Clinical Imaging
Volume 36, Issue 1 , Pages 1-7, January 2012

Role of apparent diffusion coefficient values and diffusion-weighted magnetic resonance imaging in differentiation between benign and malignant thyroid nodules

Received 5 January 2011; accepted 30 March 2011. published online 18 July 2011.

Article Outline

Abstract 

Objective

The purpose of the study was to differentiate between benign and malignant thyroid nodules using nodule-spinal cord signal intensity and nodule apparent diffusion coefficient (ADC) ratios on diffusion-weighted magnetic resonance imaging (DW-MRI).

Materials and Methods

Forty-four patients (27 females, 17 males; mean age, 49 years) with nodules who underwent DW-MRI were included in this study. The images were acquired with 0, 50, 400 and 1000 s/mm2 b values. ADC maps were calculated afterwards. Fine needle aspiration biopsies (FNAB) were performed at the same day with DW-MRI acquisition. The diagnosis in patients where malignity was detected after FNAB was confirmed by histopathologic analysis of the operation material. The signal intensities of the spinal cord and the nodule were measured additionally, over b-1000 diffusion-weighted images. Nodule/cord signal intensity (SI) ratios were obtained and the digital values were calculated by dividing to ADC values estimated for each nodule. Statistical analysis was performed.

Results

The (nodule SI-cord SI)/nodule ADC ratio is calculated in the DW images, and a statistically significant relationship was found between this ratio and the histopathology of the nodules (P<.001). The ratio was determined as 0.27 in benign and 0.86 in malignant lesions. The result of receiver operating characteristic (ROC) analysis was statistically significant, and the area under curve (100%) was considerably high. The threshold value was calculated as 0.56 according to the ROC analysis. According to this threshold value, the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy rates for (nodule SI/cord SI)/ADC ratios in differentiating benign from malignant thyroid nodules are calculated as 100%, 97%, 83%, 100%, and 98%, respectively.

Conclusion

We have found that (nodule/cord SI)/nodule ADC ratio has the highest values for sensitivity and specificity among the tests defined for characterization of nodules.

Keywords: Thyroid nodule, Diffusion weighted magnetic resonance imaging, ADC ratio

 

Back to Article Outline

1. Introduction 

Palpable nodules in the thyroid gland are detected in approximately 4–7% of the adult population, and they occur more frequently in females than males [1], [2]. Thyroid nodules may be incidentally spotted in 14.5% of patients during neck imaging performed for other reasons. Most of these nodules are benign. Nodular thyroid disease is highly prevalent, whereas cancer of the thyroid is rare and they consist 1% of all malignant cancers [3]. The probability of cancer in a nodule is influenced by various risk factors. Malignancy prevalence is higher in patients aged over 60 or below 20 years. Cancer history in the family and neck radiation history increase the risk of malignancy in the thyroid nodules. Radiation exposure is present in the history in 20–30% of patients with a palpable thyroid disease [4].

Ultrasonography, an inexpensive and noninvasive method, is the most widely used modality in defining the lesions. A nodule is detected by sonographic examination in 10–40% of the population [5]. The success of sonography in differentiating benign from malignant nodules of the thyroid is evaluated in many studies. Although none of the sonographic properties have a reliable and sufficient sensitivity, specificity, and positive predictive value (PPV) in defining malignancy of the nodules, the sonographic characteristics of the nodules may help in selecting the nodules for needle aspiration. Thyroid scintigraphy can be used in patients with thyroid nodules to determine the presence of functional-hot or nonfunctional-cold lesions. Solitary cold nodules carry a malignancy risk ranging between 15% and 25%, whereas a cold nodule in a multinodular gland carries a 1% risk of malignancy [6].

Ultrasound-guided fine-needle aspiration biopsy (FNAB) has been widely well-accepted as an accurate diagnostic method for evaluation of thyroid nodules. Cancer incidence in individuals with a thyroid nodule has been determined as 9.2-13% with FNAB [7]. FNAB is an efficient method in differentiating malignant from benign nodules [8].

