has high T1 signal and intermediate T2 signal, but over 40% has decreasing signal on T1 and T2. The surface area of the calcium particle also has an effect, with large surface area resulting in increased T1 signal (Henkelman, Watts, and Kucharczyk 1991).

(b) Depends on the protein concentration (complex cysts, abscess). CSF = cerebrospinal fluid.

echo. This sequence is very susceptible (more than spin-echo T2) to magnetic field inhomogeneity and is commonly used in the brain to identify blood products and metal deposition such as iron, manganese, and non-metals such as calcium. This sequence is very sensitive but not specific. The "flip angle" used in obtaining GRE can be altered resulting in either T1 weighted (long flip angle) or T2 weighted (short flip angle) images (Figure 2.10).

Short Tau Inversion Recovery (STIR)

Short tau inversion recovery is commonly acquired because of its very high sensitivity to fluid and ready detection of subtle edema in tissues. When acquired in the conventional method, STIR also results in nulling the fat signal, thereby further increasing the signal of tissue fluid relative to background. This is the best sequence for edema, particularly when trying to determine bone invasion by tumors. It can also be useful in assessing skull base foramina (Figures 2.11 and 2.12).

Gadolinium (Gd) Contrast

Intravenous contrast with gadolinium, a paramagnetic element, alters (shortens) T1 and T2 relaxation times, which results in a brighter signal. Its

Figure 2.10. Axial MRI GRE image.

Figure 2.11. Axial MRI STIR image at the skull base demonstrating the high signal of CSF but suppression of subcutaneous fat signal.
Parotid Gland Deep
Figure 2.12. Sagittal MRI STIR image at the level of the parotid gland demonstrating the deep lobe seen through the stylomandibular tunnel (arrows). Note the parotid gland extending superiorly to the skull base.
Irm Hippocampe

Figure 2.13. Coronal MRI T1 post-contrast fat saturated image of the skull base demonstrating a mass in the left parotid gland extending to the stylomastoid foramen (arrow). Note the mild vascular enhancement and suppression of fat high signal on T1 weighted image.

effect is greater on T1 than on T2 weighted images. Areas of tissue that accumulate Gd will have a higher or brighter signal and "enhance." In the head and neck, post-contrast T1 images should be performed with fat saturation to null the fat signal and therefore increase the signal of Gd accumulation (Figure 2.13).

Fluid Attenuation Inversion Recovery (FLAIR)

Fluid attenuation inversion recovery is not as commonly used in the neck but is a necessity in brain imaging. By nulling the CSF signal, brain tissue edema from a variety of causes stands out and is easily identified. It is, however, not specific. FLAIR can be useful for assessing skull base or foraminal invasion by tumors. However, artifacts can result from CSF pulsation or high FiO2 administration and can mimic pathologic processes such as sub-arachnoid hemorrhage or meningitis (bacterial, carcinomatous, viral, or aseptic) (Figure 2.14).

Diffusion Weighted Images (DWI)

Diffusion weighted images are not routinely clinically used in the neck or head but are indispens-

Right Prepontine Cistern
Figure 2.14. Axial MRI FLAIR image at the skull base demonstrating CSF flow-related artifactual increased signal in the right prepontine cistern.

able in the brain. Typical intracranial application is for assessing acute stroke, but can be applied for the assessment of active multiple sclerosis (MS) plaques, and abscesses (Figure 2.15). The concept of DWI is based on the molecular motion of water and the sensitivity of certain MRI sequences to detect the diffusion or movement of water in tissues at the cellular level.

The use of DWI and specifically apparent diffusion coefficient (ADC) values and maps for salivary gland imaging are under investigation and show promise in differentiating benign from malignant tissues (Abdel Razek, Kandeel, and Soliman et al. 2007; Eida, Sumi, and Sakihama et al. 2007; Habermann, Gossrau, and Kooijman et al. 2007; Shah et al. 2003). The ADC values are affected by technical factors (b-value setting, image resolution, choice of region-of-interest, susceptibility artifacts, and adequate shimming) as well as physiologic factors (biochemical composition of tumors, hemorrhage, perfusion, and salivary flow) (Eida, Sumi, and Sakihama et al. 2007). The ADC values of salivary glands change with gustatory stimulation. Although there are mixed results reported, there is generally an increase in the ADC value from pre-stimulation to post-stimulation measurements (Habermann, Gossrau, and Kooijman et al.

Figure 2.15. Axial MRI DWI image at the skull base demonstrating susceptibility artifact adjacent to the left temporal bone (arrow).

