Ultrasound versus magnetic resonance arthrography in acetabular labral tear diagnostics: a prospective comparison in 20 dysplastic hips

Background: Acetabular labral tears are highly associated with hip dysplasia. Magnetic resonance arthrography (MR arthrography) is the expensive and time-consuming contemporary gold-standard method in the radiological assessment of acetabular labral tears.

Purpose: To assess the diagnostic ability of noninvasive ultrasound (US) examination compared to MR arthrography in diagnosing acetabular labral tears in dysplastic hip joints.

Material and Methods: The study compared US examination and MR arthrography diagnosis of labral tears in 20 consecutively referred dysplastic hip joints.

Results: The ability to diagnose acetabular labral tears upon US examination was calculated: sensitivity 44%, specificity 75%, positive predictive value 88%, and negative predictive value 25%.

Conclusion: The ability of US examination in diagnosing acetabular labral tears is not yet good enough. The technique is still to be developed, and more experience, especially with the interpretation of US examinations, is needed.     

A new computed tomography-based radiographic method to detect early loosening of total wrist implants

Background: Diagnosis of loosening of total wrist implants is usually late using routine radiographs. Switching modality to computed tomography (CT) should aid in early diagnosis.

Purpose: To propose and evaluate the accuracy of a new CT method for assessing loosening of the carpal component in total wrist arthroplasty.

Material and Methods: A protocol encompassing volume registration of paired CT scans of patients with unexplained pain in a prosthetically replaced wrist (used in clinical routine) is presented. Scans are acquired as a dynamic examination under torsional load. Using volume registration, the carpal component of the prosthesis is brought into spatial alignment. After registration, prosthetic loosening is diagnosed by a shift in position of the bones relative to the prosthesis. This study is a preclinical validation of this method using a human cadaverous arm with a cemented total wrist implant and tantalum markers. Seven CT scans of the arm were acquired. The scans were combined into 21 pairs of CT volumes. The carpal component was registered in each scan pair, and the residual mismatch of the surrounding tantalum markers and bone was analyzed both visually and numerically.

Results: The detection limit for prosthetic movement was less than 1 mm.

Conclusion: The results of this study demonstrate that CT volume registration holds promise to improve detection of movement of the carpal component at an earlier stage than is obtainable with plain radiography.     

64-slice computed tomography assessment of coronary artery stents: a phantom study

Purpose: To compare the use of a new 64-slice computed tomography (CT) scanner with 16-slice CT in the visualization of coronary artery stent lumen.

Material and Methods: Eight different coronary artery stents, each with a diameter of 3 mm, were placed in a static chest phantom. The phantom was positioned in the CT gantry at an angle of 0° and 45° towards the z-axis and examined with both a 64-slice and a 16-slice CT scanner. Effective slice thickness was 0.6 mm with 64-slice CT and 1 mm with 16-slice CT. A reconstruction increment of 0.3 mm was applied in both scanners. Image quality was assessed visually using a 5-point grading scale. Stent diameters were measured and compared using paired Wilcoxon tests.

Results: Artificial lumen reduction was significantly less with 64-slice than with 16-slice CT. Average visible stent lumen was 53.4% using 64-slice CT and 47.5% with 16-slice MSCT. Most severe artifacts were seen in stents with radiopaque markers. Using 64-slice CT, image noise increased by approximately 30% due to thinner slice thickness.

Conclusion: Improved spatial resolution of 64-slice CT resulted in superior assessment of coronary artery stent lumen compared to 16-slice CT. However, a relevant part of the stent lumen is still not assessable with multi-slice CT.     

Model studies on acetabular component migration in total hip arthroplasty using CT and a semiautomated program for volume merging

Purpose: 
Validation of a non-invasive CT method for detection of acetabular cup migration after total hip arthroplasty in a phantom study. Material and Methods: 
26 CT examinations were obtained of a pelvic model while altering the position of the acetabular cup. Using a previously described program for volume merging, the pelvi in different examinations were fused and the 3D alterations of the position of the acetabular cup were evaluated visually and numerically and correlated to direct measurements on the model. Results: 
Visually, two independent examiners differentiated between 0, 1 and 2 to 3 mm migration with 100% specificity and sensitivity. Numerically, the mean error over all cases between model and CT measurements was 0.04 mm (SD ± 0.33). The mean absolute error between model and CT data was 0.26 mm (SD ± 0.19). Intra- and interobserver 95% accuracy and repeatability limits were below 0.5/0.7 mm, respectively. No significant interobserver difference occurred. The data were normally distributed and not dependent on observer. Conclusion: 
The accuracy of this non-invasive method out-performs routine plain radiography. The method gives both visual and numerical correlates to migration and can be used in clinical practice.     

