This study compares clinical, ultrasound, computed tomography (CT), magnetic resonance imaging (MRI), and pathology findings in 16 prospectively, and seven retrospectively recruited dogs presented for suspected thyroid carcinoma. Of these, 17 were confirmed thyroid carcinoma, while six were initially misdiagnosed. These included four carotid body tumors, one para-esophageal abscess, and one undifferentiated squamous cell carcinoma. Thyroid carcinomas occurred in older dogs without evidence of sex predilection, and were more often unilateral. All were large, heterogeneous, moderately to strongly vascularized, and most commonly contained areas of dystrophic mineralization and/or fluid accumulations. On MRI, thyroid carcinomas appeared hyperintense compared to surrounding musculature in all imaging sequences used, while on CT they had a lower attenuation value than normal thyroid gland tissue. Histologically confirmed tumor capsule disruption with invasion of the surrounding structures was most commonly detected with MRI. Palpation was not an accurate predictor of locally invasive vs. well-encapsulated masses. Computed tomography had the highest specificity (100%) and MRI had the highest sensitivity (93%) in diagnosing thyroid carcinoma, while ultrasound had considerably lower results. We conclude that ultrasound is adequate for use as a screening tool for dogs with suspected thyroid carcinoma, but recommend either CT or MRI for preoperative diagnosis and staging.
Diagnosis and staging of thyroid carcinoma can be challenging in dogs. Often the disease remains clinically undetected until masses are large enough to be palpated and/or compress laryngeal and pharyngeal structures, resulting in a combination of symptoms such as gagging, retching, regurgitation, vomiting, coughing, and change in bark.[1-3] The majority of canine thyroid tumors do not cause changes in thyroid hormone levels.[1-3] Palpation and ultrasonography (US) are commonly used for diagnosis and staging of suspected thyroid carcinoma in dogs and people.[2-7] However, palpation can be an insensitive predictor of tumor invasion and the diagnostic sensitivity of ultrasound can be limited by factors such as poor transmission beyond soft tissue-gas interfaces and resulting inaccessible areas dorsal to the trachea, limited field of view inherent to limited sizes of ultrasound probes, operator dependence, relatively low contrast resolution, and low accuracy in identifying the origin of large neck masses that cause displacement of normal anatomical landmarks.[2, 4, 8-10] These limitations can cause incorrect diagnosis and staging of tumor capsule integrity, degree of invasion of surrounding soft tissues, detection of intrathoracic ectopic tissue, extension of the mass dorsal to the trachea, and regional lymphadenopathy. It is important to correctly identify these characteristics preoperatively, as these are strongly correlated with patient outcome. Computed tomography (CT) has been demonstrated to clearly depict normal canine thyroid tissue, in dogs. No studies describing CT and magnetic resonance imaging (MRI) features of canine thyroid carcinoma could be found in the veterinary literature.
The purposes of this study were to (1) describe CT and MRI characteristics of canine thyroid carcinoma; (2) compare sensitivities of US, MRI, and CT for diagnosing thyroid carcinoma using pathologic diagnosis as the reference standard; and (3) compare imaging findings with clinical palpation findings for regional lymphadenopathy and local invasion.
Materials and Methods
This study conformed to the standards of the Clinical Studies Review Committee at the author's institution (CSRC 036–08), and client consent was obtained for all dogs prospectively enrolled. Dogs included in the first group (group A) were prospectively recruited and were presented for suspected thyroid carcinoma between November 2008 and December 2011. For all included dogs in this group, cervical US was performed by the first author (OT). Cervical MRI and cervical CT studies were acquired within 2 weeks following the initial ultrasound. Cytologic or histologic examination of the thyroid mass and any enlarged lymph nodes was performed within the same 2-week time frame. The same observer (OT) reviewed all CT and MRI examinations. The author was aware of the ultrasound findings at the time of review, but was unaware of the pathologic diagnosis. Dogs included in the second group (group B) were retrospectively recruited from dogs presented between September 2005 and November 2008. All included dogs in this group had cervical US and either CT or MRI examinations, and all had an imaging report diagnosis of suspected thyroid carcinoma. The same observer (OT) reviewed medical records, all available ultrasound recordings, CT and MRI studies, and imaging reports for these dogs.