2007). The normal parotid, submandibular, and sublingual glands have measured ADC values of 0.63 ± 0.11 x 10-3 mm2/s, 0.97 ± 0.09 x 10-3 mm2/s, and 0.87 ± 0.05 x 10-3 mm2/s (Eida, Sumi, and Sakihama et al. 2007). In pleomorphic adenomas the ADC maps demonstrate areas of cellular proliferation to have intermediate ADC levels and areas of myxomatous changes to have high ADC values (Eida, Sumi, and Sakihama et al. 2007). Warthin's tumor showed lymphoid tissue to have a very low ADC, necrosis with intermediate ADC, and low ADC in cysts among the lymphoid tissue (Eida, Sumi, and Sakihama et al. 2007). Among the malignant lesions, mucoepidermoid carcinoma shows low ADC in a more homogenous pattern, whereas the adenoid cystic carcinomas demonstrated a more speckled pattern with areas of low and high ADC likely from multiple areas of cystic or necrotic change (Eida, Sumi, and Sakihama et al. 2007). Lymphoma in salivary glands has been demonstrated to have a diffuse extremely low ADC likely from the diffuse uniform cellularity of lymphoma (Eida, Sumi, and Sakihama et al. 2007). In general, cystic, necrotic, or myxomatous changes tend to have higher ADC, and regions of cellular-ity, low ADC. Malignant tumors tend to show very low to intermediate ADC, whereas benign lesions have higher ADC, but with a heterogenous pattern. Overlaps do occur, for example, with Warthin's tumor demonstrating very low ADC regions and adenoid cystic carcinoma with areas of high ADC (Eida, Sumi, and Sakihama et al. 2007).

Evaluating postoperative changes for residual or recurrent tumors is also an area where DWI and ADC may have a significant impact. In general (with overlap of data), residual or recurrent lesions have been shown to have ADC values lower than post-treatment changes (Abdel Razek, Kandeel, and Soliman et al. 2007). The lower ADC may be a result of smaller diffusion spaces for water in intracellular and extracellular tissues in hypercel-lular tumors. The benign post-treatment tissue with edema and inflammatory changes has fewer barriers to diffusion and increased extracellular space resulting in a higher ADC (Abdel Razek, Kandeel, and Soliman et al. 2007).

Evaluation of connective tissue disorders with DWI has demonstrated early changes with an increase in ADC prior to changes on other MRI sequences. This may be a result of early edema and/or early lymphocellular infiltration (Patel et al. 2004). Therefore, DWI and ADC may play an important role in early assessment of connective tissue disorders and preoperative evaluation of salivary tumors, as well as surveillance for recurrent disease.

MR Spectroscopy (MRS)

Magnetic resonance spectroscopy falls under the category of functional MRI (fMRI), which contains a variety of different exams created to elucidate physiologic functions of the body. DWI, spectroscopy, perfusion weighted imaging (PWI), and activation studies are examples of fMRI. Of these, MRS of brain lesions is the most commonly performed functional study in clinical imaging. Spectroscopy is, after all, the basis for MRI. MRS attempts to elicit the chemical processes in tissues. Although a variety of nuclei may be interrogated, protons, demonstrating the highest concentration in tissues, are the most practical to evaluate. The majority of MRS studies are performed for the brain, but several recent studies have evaluated head and neck tumors. The need for a very homogenous magnetic field and patient cooperation (prevention of motion) are the keys to successful MRS. Susceptibility artifact and vascular pulsation artifact add to the challenge of MRS. With higher field strength magnets MRS shows promise in determining the biochemical nature of tissues (King, Yeung, and Ahuja et al. 2005).