The movement of water molecules in the biological tissues due to heat effect is called diffusion and Brownian motion. Diffusion weighted magnetic resonance imaging (DW-MRI) is a noninvasive technique used for measuring the diffusion of water molecules in the tissues. Routine T1- and T2-weighted images have a limited role in the evaluation of thyroid nodules [8]. Studies performed using apparent diffusion coefficient (ADC) with DW-MRI for differentiating benign from malignant thyroid nodules are reported in the literature. In a study performed by Razek et al., the sensitivity, specificity, and accuracy of ADC values in differentiating benign from malignant nodules are stated as 97.5%, 91.7% and 98.9%, respectively, according to the threshold value [9].

Our aim in this study was to assess the role of ADC values and signal intensities of the nodule and the spinal cord in differentiating benign from malignant thyroid nodules.

Back to Article Outline

2. Materials and methods 

This is a prospective study performed on 54 nodules in 44 patients who have applied for ultrasonography (US) between June 2007 and Jan 2010. While 27 of the patients were female and 17 were male, the age range was between 33 and 75 years (mean, 49 years). Written consent was obtained from all of the patients before initiation of the study.

MRI studies were performed on 1.5-T MR scanners (Avanto; Siemens, Erlangen, Germany), using superficial neck coil. A neck coil was placed in each patient, putting the thyroid gland at the central area of field of view (FOV). Conventional T1-weighted (TR/TE, 363/9.2 ms) and T2-weighted (TR/TE, 6180/64 ms) axial- and sagittal-view images are acquired with 5-mm slice thickness, 1-mm intersectional gap, 20–25 cm FOV and matrix 512×512. Diffusion-weighted MR images are acquired afterwards, using multislice single-shot echoplanar sequence. Multiple slices are acquired from the neck, including thyroid gland. Image parameters were: TR=6200 ms, TE=86 ms, NEX=1, band width 125 kHz, 192×192 matrix, 20–25 cm FOV, 5-mm slice thickness, and 1-mm intersectional gap. Diffusion gradients are applied in three ortogonal planes (X, Y and Z). Fat suppression is performed with STIR sequence to avoid chemical shift artifacts. The images were acquired with 0, 50, 400 and 1000 s/mm2 b values. ADC maps are calculated afterwards. Diffusion-weighted imaging took less than 2 min (Fig. 1).

  • View full-size image.
  • Fig. 1. 

    (A) A 35-year-old woman with multi nodular goiter. Calculation of ADC from thyroid nodules on ADC maps. (B) Calculation of signal intensity on DW images. ADC map (A) and DW image (B) shows calculation of ADC value and signal intensity of thyroid nodules in 35-year-old woman with multinodular goiter, respectively.

The FNABs in the patients were performed at the same day with DW-MRI acquisition. The nodule was divided into four quadrants so that the samples should be taken from the solid area during FNAB, and aspiration was performed from each quadrant using a 20–25-G needle. The biopsy was repeated in patients where the initial biopsy was reported as non-diagnostic. Any complication such as hemorrhage was not observed during biopsies. The diagnosis in patients where malignancy was detected after FNAB was confirmed by histopathologic analysis of the operation material. Comparison with only the FNAB result was performed in a patient with advanced medullary cancer, since an operation could not be performed.

2.1. Interpretation of the images 

DW-MRI images are transferred to DICOM viewer Intage Realia for data processing. Round regions of interest (ROI) were determined in each anatomical area, to measure the ADC values of the nodules. Streak and motion artifacts were cautiously excluded during determination of these regions. ADC values were obtained from the normal parenchyma as well, in addition to the nodule. Three ROI measurements were performed for each ADC value and the mean for the closest two measurements was calculated. Measurement of ADC values from cystic areas was exclusively avoided in semisolid nodules. The signal intensities (SI) of the spinal cord and the nodule were measured additionally over b-1000 DW images. Nodule SI/cord SI ratios were obtained and the digital values were calculated by dividing to ADC values estimated for each nodule. Three nodules containing substantial hyperintense areas corresponding to hemorrhage in T1-weighted sequences in MRI were excluded from data analyses to avoid misinterpretation of ADC values.