Recorded clinical information consisted of patient history and physical examination recordings, presenting complaints (coughing, gagging, dysphagia, dysphonia, regurgitation, retching, stridor, or others), laterality of mass on palpation, approximate size on palpation, mobility of the mass, palpation of peripheral lymphadenomegaly, presence of radiographic signs consistent with pulmonary metastatic disease, thyroid function status, and other relevant clinical findings. Clinical mobility of the mass on palpation, and findings from surgery and/or necropsy were recorded for all dogs.
A broadband linear transducer1 with a frequency range of 5–12 MHz was used for US examinations in group A dogs. Dogs were positioned in dorsal recumbency and scanned using a previously described US protocol. The following US characteristics were recorded from gray-scale images for each dog: origin (thyroid vs. ectopic thyroid), side (left, right, bilateral, or midline), size (height, width, length) of the suspected thyroid mass, presence of isthmus, isthmus invasion by the mass, tumor capsule integrity (intact, possibly interrupted, definitively interrupted), invasion of adjacent cervical structures, echotexture (homogeneous vs. heterogeneous, presence of areas of mineralization, and presence of intratumoral fluid accumulation), and echogenicity of the mass relative to surrounding cervical musculature (isoechoic, hypoechoic, or hyperechoic). Degree of vascularization was subjectively rated as mild, moderate, or strong, and based on Color and/or Power Doppler. Regional lymphadenopathy (mandibular, medial retropharyngeal, and deep cervical lymph nodes) was recorded when suspected and based on shape, echogenicity alterations, and/or asymmetry relative to the contralateral node. Visibility of ipsilateral parathyroid glands, presence of invasion by the thyroid mass, and displacement of other anatomical structures in the neck were also recorded. Volume of the mass was calculated using an ellipsoid method (max. length × max. height × max. width × π/6) for each lobar mass. When more than one thyroid lobe was affected, the sum of both lobes and/or ectopic tissue were used.
MRI preceded CT imaging for all group A dogs, in order to avoid artifacts resulting from intravenous contrast injection of iodinated contrast medium. Total anesthesia time for combined MRI and CT imaging was minimized as much as possible so that subsequent surgery could be performed under the same general anesthesia session. All MRI studies were acquired using the same 1.5T scanner,* with the dogs maintained under general anesthesia and positioned in dorsal recumbency. A combination of head and neck array coils were used in all cases. Sequences acquired were: dorsal fast-spin echo (FSE) T2, sagittal FSE T2, transverse FSE T2, and spin echo (SE) T1. A three-dimensional (3D) magnetic resonance angiography (MRA) was acquired at the end of bolus intravenous (IV) contrast injection of gadopentetate dimeglumine2 at a dose of 0.2 ml/kg. This was immediately followed by transverse SE T1 postcontrast images, and a T1-weighted fat-saturated volume interpolated gradient recalled echo sequence (3D VIBE [volume interpolated gradient echo]). Technical parameters for each MRI sequence are described in Table 1. The field of view included the entire neck on both dorsal and sagittal planar images. Transverse slices were limited to the area of the cervical mass. Recorded characteristics from MRI studies were the same as those recorded from US studies and also included suspected tissue of origin (thyroid, ectopic thyroid, or other), and signal characteristics relative to the surrounding musculature. Degree of vascularization was subjectively rated as mild, moderate, or strong, and based on comparing signal intensities of the mass between pre- and postcontrast T1w images.
Table 1. MRI Technique Settings Used for 16 Prospectively Recruited Dogs with Suspected Thyroid Carcinoma
TR, relaxation time; TE, time to echo; FA, flip angle; ST, slice thickness; SI, slice interval.
TRANS T1 (+C)
For CT studies, dogs were scanned under the same general anesthesia session as that used for MRI, and had similar positioning as that used for MRI. A single slice CT scanner3 was used for the first four patients, after which a 16-slice CT scanner4 was used for the remainder of the dogs in group A. The scanning protocol was kept the same after introduction of the new scanner. Images were either acquired5 or reconstructed,6 as contiguous 3-mm transverse slices using both soft tissue and high detail algorithms. The scanned area covered an area from the last molars to the thoracic inlet, and the same scan area was evaluated before and after intravenous injection of iohexol7 at a dose of 600 mg I/kg. Parameters used for CT scans were 110 kV, 150 mA, and 1-s tube rotation time,† or 120 kV, 200 mA, 0.5 s.†† Images were interpreted using a window level and window width of 120 and 220, respectively. Recorded CT characteristics were similar to those recorded for US and MRI studies. In addition, mean CT attenuation values of the mass were recorded from pre- and postcontrast images using a soft tissue algorithm, and avoiding areas of mineralization or suspected hypoperfusion. Vascularization of the mass was subjectively rated as mild, moderate, or strong, and based on attenuation differences of <50 HU, 50–100 HU, and >100 HU, respectively.