Screening for carotid injury in trauma patients: image quality of 16-detector-row computed tomography angiography

Background: The introduction of multidetector-row computed tomography (MDCT) has revolutionized the initial management of multiply injured patients. This technology has the potential to improve the imaging of traumatic vascular injuries.

Purpose: To evaluate the quality of multidetector-row computed tomography angiography (MDCTA) of the carotid arteries in the setting of a routine whole-body trauma scan.

Material and Methods: 87 trauma patients underwent a routine whole-body CT scan in a 16-detector-row scanner including an MDCTA with a reconstructed axial slice thickness of 3 mm. Images were reviewed by three experienced radiologists with emphasis on image quality. Contrast density, severity, and origin of artifacts and the occurrence of vessel lesions were assessed for different vessel segments.

Results: 3642 separate vessel segments were evaluated. Contrast density was rated good or sufficient for diagnosis in 99.8%. A total of 67.3% of vessel segments were free of artifacts, while 27.9% of vessel segments showed minor artifacts not impairing diagnostic evaluation. Clinically relevant artifacts obscuring a vessel segment occurred in 4.7% and were mostly caused by dental hardware. Four dissections of the internal carotid artery were diagnosed by all three radiologists.

Conclusion: As a rapid screening test for blunt carotid artery injury, integration of MDCTA in the routine imaging workup of trauma patients utilizing a whole-body CT trauma scan is possible and practicable. Image quality is mostly sufficient for diagnosis, but impaired in a few cases by artifacts deriving primarily from dental hardware.     

High-resolution computed tomography with single-slice computed tomography and 16-channel multidetector computed tomography: a comparison regarding visibility and motion artifacts

Background: High-resolution computed tomography is the image procedure of choice in the evaluation of interstitial lung disease. Multidetector-row computed tomography provides the possibility of simultaneous reconstruction of thin and thick slices from the same raw data, acquired from one single series. Thus, it may be tempting to exclude the step-and-shoot series.

Purpose: To compare high-resolution computed tomography (HRCT step-and-shoot) from single-slice CT (SSCT) and 16-channel multidetector CT (MDCT) in terms of visibility and motion artifacts, and to investigate whether thin images reconstructed from helical MDCT are equal to or better than conventional HRCT by SSCT in terms of visibility and motion artifacts.

Material and Methods: 20 patients underwent HRCT step-and-shoot by SSCT (SSCT step-and-shoot) and MDCT (MDCT step-and-shoot), and a helical MDCT acquisition (MDCT helical). Images from four anatomical levels were analyzed in random order regarding visibility and motion artifacts.

Results: Visibility using MDCT step-and-shoot was significantly better than or equal to SSCT step-and-shoot for segmental bronchi and fissures, but not for subsegmental bronchi. For MDCT helical, visibility was equal to or better than SSCT step-and-shoot for segmental bronchi, but not for fissures and subsegmental bronchi. Concerning motion artifacts, MDCT step-and-shoot and MDCT helical were significantly better than or equal to SSCT step-and-shoot.

Conclusion: The image quality (accounting for motion artifacts and visibility) of SSCT step-and-shoot and MDCT step-and-shoot is comparable. The visibility of anatomic structures in images from MDCT helical is inferior to HRCT step-and-shoot.     

Hyperperfusion on perfusion computed tomography following revascularization for acute stroke

Purpose: To describe the findings of hyperperfusion on perfusion computed tomography (CT) in four patients following revascularization for acute stroke.

Material and Methods: In 2002-2003, among a series of 6 patients presenting with an acute stroke and treated with intra-arterial thrombolysis, we observed the presence of hyperperfusion in 3 patients on the follow-up CT perfusion. We included an additional patient who was treated with intravenous thrombolysis and who had hyperperfusion on the follow-up CT perfusion. We retrospectively analyzed their CT perfusion maps. Cerebral blood volume (CBV) and cerebral blood flow (CBF) maps were compared between the affected territory and the normal contralateral hemisphere.