2.2. Statistical analyses 

All statistical analyses are performed with SPSS (Statistical Package for the Social Sciences) for Windows ver. 15.0 software. One-sample Kolmogorov-Smirnov test and one-way analysis of variance were performed for data analyses in our study. The parametric independent-samples t test was applied for comparing the histopathology of thyroid nodules and ADC, nodule SI/cord SI and nodule ADC/thyroid parenchyma ADC values, since the data distribution was normal and homogeneous. On the other hand, the relationships between thyroid parenchyma ADC and nodule-cord SI/nodule ADC ratio and nodule histopathologies were tested by non-parametric Mann–Whitney U test, since the data distribution was abnormal and non-homogeneous. The threshold values for differentiating malignant from benign thyroid nodules were obtained using ROC curve. Statistical significance level was set at P<.05.

Back to Article Outline

3. Results 

The image quality level in our study was acceptable. A total of 51 nodules were analyzed statistically. Three nodules containing substantial hyperintense signal character corresponding to hemorrhage in T1-weighted sequences in MRI were excluded from data analyses, since they could lead to ADC values lower than expected; however, their internal ADC and SI values were also calculated. The diameter of the nodules varied between 0.9 and 6 cm (mean 2 cm). Forty-six (90.2%) of the nodules were benign, and 5 (9.8%) were malignant lesions. Histopathologic typing was not done in the benign nodules. The malignant lesions were confirmed postsurgically by histopathology. Due to the advanced stage, the diagnosis was made by FNAB alone in one case with cancer (Fig. 2, Fig. 3).

  • View full-size image.
  • Fig. 2. 

    (A) A 44-year-old woman with goiter. Hyperintense signal in nodular lesion on isthmus of thyroid gland on DW images. (B) Diffusion restriction is seen on ADC maps with an ADC value of 0.49×10–3 mm2/s. Nodule was diagnosed malignant by biopsy and surgery.

  • View full-size image.
  • Fig. 3. 

    (A) A 50-year-old female patient with complaints of hoarseness and having enlarged thyroid glands, rightward trachea deviation, obstruction on left jugular vein on T1 weighted images. (B) A heterogeneous mass lesion having lobular contour on coronal T2 images causes tracheal pressure. (C) Mass lesion is hyperintense on DW images. (D) Mean ADC values are measured 1.1×10–3 mm2/s on ADC map. FNAB of the nodule was diagnosed as medullary thyroid cancer.

A statistically significant difference was found between the ADC values of malignant and benign lesions (P<.001). The mean ADC values detected were 1.6±0.1×10–3 mm2/s in the benign and 0.8±0.2×10–3 mm2/s in the malignant lesions. The result of ROC analysis was statistically significant. The threshold value found was 1.0×10–3 mm2/s by ROC analysis. According to this threshold value, the sensitivity, specificity, PPV, negative predictive value (NPV) and accuracy rates for ADC values in differentiating benign from malignant thyroid nodules are calculated as 80%, 97%, 80%, 97% and 96%, respectively.

Normal thyroid gland parenchyma ADC value ranges between 0.7 and 1.45 (mean 0.98). There is not a statistically significant relationship between ADC values of malignant and benign nodules and normal parenchyma (P=.474).

A statistically significant difference was found between the nodule ADC/thyroid parenchyma ADC ratios for malignant and benign thyroid nodules (P=.004). This ratio was 1.67±0.14 in benign and 0.98±0.26 in malignant lesions. The result of ROC analysis was statistically significant. The threshold value was determined as 1.11 in ROC analysis. According to this threshold value, the sensitivity, specificity, PPV, NPV and accuracy rates for nodule ADC/thyroid parenchyma ADC ratios in differentiating benign from malignant thyroid nodules are calculated as 80%, 95%, 66%, 97% and 94%, respectively.

The nodule SI/cord SI ratios were calculated in the DW images and a statistically significant relationship was detected between this ratio and the histopathology of the nodules (P<.001). This ratio was 0.4 in benign and 0.7 in malignant lesions. The result of ROC analysis was statistically significant, and the area under ROC curve (96%) was considerably high (P<.001). The threshold value was calculated as 0.56 according to the ROC analysis. According to this threshold value, the sensitivity, specificity, PPV, NPV, and accuracy rates for nodule SI/cord SI ratio in differentiating benign from malignant thyroid nodules are calculated as 100%, 86%, 45%, 100%, and 88%, respectively.