A board-certified anatomical pathologist (RP) reviewed all available histology samples. Recorded characteristics from histology samples included: origin of the mass, tumor capsule invasion (defined as presence of tumor cells within the capsule) or disruption (defined as interrupted capsule outline with presence of tumor cells beyond the capsule margins), presence of metastatic disease in submitted lymphoid tissue, and presence of parathyroid gland involvement.
The following characteristics were compared for US, MRI, and CT in group A dogs: origin of the mass, mass size, capsule disruption (defined as interruption of the tumor capsule and focal bulging of the contour of the lobe), local tissue invasion (defined as presence of tissue with imaging characteristics similar to the mass beyond the tumor capsule), lymphadenopathy with or without metastatic disease, and degree of vascularization. Findings from imaging studies were also compared with clinical palpation findings for all dogs. Sensitivity, specificity, positive predictive, and negative predictive values for diagnosing thyroid carcinoma were calculated for each modality, using cytology or histology as the reference standard.
A total of 23 dogs were included in the study, with 16 in group A and seven in group B. Represented breeds were four German Shepherd Dogs, three Labrador Retrievers, three Golden Retrievers, two Beagles, and one each of the following breeds: Whippet, Great Dane, Maltese, English Setter, Terrier, Standard Poodle, Siberian Husky, Pekinese, Brittany Spaniel, and one mongrel. Mean body weight for all dogs was 27 kg, and mean age was 10 years. There were 12 males and 11 females.
Presenting clinical signs, in decreasing order of occurrence, were coughing (8), dyspnea (7), gagging (6), stertor/stridor (5), lethargy (3), dysphonia (3), regurgitation (2), retching (2), vomiting (1), anorexia (1), reverse sneezing (1), sneezing (1). Other clinical findings were peripheral lymphadenomegaly (6) (medial retropharyngeal: 4, mandibular: 1, superficial cervical: 1), laryngeal paralysis (5), other neoplasia (2), fever (1), cyanosis (1), prior thyroid carcinoma resection (1), and anemia as a result of hemorrhage from the cervical mass (1). Pulmonary metastases were identified at time of presentation in one dog.
For the 16 dogs in group A, pathologic diagnoses were based on surgical biopsy and histology in 10 of 16, necropsy and histology in one of 16, and cytology in five of 16. Ten of the 11 dogs evaluated with histology had a diagnosis of thyroid carcinoma and one had a para-esophageal abscess. Three of the five dogs evaluated with cytology had a diagnosis of thyroid carcinoma, one had carotid body tumor, and one had undifferentiated squamous cell carcinoma. For the 10 dogs with a histologic diagnosis of thyroid carcinoma, tumor capsule invasion was detected in all and tumor capsule disruption was diagnosed in six of 10. Metastasis was diagnosed in one of the five histologically evaluated lymph nodes (medial retropharyngeal), and in none of the cytologically evaluated lymph nodes. Parathyroid gland invasion was diagnosed by histology in one case. Eleven dogs in group A had recent thyroid function tests. Results were normal in 10 dogs and abnormal in one dog, a 13-year-old Standard Poodle with hypothyroidism. Prior thyroid function tests in this dog had been normal. This dog had the largest volume of all thyroid masses, and was confirmed to have bilateral thyroid carcinoma (Fig. 1). Another dog with a severely cystic, unilateral, thyroid carcinoma had a small contralateral lobe with decreased CT attenuation, decreased US echogenicity, and heterogeneous US parenchyma. These imaging findings were consistent with previously reported characteristics of hypothyroidism[14-17], but no recent thyroid function tests were available for this dog.