Results: In the four patients, the mean CBV and CBF were 3.6±2.0 ml/100 g and 39±25 ml/100 g/min in the affected territory compared to the normal side (mean CBV = 2.7±2.1 ml/100 g, mean CBF = 27±23 ml/100 g/min). There was no intracranial hemorrhage in the hyperperfused territories. At follow-up CT, some hyperperfused brain areas progressed to infarction, while others retained normal white to gray matter differentiation.

Conclusion: CT perfusion can demonstrate hyperperfusion, which can be seen in an ischemic brain territory following recanalization.     

Cervical lymph node metastases of unknown origin: primary tumor detection with whole-body positron emission tomography/computed tomography

Background: Identification of primary tumor in patients with cervical lymph node metastasis of unknown primary (MUO) has a great impact on therapy approach and potentially on patient prognosis.

Purpose: To assess the diagnostic accuracy of combined positron emission tomography (PET)/computer tomography (CT) for primary tumor detection in cervical metastases of unknown origin compared to PET, CT, and PET+CT side-by-side evaluation.

Material and Methods: 39 consecutive patients (eight women, 31 men; mean age 59.9±11.2 years) with MUO were enrolled in this study. PET/CT images were obtained 1 hour after injection of 350 MBq of fluorodeoxyglucose. Oral and intravenous contrast agents were administered in all patients to ensure diagnostic CT data. Fused PET/CT data were evaluated for primary tumor detection. Diagnostic accuracy was calculated and compared with CT alone, PET alone, and side-by-side PET+CT evaluation. Statistical analysis of differences in diagnostic performance between the different imaging procedures was based on the McNemar test.

Results: Fused PET/CT depicted the primary tumor in 11 of 39 (28%) patients. In 28 (72%) patients, the primary tumor remained occult. CT revealed the primary in five (13%), PET alone in 10 (26%), and side-by-side evaluation of PET+CT in 10 (26%) of 39 patients. Statistical analysis showed no significant differences between the imaging modalities.

Conclusion: PET, side-by-side PET+CT, and PET/CT revealed similar detection rates for primary tumors in cervical MUO patients. Therefore, cervical metastases of an unknown primary may be assessed with either of these imaging modalities. Detection rates with CT were substantially lower. Thus, inclusion of functional data for assessment of cervical MUO patients must be recommended.     

Imaging of four-corner fusion (SLAC arthrodesis) of the wrist with 64-slice computed tomography

Purpose: To find out whether it is possible to evaluate the healing of wrist arthrodesis, carried out with a metallic spider plate, by means of 64-slice computed tomography (CT).

Material and Methods: 18 CT examinations were performed in 12 patients 2 weeks to 37 months following scapholunate advanced collapse (SLAC) arthrodesis fixed with a metallic plate. Ten patients also had plain films of the wrist. Radiation doses were estimated.

Results: Plain films were difficult to evaluate due to overprojection of the spider plate. With 64-slice CT, however, it was possible to evaluate the healing process in all patients in spite of metallic artifacts. Radiation doses were low.

Conclusion: The healing of SLAC arthrodesis of the wrist is difficult to evaluate with conventional radiography due to the metallic plate. By means of 64-slice CT, however, it was possible to “see under” the plate in all 12 patients.     

Validation of the rotation ratios method

Background: The rotation ratios method describes rotations between pairs of sequential pelvic radiographs. The method seems promising but has not been validated.

Purpose: To validate the accuracy of the rotation ratios method.

Material and Methods: Known pelvic rotations between 165 radiographs obtained from five skeletal pelvises in an experimental material were compared with the corresponding calculated rotations to describe the accuracy of the method. The results from a clinical material of 262 pelvic radiographs from 46 patients defined the ranges of rotational differences compared. Repeated analyses, both on the experimental and the clinical material, were performed using the selected reference points to describe the robustness and the repeatability of the method.

Results: The reference points were easy to identify and barely influenced by pelvic rotations. The mean differences between calculated and real pelvic rotations were 0.0° (SD 0.6) for vertical rotations and 0.1° (SD 0.7) for transversal rotations in the experimental material. The intra- and interobserver repeatability of the method was good.

Conclusion: The accuracy of the method was reasonably high, and the method may prove to be clinically useful.     