The nodule SI-cord SI/nodule ADC ratio is calculated in the DW images and a statistically significant relationship was found between this ratio and the histopathology of the nodules (P<.001). The ratio was determined as 0.27 in benign, and 0.86 in malignant lesions. The result of ROC analysis was statistically significant, and the area under ROC curve (100%) was considerably high. The threshold value was calculated as 0.56 according to the ROC analysis. According to this threshold value, the sensitivity, specificity, PPV, NPV and accuracy rates for nodule SI-cord SI/ADC ratios in differentiating benign from malignant thyroid nodules are calculated as 100%, 97%, 83%, 100%, and 98%, respectively.

Back to Article Outline

4. Discussion 

Palpable thyroid nodules are detected in 4–7% of the adult population, and they occur more frequently in females than males [1], [2]. Thyroid nodules can be determined incidentally in 14.5% of patients undergoing neck imaging for other reasons and up to 38–50% in autopsy series [10]. Of the nodules in the thyroid gland detected at the US evaluation, two thirds are multiple and one third are solitary nodules. Around 85–90% of these nodules are benign [9], [11]. Although 80% of the benign nodules are hyperplastic nodules, the remaining 20% consist of adenomas, cysts and focal thyroiditis. Among the 51 nodules in our study, 46 (90.2%) were benign and 5 (9.8%) were malignant lesions, with a similar rate reported in the literature. While papillary cancer comprises 75–90% of the thyroid cancers, the other types are follicular, medullary, anaplastic, metastasis, and lymphoma. Of the six patients with cancer in our study, five had papillary and the other one had medullary cancer [12].

When thyroid nodules are detected by imaging methods, we need to differentiate benign from malignant lesions. Early diagnosis of thyroid cancers decreases morbidity and mortality, and detecting the benign nodules helps avoid unnecessary operations. The sensitivity, specificity, and accuracy of US in differentiating benign from malignant nodules range between 63–94%, 61–95% and 80–90%, respectively, and its reliability is unknown [13]. US has an important place, especially in guidance for FNAB and selecting the nodule for biopsy.

FNAB is the principle and mostly used method in differentiating benign from malignant nodules and indicating for surgery. Its sensitivity and specificity are found to be approximately 85% and 99%, respectively, in large and experienced centers. The experience of persons performing the biopsy and cytopathologic evaluation has vast importance. Satisfactory aspiration up to 97% can be achieved with FNAB performed by experienced individuals [14]. Around 18% of patients with thyroid nodule(s) will undergo an operation due to positive, suspected, and nondiagnostic FNAB and, while most of these patients have benign lesions, the malignant lesions comprise approximately 15–32%. In other words, most of the patients who undergo an operation have benign nodules [15].

Repeat biopsies are performed in case of a suspected or nondiagnostic FNAB result. Since it is an invasive procedure, it is undesirable for the patients. Additionally, it has some complications. The most common complication consists of venous hemorrhages. Performing the biopsy from an area where there are a lot of cancer cells should increase the reliability of FNAB. Therefore, we need a noninvasive imaging method which can decrease the repeat biopsies and help US in selection of the nodule for biopsy and spotting the exact place for FNAB. DW-MRI helps US in these aspects and, additionally, provides information not only anatomically but also physiologically. While anatomical structures at the thorax entrance can not be assessed clearly in the US examination, pathological thyroid tissue and lymph nodes in these regions can be evaluated with DW-MRI.

Randomized movement of water molecules is used in DW-MRI. The completely randomized movement of water molecules in an unrestricted area is called free diffusion or Brownian motion [16]. Movement of water molecules in the tissues is not completely randomized; more precisely, movement of water is restricted due to tissue compartments, cell membranes and intracellular organelles. An increase in the number of cells in the tissue in the intact cell membranes and nucleus/cytoplasm ratio within the cell will lead to restriction of movement of water molecules in the intracellular and extracellular compartments [17]. Diffusion restriction will occur in DW-MRI of the malignant thyroid nodules due to these reasons. The quantitative values for the diffusion are obtained over ADC maps and their relationships with the histopathologies are assessed in our study. Additionally, the relationships between nodule SI/cord SI, nodule-cord SI/nodule ADC values, normal thyroid parenchymal ADC and nodule ADC/parenchymal ADC are also assessed in the DW-MRI examination.