For the seven dogs in group B, one of seven had a nondiagnostic pathologic sample, two of seven had no pathologic sample acquired, and four of seven had an initial histology diagnosis of thyroid carcinoma. The latter four dogs had a CT examination, and the initial diagnosis of thyroid carcinoma was questioned in three of them because of the visibility of two normal thyroid lobes and an unusual rostral location of the mass. Immunohistochemistry was performed for these three dogs, and did not detect thyroglobulin or calcitonin staining. Chromogranin A stained positive and, based on this and the location of the mass, the pathologic diagnosis for these three dogs was changed to carotid body tumor. The one dog with histologically confirmed thyroid carcinoma had both tumor capsule invasion and disruption.
Sensitivity, specificity, positive, and negative predictive values for diagnosing thyroid carcinoma using US, CT, and MRI are summarized in Table 2. Graphical comparisons between US, CT, and MRI characteristics of histologically or cytologically confirmed thyroid carcinoma in group A dogs are displayed in Fig. 2. Thyroid carcinomas appeared as unilateral (60%) or bilateral (40%) space-occupying structures along the dorsolateral aspect of the trachea, just caudal to the larynx, and displaced the common carotid arteries laterally (77%). Mean volume of thyroid carcinomas was 47 cm3 on ultrasound, 65 cm3 on CT, and 55 cm3 on MR. Heterogeneous parenchymal texture was identified in all cases with US and MRI, and often with CT (80%). Areas of mineralization within the mass were identified in 63% and fluid pockets in 77%. The average CT attenuation values were 56 HU precontrast and 132 HU postcontrast images. On MRI, thyroid carcinomas were hyperintense on T2w and 3D VIBE postcontrast sequences in all cases, and hyperintense on T1w in 95%. Capsule disruption was the most variable parameter among the three imaging modalities, and identified on MRI with the highest frequency (Figs. 3 and 4). The frequency of capsule disruption based on MRI was 62% and the frequency based on histology was 67%. One case with a diagnosis of capsule disruption on MRI was not confirmed on histology. Two cases with a diagnosis of capsule disruption on histology were not seen with MRI. For all group A cases with histologically confirmed thyroid carcinoma and capsule disruption, invaded structures were limited to the fascial planes (i.e., cervical connective tissue). Focal tracheal invasion was identified on CT for one histologically confirmed thyroid carcinoma in group B. Parathyroid glands on the side of the cervical mass were not seen with any of the imaging modalities. Contrast-enhanced MRA was available for eight cases of confirmed thyroid carcinoma. These MRA images, acquired during the arterial phase of vascularization, were considered excellent at depicting the deviated common carotid arteries, engorged cranial and caudal thyroid arteries, and vascularization and perfusion of the thyroid masses (Fig. 5). None of the MRA studies demonstrated vascular invasion of the common carotid artery or contrast leakage. Three cases of thyroid carcinoma were located on midline, ventral to the larynx. Based on their midline location, the presence of bilaterally normal-appearing thyroid lobes in two of them, and the absence of metastases elsewhere; these three masses were considered to represent neoplastic transformation of ectopic thyroid tissue. Two of the three masses invaded the hyoid apparatus (Fig. 6A) and had otherwise normal thyroid lobes on CT (Fig. 6B). The third ectopic thyroid mass was small, well defined, noninvasive, and was concurrent with bilateral thyroid masses.
Table 2. Sens, Spec, PPV and NPV for Diagnosing Histologically Or Cytologically Confirmed Thyroid Carcinoma in 13 Dogs of Group A, Using US, CT, and MRI
Ultrasound and CT characteristics of the three histologically confirmed and one cytologically suspected cases of carotid body tumor were similar to characteristics of thyroid carcinomas. These masses were also moderately enhancing and had a heterogeneous parenchyma with occasional dystrophic mineralization and small focal fluid accumulations. The pre- and postcontrast attenuation values and relative echogenicity of the parenchyma were also similar to thyroid carcinomas. All four of the carotid body tumors were located in more rostral positions than thyroid carcinomas, dorsolateral to the larynx. All surrounded and markedly displaced the common carotid artery bifurcation. All were associated with compression and ventral displacement of the larynx, and with narrowing of the laryngeal lumen. All had irregular margins with a multilobulated appearance (Fig. 7). Both thyroid lobes were normal in size, attenuation value, and location in all of these cases. One carotid body tumor invaded the ipsilateral jugular vein and was associated with metastatic mandibular lymphadenopathy. One carotid body tumor invaded local musculature and fascial planes and had histologically confirmed metastases to the medial retropharyngeal lymph nodes. Affected breeds were German shepherd cross (2), Siberian husky (1), and Pekinese (1). Average age for these four dogs was 8 years, and mean body weight was 21 kg. Stertor and/or dyspnea were clinically reported in all of them.