As in brain tumors, the most reliable markers for tumors are choline and creatine. Choline is considered to be an important constituent of cell membranes. Increased levels of choline are thought to be related to increased biosynthesis of cell membranes, which is seen in tumors, particularly those demonstrating rapid proliferation. The choline signal is comprised of signals from choline, phos-phocholine, phosphatidylcholine, and glycero-phosphocholine. Elevation of the choline peak in the MR spectra is associated with tumors relative to normal tissue. This unfortunately can be seen in malignant lesions, inflammatory processes, and hypercellular benign lesions (King, Yeung, and Ahuja et al. 2005). Another important constituent is creatine, a marker for energy metabolism. Its peak is comprised of creatine and phosphocreatine. The reduction of the creatine peak in neoplasms may represent the higher energy demands of neoplasms. The elevation of choline and more importantly the elevation of the ratio of choline to creatine has been associated with neoplasms relative to normal tissue. The elevation of choline is not tumor specific and may be seen with squamous cell carcinomas as well as a variety of salivary gland tumors, including benign tumors. It has been described in Warthin's tumor, pleomor-phic adenomas, glomus tumors, schwannomas, inflammatory polyps, and inverting papillomas (Shah et al. 2003). In fact, Warthin's tumor and pleomorphic adenoma demonstrate higher choline to creatine ratios than other tumors (King, Yeung, and Ahuja et al. 2005). King, Yeung, and Ahuja et al. also evaluated choline to water ratios and suggest that this may be an alternative method. Although the role of MRS in distinguishing between benign and malignant tumors may be limited, it nevertheless remains an important biomarker for neoplasms and plays a complementary role to other functional parameters and imaging characteristics (Shah et al. 2003). An area where MRS may play a more significant role is in a tumor's response to therapy and assessment of recurrence. Elevation of the choline to creatine ratio is seen in recurrent tumors, whereas the ratio remains low in post-treatment changes. Progressive reduction of choline is seen with a positive response to therapy and persistent elevation is seen in failure of therapy (Shah et al. 2003). Use of artificial intel ligence and neural network analysis of MR spectroscopy has demonstrated improved diagnostic accuracy of MRS using neural network analysis over linear discriminate analysis (Gerstle et al. 2000). Currently MRS of salivary gland tumors is under study and not employed clinically.

Dynamic Contrast Enhanced Magnetic Resonance Imaging

Dynamic contrast enhanced MRI has demonstrated improved diagnostic capability of tumor masses in the salivary glands and elsewhere in the body. Distinct enhancement curves can be generated based on the time points of acquisition resulting in improved differentiation of tumors (Alibek et al. 2007; Shah et al. 2003; Yabuuchi, Fukuya, and Tajima et al. 2002). However, data demonstrates similar characteristics in Warthin's tumor and malignant tumors, with a rapid increase in the signal intensity post-contrast. Pleomorphic adenoma demonstrates a more gradual increase in intensity (Alibek et al. 2007; Yabuuchi, Fukuya, and Tajima et al. 2002). Primary salivary duct carcinomas have also demonstrated the rapid enhancement as well as low ADC values as are seen with the more common primary malignancies of the salivary glands (Motoori, Iida, and Nagai et al. 2005).

Other Magnetic Resonance Imaging Techniques

In order to replace the invasive technique of digital subtraction sialography, attempts have been made to develop MR sialography. The techniques are based on acquiring heavily T2 weighted images in order to depict the ducts and branches. The lower spatial resolution and other technical factors have not allowed MR sialography to become a standard of care. This may change with newer single shot MR sequences and higher field strength magnets (Kalinowski et al. 2002; Shah et al. 2003; Takagi et al. 2005b). Dynamic MR sialography has also been used to assess function of parotid and sub-mandibular glands at rest and under stimulation (Tanaka, Ono, and Habu et al. 2007).

An extension of this concept is MR virtual endoscopy. MR virtual endoscopy can provide high-resolution images of the lumen of salivary ducts comparable to sialoendoscopy (Su et al. 2006). Although this initial experience was a preoperative assessment of the technology, it appears to be a promising method of non-invasive assessment of the ducts. In a similar manner, MR microscopy is a high-resolution imaging technique employing tiny coils enabling highly detailed images of the glands (Takagi, Sumi, and Sumi et al. 2005a). This technique was used to demonstrate morphologic changes in Sjogren's syndrome.

Use of supraparamagnetic iron oxide particle MR contrast agents has been under investigation for several years. The particles used for evaluation of lymph nodes are 20 nm or less. These are intravenously injected and are taken up by the cells in the reticuloendothelial system (RES). Since normal lymph nodes have a RES that is intact they readily take up the iron oxide agents. MR imaging using T2 and T2* weighted images demonstrate susceptibility to the iron oxide and result in signal loss at sites of iron accumulation. Therefore, normal lymph nodes lose signal, whereas metastatic lymph nodes whose RES has been replaced by metastases do not take up the particles and do not lose signal (Shah et al. 2003). Although not a direct imaging technique for the salivary glands, it may prove to be useful in the evaluation of nodal metastases.