Angulation, length, and morphology of the styloid process of the temporal bone analyzed by multidetector computed tomography

Purpose: To investigate the angulation, length, and structural variations of the styloid process (SP) by multidetector computed tomography (MDCT).

Material and Methods: MDCT scans were performed in 283 cases (127 M and 156 F, age range 18-77 years). The length of the SP and its angulation on the transverse and sagittal planes were measured. Structural variations of the SP were observed by means of three-dimensional (3D) and multiplanar reconstruction (MPR) images.

Results: The length of the bony SP on both sides varied from 0 to 62 mm (mean 26.8±10.0 mm). Angulation ranged between 55° and 90.5° (7 2.7±6.6) in the transversal plane and between 76° and 110° (93.5±6.9) in the sagittal plane. Morphologically, the SP showed a considerable amount of variation. A solitary SP was present in 168 individuals (59.4%). In 9 individuals (3.1%), the SP was duplicated (4 unilateral and 5 bilateral). Sixty-one persons (21.6%) showed an incomplete ossified SP (42 unilateral and 19 bilateral), whereas in 7 individuals (2.5%) a bony SP was absent entirely (7 unilateral). In 38 individuals (13.4%), the stylohyoid ligament was ossified (16 entirely, 22 partial). In all individuals, 3D and MPR images showed the SP in its entire length.

Conclusion: MDCT with 3D CT and MPR of SP may show further detailed information related to SP. Not only the length of the SP, but also its 3D orientation, should be in focus in anatomical and clinical studies.     

Split renal function in patients with suspected renal artery stenosis: a comparison between gamma camera renography and two methods of measurement with computed tomography

Purpose: To validate a method for calculating split renal function from computed tomography (CT) compared with gamma camera renography, and to test a new method for the measurement based on a volume-rendering technique.

Material and Methods: Thirty-eight patients, aged 65.7±11.6 (range 37.8-82.1) years, who had undergone both CT angiography and gamma camera renography for a suspected renal artery stenosis were included in this study. Split renal function was calculated from the CT examinations by measuring area and mean attenuation in the image slices of the kidneys, and also by measuring volume and mean attenuation from a 3D reconstruction of the kidneys. Gamma camera renography with ^99mTc-MAG3 with or without captopril enhancement was used as a reference.

Results: The 2D CT method had good correlation with renography (r = 0.93). Mean difference was 4.7±3.6 (0-12) percentage points per kidney. There was also excellent correlation between the two CT methods (r = 1.00).

Conclusion: CT is equivalent to renography in determining split renal function, and the measurement from the CT examination can be made more quickly and equally accurately with a 3D technique.     

Pediatric multidetector computed tomography using tube current modulation and a patient image gallery

Background: Dose reduction is crucial in pediatric multidetector computed tomography (MDCT).

Purpose: To perform pediatric 16-slice MDCT using tube current modulations and to adjust prospectively the tube current using a patient image gallery (IG) providing simulated dose-reduced protocols; and to evaluate and compare the image quality of the IG and the clinical MDCT.

Material and Methods: 30 examinations (thorax, n=15; abdomen, n=8; pelvis, n=7) in 20 patients (nine male, age 8.05±7.33 years, weight 29.8±24.02 kg) were performed according to an IG on a 16-slice MDCT with additional use of online tube current modulation (CARE Dose). Three radiologists visually assessed image quality from the IG and actual 16-slice MDCT scans. For objective analysis, image noise was determined.

Results: Statistical analysis showed moderate concordance in objective (K =0.68-0.78) and subjective (K =0.33-0.64) image assessment between the IG and clinical 16-slice MDCT scans. Depending on the weight group and clinical question, no or only minor dose reductions in the chest, but moderate to considerable reductions in the abdominal/pelvic 16-slice MDCT scans compared to previously used pediatric protocols were achieved. Extra dose reduction was achieved due to additional use of CARE Dose depending on age group and scan region (mean 8.6-23.9%).

Conclusion: The IG enabled us to prospectively reduce the tube current and adapt the required image quality to the clinical question. Additional dose reduction was achieved with application of CARE Dose; nevertheless, the images are comparable to the simulated images of the IG.     

Computed-tomography-guided punctures using a new guidance device

Purpose: To evaluate a new adjunctive guidance device, a puncture guide, constructed to simplify computed tomography (CT)-guided punctures and to make the procedure more accurate and safe.