Heterogeneous invasion of the isthmus was seen in a one case of bilateral carcinoma (Fig. 8). On MRI, all affected lobes appeared both T1 and T2 hyperintense compared to the surrounding musculature, at the exception of one lobe that appeared homogeneously T1 hypointense (Fig. 9). Unfortunately, no histopathology was obtained from that lobe. Postcontrast 3D VIBE sequences resulted in marked hyperintensity of the lobes with a marked contrast with the surrounding structures (Fig. 9).
One case of para-esophageal abscess was diagnosed in a 5-year-old Great Dane. This dog presented for coughing, regurgitation, dysphonia, fever, and abnormal breathing sounds. Imaging characteristics were similar to those seen with thyroid carcinoma. The mass was located dorsal and lateral to the esophagus in the mid-cervical area, and normal thyroid lobes could not be identified on US and MRI. Normal thyroid lobes for this case were identified in CT images.
For group A dogs, cervical masses were considered moveable based on clinical palpation in five of 16, and fixed or partially fixed in 11 of 16 dogs. Imaging characteristics of capsule disruption were identified for five of 11 of the masses considered fixed or partially fixed on palpation, and for four of five of the masses considered mobile on palpation.
Confirmed thyroid carcinomas in this sample population of dogs appeared as large, heterogeneous, space-occupying structures dorsolateral to the trachea, and immediately caudal to the larynx. Masses most frequently affected one lobe. The parenchyma was considered moderately to strongly vascularized in most cases. The masses were often multicavitated, and areas of mineralization were seen in more than half the cases. MRA was considered helpful in mapping the displacement of the cervical vasculature and warrants further investigation as a possible preoperative tool. Macroscopic vascular invasion was not assessed or confirmed in this study. Thyroid carcinoma CT attenuation values were much lower than those previously reported for normal thyroid tissue in precontrast images (56 vs. 108 HU), and somewhat lower than those reported for normal thyroid postcontrast images (132 vs. 169 HU).[3, 12, 18, 19] The most likely explanation for this finding is decreased uptake, metabolism, and storage of iodine in neoplastic vs. normal thyroid tissue.[20, 21] With appropriate CT windowing, absence of visualization of one or both normal hyperattenuating thyroid lobes therefore warrants a high diagnostic suspicion for a thyroid gland neoplasm.
Average thyroid mass sizes in our study were approximately 50 times the size of reported normal canine thyroid gland sizes, and measured smaller with US vs. CT and MR. However, a lack of confirmed thyroid mass volumes postsurgery precluded assessment of tumor volume accuracies for the three imaging modalities.
Signalment and presenting signs for dogs with confirmed thyroid carcinoma in this study were consistent with previous reports. Most dogs were euthyroid, while hypothyroidism was documented using blood chemistry in one dog that had a large bilateral thyroid carcinoma. In this dog, hypothyroidism was most likely due to destruction of normal thyroid tissue. Another dog, with a unilateral cystic mass had a small irregular contralateral gland on US. This was considered to be suggestive for hypothyroidism but no thyroid test results were available for confirmation. Imaging characteristics of this mass were similar to those previously described for cats with hyperthyroidism due to cystic thyroid masses. We suspected that this dog's thyroid carcinoma may have been secreting excessive levels of thyroid hormones and may have therefore suppressed the contralateral thyroid lobe.
In our study, thyroid carcinomas and carotid body tumors had similar parenchymal imaging characteristics and histologic features. Histologic differentiation in four of our cases required use of special staining. Distinctive imaging characteristics of the confirmed carotid body tumors in our study were a dorsolateral location to the larynx at the level of the bifurcation of the common carotid artery, compression and narrowing of the lumen of the larynx, and evidence of local tissue invasion and metastatic disease to regional lymph nodes in half of them. Clinical signs of upper airway obstruction were also more frequent in dogs with this tumor type. Carotid body tumor CT, MRI, and US characteristics in our dogs were similar to those describe in people[26-29], however evidence of metastatic disease was higher.[5, 12, 30-33] The three ectopic thyroid carcinomas in our study were located in midline, ventral to the hyoid apparatus, and invaded the hyoid bones in two dogs. These were more smoothly outlined than the carotid body tumors.