Ultrasound is performed infrequently for head and neck imaging relative to CT and MRI. Although US is able to depict normal anatomy and pathology in the major salivary glands, it is limited in evaluation of the deep lobe of the parotid and subman-dibular glands (Figures 2.16 and 2.17). US is operator dependent and takes significantly longer to perform on bilateral individual salivary glands

Mylohyoid Muscle Images Ultrasound

Figure 2.16. Ultrasound of the submandibular gland (black arrow) adjacent to the mylohyoid muscle (white arrow).


Figure 2.16. Ultrasound of the submandibular gland (black arrow) adjacent to the mylohyoid muscle (white arrow).

when compared to contrast enhanced CT of the entire neck. US is quite effective at delineating cystic from solid masses and determining degree of vascularity. US can be used to image calculi and observe the resulting ductal dilatation. Normal lymph nodes and lymphadenopathy can also be reliably distinguished. US can be used to initially stage disease. It is not, however, optimal for post-therapy follow-up, be it radiation or surgery. When compared with CT or MRI, US significantly lacks in soft tissue resolution and contrast. Because of its real-time imaging capability and ease of handheld imaging, US is quite good at image guided fine needle aspiration and biopsy. The application of color Doppler or power Doppler US can distinguish arteries from veins, which is critical for image guided biopsy (Figures 2.18 and 2.19). Eigh-

Normal Volume Parotid Gland
Figure 2.17. Ultrasound of the parotid gland demonstrating a normal intraparotid lymph node on a hyperechoic background. The lymph node is round and has a hypoechoic rim but demonstrates a fatty hyperechoic hilum (arrow).

«Mt 1 'Vi I !*!t PvhplhL !4mi PK <M*t *£rrr F'¥ i>V :


LTSAG PAq-jTltl [CJl zaii^

Figure 2.18. Ultrasound of the parotid gland in longitudinal orientation demonstrating the Doppler signal of the external carotid artery.

Intraparotid Nodes Ultrasound
Figure 2.19. Ultrasound of the parotid gland in longitudinal orientation demonstrating the Doppler signal of the retromandibular vein.

teen gauge core biopsies of the parotid may be safely performed under US guidance (Wan, Chan, and Chen 2004).

US Technique

High-frequency transducers such as 5, 7.5, or 10 MHz are typically applied to image superficial small parts. Real-time imaging and image acquisition is performed by a technologist or physician. Doppler US may be applied to observe the vascularity of the glands (increased in inflammatory conditions) or tumors within the glands. Doppler US can easily determine arterial from venous channels.


Radionuclide imaging has, throughout its history, been a functional imaging modality without the quality of anatomic depiction when compared with CT, MRI, or even US. The majority of radionuclide imaging has been performed with planar imaging systems that produce single view images of functional processes. All RNI exams employ a radioactive tracer either bound to a ligand (radiopharmaceutical) or injected directly (radio-nuclide). As the radionuclide undergoes radioactive decay it emits either a gamma ray (photon) and/or a particle such as an alpha particle (helium nucleus), beta particle (electron), or a positron (a positively charged electron). Gamma rays differ from X-rays in that gamma rays (for medical imaging) are an inherent nuclear event and are emitted from the nucleus of an unstable atom in order to achieve stability. X-rays (in the conventional sense) are produced in the electron cloud surrounding the nucleus. In medical imaging X-rays are artificially or intentionally produced on demand, whereas gamma rays (and other particles) are part of an on-going nuclear decay enabling unstable radioactive atoms to reach a stable state. The length of time it takes for one-half of the unstable atoms to reach their stable state is called their half-life. Radionuclide imaging involves the emission of a photon, which is imaged using a crystal or solid state detector. The detector may be static and produces images of the event in a single plane or the detector may be rotated about the patient in order to gather three-dimensional data and reconstruct a tomographic image in the same manner as a CT scanner. This is the basis for SPECT. Examples of planar images used in salivary gland diseases include Gallium (67Ga) for evaluation of inflammation, infection, and neoplasms (lymphoma). SPECT, which produces tomographic cross-sectional images, is less commonly used in oncologic imaging, although novel radionuclides and ligands are under investigation. The recent introduction of SPECT/CT, a combined functional and anatomic imaging machine, may breathe new life into SPECT imaging.