Material and Methods: 17 patients referred for CT-guided punctures were included in the study. There were 10 thoracic and 7 abdominal or pelvic lesions with a mean maximum diameter of 29±18 mm. All punctures were performed using a laser guide combined with the new device. The needle guide created a streak artefact in the image, indicating the needle path.

Results: The puncture was successful at the first attempt in 15 of the 17 patients. The artefact was visible in all patients, and in the majority there was a distinct artefact reaching from the entry point to the lesion. The deviation between the angle of the streak artefact and the final angle of the needle was 1.1°.

Conclusion: The benefits of the puncture guide were the artefact pointing at the target, the needle support, and accuracy when performing CT-guided punctures.     

Pulmonary cement embolism: a complication of percutaneous vertebroplasty

Background: Percutaneous vertebroplasty (PV) has recently become a very common procedure for vertebral compression fractures. Extravasation of cement, a common event associated with vertebroplasty, may lead to cement emboli in the lungs.

Purpose: To determine the frequency of pulmonary cement embolism after percutaneous vertebroplasty.

Material and Methods: Between 2002 and 2006, 128 percutaneous vertebroplasties were performed in 73 patients (56 women and 17 men) in our institution. Postprocedural chest radiographs were obtained for all patients and assessed for the presence of pulmonary cement emboli.

Results: Pulmonary cement embolism was detected on chest radiographs and confirmed with chest computed tomography (CT) in four patients treated with percutaneous vertebroplasty for osteoporotic collapse and one patient treated for multiple myeloma. The imaging finding of pulmonary cement embolism was solitary or multiple fine radiodense lines with occasional branching patterns. The frequency of pulmonary cement embolism was 6.8%.

Conclusion: An incidence of pulmonary cement embolism of 6.8% during PV was found. Close clinical follow-up, postprocedural chest radiographs, and chest CT scans, if necessary, are important for the detection of pulmonary cement embolism at an early stage.     

What are radiologists doing to prevent contrast-induced nephropathy (CIN) compared with measures supported by current evidence? A survey of European radiologists on CIN associated with computed tomography

Background: Contrast-induced nephropathy (CIN) is a serious complication of the use of iodinated contrast media (CM), and is associated with increased morbidity and mortality.

Purpose: To investigate whether radiologists take sufficient measures to prevent CIN in computed tomography (CT).

Material and Methods: 2005 survey of 509 European radiologists who had ≥3 years' experience and performed ≥50 CT scans/week.

Results: The most common methods used to identify patients at risk of CIN were renal function measurements (64%), clinical judgment (55%), and patient questionnaires (31%); 9% made no routine attempt to identify at-risk patients. The most common preventive protocols used in at-risk patients included: intravenous (i.v.) saline volume repletion (59%) or oral hydration (52%) before/after CT; use of low-osmolar CM (LOCM; 40%) or isosmolar CM (IOCM; 36%); and N-acetylcysteine (20%); 8% used no hydration regimen. While 78% of respondents used ≤100 ml of CM in high-risk patients, 14% used ≤150 ml, and 9% set no volume limit. For 57% of respondents, osmolality was the most important attribute in choosing an iodinated CM in at-risk patients; 41% agreed that CIN risk is lower with IOCM versus LOCM (31% disagreed).

Conclusion: A European radiologist survey identified a need for increased implementation of evidence-based protocols to improve CIN prevention: routine identification of at-risk patients; withdrawal of nephrotoxic drugs; use of volume repletion regimens; lowest possible volume of CM; and appropriate CM.     

Low-field magnetic resonance imaging of the lumbar spine: reliability of qualitative evaluation of disc and muscle parameters

Purpose: To determine the intra- and interobserver reliability in grading disc and muscle parameters using low-field magnetic resonance imaging (MRI).

Material and Methods: MRI scans of 100 subjects representative of the general population were evaluated blindly by two radiologists. Criteria for grading lumbar discs were based on the spinal nomenclature of the Combined Task Force and the literature. Consensus in rating was achieved by evaluating 50 MRI examinations in tandem. The remaining 50 examinations were evaluated independently by the observers to determine interobserver agreement and re-evaluated by one of the observers to determine intra-observer agreement.

Results: Intra- and interobserver agreement was substantial when grading changes in the lumbar discs. Interobserver agreement was fair to moderate in grading the lumbar muscles, whereas intra-observer agreement was almost perfect.