Ultrasound, CT, and MRI findings were similar for location of thyroid carcinomas in all our cases. Both US and MRI had a false-positive diagnosis of thyroid carcinoma in one case with para-esophageal abscess, while CT had no false positive diagnoses. Signal characteristics, location of the mass, and nonvisualization of normal thyroid tissue contributed to the MRI misdiagnosis. The normal thyroid gland in this case was markedly displaced and compressed, resulting in silhouetting with surrounding soft tissues. In precontrast CT images for this case, we noticed markedly hyperattenuating normal thyroid tissue adjacent to the mass. These findings justify the use of CT for cases where the origin of a large cervical mass is uncertain. The high iodine content of normal thyroid tissue and resulting high CT attenuation would be expected to improve detection of normal thyroid lobes even when they are displaced or compressed by an adjacent mass. However, correct diagnosis of tumor origin may be difficult when there is a neoplastic mass originating from ectopic thyroid tissue and otherwise normal thyroid lobes bilaterally. In this situation, identifying the mass in a region consistent with ectopic thyroid tissue (i.e., ventral to the larynx) may be helpful.
In our study, US had the lowest sensitivity and specificity for correctly identifying thyroid carcinoma. Also, US predictions of tumor capsule integrity and size were lower than those obtained using CT and MRI. Findings indicated that US may be useful as a screening tool, but may not be sufficiently accurate for preoperative diagnosis and staging of canine thyroid carcinomas. Magnetic resonance imaging yielded the lowest number of false-negative diagnoses for thyroid carcinoma and a frequency value that most closely matched histology for local tissue invasion. These findings justify the use of MRI as a preferred modality for preoperative diagnosis and staging of clinically suspected thyroid carcinomas.
Clinical palpation was a poor predictor for imaging characteristics of tumor invasion in our study. Imaging characteristics of capsule disruption were identified for almost half of the masses considered fixed or partially fixed on palpation, and for nearly all of the masses considered mobile on palpation. These findings indicate that palpation alone may not be sufficiently accurate for assessing tumor invasion in dogs with suspected thyroid carcinoma.
One limitation of our study was a possible sample population bias. Dogs were included only if they had suspected thyroid carcinoma based on clinical and US characteristics. It is possible that the level of suspicion may have changed between retrospective and prospective studies and may have also changed during the course of our prospective study. Another limitation of our study was a lack of statistical comparisons for diagnostic sensitivity values and frequencies of imaging characteristics among the three modalities. The decision not to perform statistical comparisons was based on the small sample population size, high variability between samples, and concerns that cytology and routine histology may not be reliable reference standards. Cytology findings rely on aspirate locations. Histology findings rely on the locations of the cuts obtained from the mass and may not discriminate thyroid carcinomas from carotid body tumors if special stains are not used.
Another possible limitation of this study was the assumption that thyroid tissue masses located ventral to the larynx were due to neoplastic transformation of ectopic thyroid tissue rather than local metastasis. As both would be expected to look similar on histology, we based our assumption on the fact that there was no primary neoplasia at the location of the normal thyroid in two of three dogs, that a location ventral to the larynx has not been previously reported for local metastasis of thyroid carcinomas in dogs, that no metastases were found in the subcutaneous tissues of affected dogs, and that no pulmonary metastases were found. We also based our assumption on the fact that these ventrally located thyroid tissue masses were found along the path of the thyroglossal duct, and that this is a commonly reported site of ectopic thyroid tissue in people.[5, 30-33]
In conclusion, findings from our study support the use of US as an initial screening tool for evaluating suspected thyroid carcinomas in awake dogs. However, we recommend the additional use of either MRI or CT for preoperative diagnosis and staging purposes. All three imaging methods are likely to be more reliable than clinical palpation for assessment of local tissue invasion.
We would like to thank Dr. S. Rowell and Dr. M. Bucknoff for their help with data collection, and Linda Kinney, Brenda Tilley, and Joanne Morin for their technical assistance. We also wish to thank all small animal clinicians of the Foster Hospital for Small Animals, in particular Dr. S. Mitchell (in memoriam), to help recruiting patients for this study.
Philips iU22 Ultrasound system, Philips Medical Systems, Eindhoven, The Netherlands.