Positron emission tomography is a unique imaging modality that records a series of radioactive decay events. Positron emission is a form of radioactive decay in which a positron (positively charged electron) is emitted from the unstable atom in order to achieve a more stable state. The positron almost immediately collides with an electron (negatively charged) and undergoes an annihilation event in which both particles are destroyed and converted into pure energy. The annihilation event produces two gamma rays, each with 511 Kev (kilo electron volt) of energy, and traveling in 180-degree opposition. By using sophisticated solid state detectors and coincidence circuitry, the PET system is able to record the source of the event, thereby localizing the event in three-dimensional space. Using a complex algorithm similar to SPECT and CT, a three-dimensional block of data is produced and can be "sliced" in any plane, but most commonly in axial, coronal, and sagittal planes, as well as a maximum intensity projection (MIP) rendering.

PET radionuclides are produced in a cyclotron and are relatively short lived. Typical radionuclides include 18F, UC, 15O, 82Rb, and 13N. A variety of ligands have been labeled and studied for the evaluation of perfusion, metabolism, and cell surface receptors. The most commonly available is 18F-deoxyglucose (FDG), which is used to study glucose metabolism of cells. Most common uses of FDG include oncology, cardiac viability, and brain metabolism. PET has a higher spatial resolution than SPECT. Both systems are prone to multiple artifacts, especially motion. Acquisition times for both are quite long, limiting the exam to patients who can lie still for prolonged periods of time. Both systems, but PET in particular, are very costly to install and maintain. Radiopharmaceuti-cals are now widely available to most institutions through a network of nuclear pharmacies.

The oncologic principle behind FDG PET is that neoplastic tissues can have a much higher metabolism than normal tissues and utilize glucose at a higher rate (Warburg 1925). Glucose metabolism in the brain was extensively studied using autoradiography by Sokoloff and colleagues at the National Institutes of Health (NIH) (Sokoloff 1961). The deoxyglucose metabolism is unique in that it mimics glucose and is taken up by cells using the same transporter proteins. Both glucose and deoxy-glucose undergo phosphorylation by hexokinase to form glucose-6-phosphate. This is where the similarities end. Glucose-6-phosphate continues to be metabolized, eventually to form CO2 and H2O. Deoxyglucose-6-phosphate cannot be further metabolized and becomes trapped in the cell, as it cannot diffuse out through the cell membrane. Therefore, the accumulation of FDG reflects the relative metabolism of tissues (Sokoloff 1986). The characteristic increased rate of glucose metabolism by malignant tumors was initially described by Warburg and is the basis of FDG PET imaging of neoplasms (Warburg 1925).

FDG PET takes advantage of the higher utilization of glucose by neoplastic tissues to produce a map of glucose metabolism. Although the FDG PET system is sensitive, it is not specific. Several processes can elevate glucose metabolism, including neoplastic tissue, inflammatory or infected tissue, and normal tissue in a high metabolic state. An example of the latter includes uptake of FDG in skeletal muscle that was actively contracting during the uptake phase of the study. (Figure 2.20a-d). Another peculiar hypermetabolic phenomenon is brown adipose tissue (BAT) FDG uptake (Figure 2.21a-d). BAT is distributed in multiple sites in the body including interscapular, paravertebral, around large blood vessels, deep cervical, axillary, mediastinal, and intercostal fat, but is concentrated in the supraclavicular regions (Cohade et al. 2003; Tatsumi, Engles, and Ishimori et al. 2004). BAT functions as a thermogenic organ producing heat in mammals, and most commonly demonstrates uptake in the winter (Tatsumi, Engles, and Ishimori et al. 2004). BAT is innervated by the sympathetic nervous system, has higher concentration of mitochondria, and is stimulated by cold temperatures (Cohade, Mourtziks, and Wahl 2003; Tatsumi, Engles, and Ishimori et al. 2004). Administration of ketamine anesthesia in rats markedly increased FDG uptake presumably from sympathetic stimulation (Tatsumi, Engles, and Ishimori et al. 2004). Although typically described on FDG PET/CT exams, it can be demonstrated with 18F-Fluorodo-pamine PET/CT, 99mTc-Tetrofosmin, and 123I-MIBG SPECT as well as 201TlCl and 3H-l-methionine (Baba, Engles, and Huso et al. 2007; Hadi, Chen, and Millie et al. 2007). Propranalol and Reserpine administration appears to decrease the degree of FDG uptake, whereas diazepam does not appear to have as significant an effect (Tatsumi, Engles, and Ishimori et al. 2004). Exposure to nicotine and ephedrine also resulted in increased BAT uptake; therefore, avoiding these substances prior to PET scanning can prevent or reduce BAT uptake (Baba, Engles, and Huso et al. 2007). Preventing BAT uptake of FDG can be accomplished by having the patient stay in a warm ambient temperature for 48 hours before the study and by keeping the patient warm during the uptake phase of FDG PET (Cohade, Mourtzikos, and Wahl 2003; Delbeke, Coleman, and Guiberteau et al. 2006). Although somewhat controversial, diazepam or lorazepam and propranalol can reduce BAT uptake by blocking sympathetic activity as well as reducing skeletal muscle uptake from reduced anxiety and improved relaxation (Delbeke, Coleman, and Guib-erteau et al. 2006). Understanding the distribution of BAT and the physiology that activates BAT, as well as recognizing the uptake of FDG in BAT in clinical studies, is critical in preventing a false positive diagnosis of supraclavicular, paraverte-bral, and cervical masses or lymphadenopathy.