Conclusion: Convincing reliability was found in the evaluation of disc- and muscle-related MRI variables.     

Neonatal hip instability: a prospective comparison of clinical examination and anterior dynamic ultrasound

Background: Ultrasound is increasingly being used to complement the clinical examination in assessing neonatal hip instability. The clinical examination, although highly sensitive in detecting hip instability, can lead to considerable overtreatment.

Purpose: To compare anterior dynamic ultrasound and clinical examination in the assessment of neonatal hip instability and regarding treatment rates.

Material and Methods: 536 newborn infants (out of a population of 18,031) were selected, on the basis of a combination of risk factors, clinical signs of hip instability or ambiguous clinical findings, to undergo an anterior dynamic ultrasound examination of the hip, by a method developed by our group. This examination, performed by one out of seven experienced examiners, was compared with the standard clinical hip examination conducted by one of four pediatric orthopedic surgeons. The clinical examination was carried out both prior to and within a few hours after the ultrasound examination.

Results: The clinical examination diagnosed 81.7% of the hips as normal, 14.5% as unstable, and 3.8% as dislocatable or dislocated. With the dynamic ultrasound method, the corresponding figures were 87.8%, 10.4%, and 1.8%, respectively. Use of the criteria of the clinical examination resulted in treatment of 147 infants. Using the dynamic ultrasound examination as a criterion meant that 87 infants would receive treatment. The calculated treatment rate was 0.85% when based on the clinical stress test and 0.49% when based on the dynamic ultrasound.

Conclusion: The dynamic ultrasound results reduced the treatment rate by over 40% when used as a basis for the decision regarding treatment.     

3D imaging of the whole spine at 3T compared to 1.5T: initial experiences

Purpose: To describe our first experiences with a recently introduced 3T system for T2-weighted isotropic 3D whole-spine imaging.

Material and Methods: Magnetic resonance imaging of the whole spine was performed by implementing an isotropic 3D fast spin-echo sequence with variable flip-angle refocusing pulses at 3T and 1.5T. Signal-to-noise ratio (SNR) was compared on both systems in eight subjects.

Results: Mean values for SNR were significantly higher at 3T (346±214) for 1.0 mm voxel size compared to 1.5T (202±41), but showed considerably higher variability at 3T. At 3T, measurements with 0.9 mm voxel size were possible with similar SNR as with 1.0 mm voxel size at 1.5T.

Conclusion: High-resolution 3D imaging of the whole spine is feasible at 3T with an increased SNR compared to 1.5T. Signal gain at 3T can be used to further increase spatial resolution. Pronounced interpatient variability of SNR at 3T may be a result of inhomogeneous RF deposition due to dielectric effects.     

Value of apparent diffusion coefficient calculation in the differential diagnosis of parotid gland tumors

Background: The differential diagnosis of parotid gland tumors is often difficult with conventional magnetic resonance imaging.

Purpose: To determine whether the calculation of the apparent diffusion coefficient (ADC) is valuable for making the differential diagnosis of parotid tumors.

Material and Methods: Thirty parotid masses in 28 patients and 24 healthy parotid glands in 12 controls were examined in this prospective study. Diffusion-weighted magnetic resonance imaging with echo-planar spin-echo sequences was used to evaluate each subject. The ADC of each tumor and each healthy parotid gland was calculated. Tumor diagnoses were confirmed by the results of histopathologic analysis.

Results: The following types of masses were identified: 11 Warthin tumors, nine pleomorphic adenomas, seven malignant tumors, one basal cell adenoma, and two benign cysts. The mean ADC value for the Warthin tumors was 0.97±0.16×10^-3 mm²/s, for the pleomorphic adenomas was 1.74±0.37×10^-3 mm²/s, for the malignant tumors was 1.04±0.35×10^-3 mm²/s, and for the normal parotid glands was 0.34±0.20×10^-3 mm²/s. The respective ADC value for the single basal cell adenoma was 1.40×10^-3 mm²/s. Statistically significant differences were identified between the subjects with pleomorphic adenoma and those with another type of parotid tumor, and between subjects with healthy parotid glands and those with a tumor.

Conclusion: Calculating the ADC appears to be useful in differentiating pleomorphic adenomas from other types of parotid gland tumors.