Increased Uptake Salivary Glands Fdg Pet

Figure 2.20. CT (a), PET (b), and fused PET/CT (c) images in axial plane, and an anterior MIP image (d) demonstrating skeletal muscle uptake in the sternocleidomastoid muscle and biceps muscle (arrows). Also note the intense uptake in the abdominal, psoas, and intercostal muscles on the MIP image. The very high focal uptake in the middle of the image is myocardial activity.

FDG uptake in all salivary glands in the normal state is usually mild and homogenous (Burrell and Van den Abbeele 2005; Wang et al. 2007) (Figures 2.22a and b and 2.23a and b). After therapy, radiation, or chemotherapy, the uptake can be very high (Burrell and Van den Abbeele 2005). Standardized uptake value (SUV), a semiquantitative measurement of the degree of uptake of a radiotracer (FDG), may be calculated on PET scans. There are many factors that impact the measurement of SUVs, including the method of attenuation correction and reconstruction, size of lesion, size of region of interest, motion of lesion, recovery coefficient, plasma glucose concentration, body habitus, and time from injection to imaging (Wang et al. 2007, Schoder, Erdi, and Chao et al. 2004 and Beaulieu, Kinaha, and Tseng et al. 2003).

A range of SUVs can be calculated in normal volunteers for each salivary gland. Wang et al.

Figure 2.21. PET image (a), corresponding CT image (b), and a fused PET/CT image (c) in the axial plane demonstrating BAT uptake in the supraclavicular regions bilaterally, which could mimic lymphadenopathy (see arrows on Figures 2.21a and 2.21b). Direct correlation enabled by the PET/CT prevents a false positive finding. Note the similar uptake on the MIP image (arrow) (d) including paraspinal BAT uptake.

a b c measured SUVs in normal tissues to determine the maximum SUV and mean SUV as well as assignment of an uptake grade ranging from none (mean SUV less than aortic blood pool), mild (mean SUV greater than mean SUV of aortic blood pool but less then 2.5), moderate (mean SUV between 2.5 and 5.0), and intense (mean SUV greater than 5.0). SUV greater than 2.5 was considered significant (Wang et al. 2007). Parotid glands (n = 97) had a range of SUVmax of 0.78-20.45 and an SUVmean range of 1.75 ± 0.79. Fifty-three percent of the SUV measurements fell into the "none" category, 33% into the "mild" category, and 14% into the "moderate" category. No SUV measurement fell into the "intense" category. Submandibular glands (n = 99) had an SUVmax range of 0.56-5.14 and an SUVmean of

2.22 ± 0.77. The uptake grades consisted of the following: 25% were in the "none" category, 44% in the "mild," and 31% were "moderate." The sublingual gland (n = 102) had an SUVmax range of 0.93-5.91 and an SUVmean of 4.06 ± 1.76. Four percent of these fell into the "none" category, 19% in the "mild," 54% in the "moderate," and 23% in the "intense" group (Wang et al. 2007). Similar work by Nakamoto et al. demonstrated a SUVmean of 1.9 ± 0.68 for the parotid gland, 2.11 ± 0.57 for the submandibular gland, and 2.93 ± 1.39 for the sublingual gland (Nakamoto, Tatsumi, and Hammoud et al. 2005). This demonstrates the wide range of normal uptake values (Table 2.3).

Although FDG does accumulate in the saliva, the concentration varies from 0.2 to 0.4 SUV but

Wr* il rll

Was this article helpful?

0 0
How To Boost Your Metabolism

How To Boost Your Metabolism

In The Next 45 Minutes You're Going To Discover How To Boost Your Metabolism And Lose Weight. Who Else Wants To Boost Their Metabolism And Shed Pounds Fast?

Get My Free Ebook

Post a comment