A statistically significant relationship was found between nodule histopathology and nodule ADC values (P<.001). Mean ADC values are determined as 1.6±0.1×10–3 mm2/s for benign and 0.8±0.2×10–3 mm2/s for malignant lesions. According to these results, it is observed that the ADC values in benign nodules are higher, when compared with those in malignant nodules. The result of ROC analysis was statistically significant. The threshold value is calculated as 1.0×10–3 mm2/s according to the ROC analysis. According to this threshold value, the sensitivity, specificity, PPV, NPV and accuracy rates for ADC values in differentiating benign from malignant thyroid nodules are calculated as 80%, 97%, 80%, 97%, and 96%, respectively. In a study performed by Razek et al., the sensitivity, specificity, and accuracy of ADC values in differentiating benign from malignant nodules are stated as 97.5%, 91.7%, and 98.9%, respectively, when compared with the threshold value [9]. When we compared our results with the results obtained by Razek et al. [9], we saw that the sensitivity for ADC values found in our study was lower; however, the other values were comparable.

Normal parenchymal ADC value of the thyroid gland ranges between 0.7 and 1.45 (mean 0.98). No statistically significant relationship was found between the ADC values of malignant and benign nodules and parenchyma (P=.474). The nodule should be localized exactly by assessing the T1- and T2-weighted images and DW-MRI together on the ADC maps to avoid any error, since the values obtained from normal thyroid tissue may be even lower than the threshold value. Since the selectivity of nodules smaller than 8 mm especially decrease and may lead to false ADC measurements, studies should be performed on larger nodules [9]. However, studies on smaller nodules may be performed in the future with novel sequences using systems that have a higher power.

A statistically significant difference was found between the nodule ADC/parenchymal ADC ratios of malignant and benign thyroid nodules (P=.004). This ratio was 1.67±0.14 in benign and 0.98±0.26 in malignant lesions. While this ratio is high in benign lesions, it is low in malignant lesions. The result of ROC analysis was statistically significant. The threshold value was determined as 1.11 according to the ROC analysis. The sensitivity, specificity, PPV, NPV, and accuracy rates for nodule ADC/parenchymal ADC ratio in differentiating benign from malignant thyroid nodules according to the threshold value are calculated as 80%, 95%, 66%, 97%, and 94%, respectively. The sensitivity and PPV values of this test were found to be less than those of nodule ADC test.

The nodule SI/cord SI ratio is calculated in DW images and a statistically significant relationship is detected between the histopathological features of the nodules (P=.000). This ratio was 0.4 in benign and 0.7 in malignant lesions. The ratio was found to be low in benign and high in malignant nodules. The result of ROC analysis was statistically significant and the area under ROC curve (96%) was considerably high, which shows the power of the test (P<.001). The threshold value according to the ROC analysis was calculated as 0.56. The sensitivity, specificity, PPV, NPV, and accuracy rates for nodule SI/cord SI ratio in differentiating benign from malignant thyroid nodules according to the threshold value are calculated as 100%, 86%, 45%, 100%, and 88%, respectively. According to these values, the sensitivity and NPV of this ratio are considerably higher than those with ADC test measurement, while its specificity and NPV are lower. Thus, this is rather a differentiating test but not a diagnostic test and has been defined for the first time in literature for thyroid nodules. The signal intensities of the nodule in DW images are used in this test. While the SI increases significantly in malignant nodules, it decreases in benign nodules.

The nodule SI-cord SI/nodule ADC ratio is calculated in the DW images and a statistically significant relationship was found with the nodule histopathology (P<.001). This ratio was calculated as 0.27 in benign, and 0.86 in malignant lesions. While this value is low in benign nodules, it was high in malignant nodules. The result of ROC analysis was statistically significant and the area under ROC curve (100%) was considerably high. The threshold value according to the ROC analysis was calculated as 0.56. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy rates for nodule SI-cord SI/ADC ratio in differentiating benign from malignant thyroid nodules according to the threshold value are calculated as 100%, 97%, 83%, 100%, and 98%, respectively. These sensitivity and specificity values are considerably high for a diagnostic test. However, the low PPV can be explained especially by the small number of patients with malignant lesions. PPV can be increased with wider series. The odds ratio for a positive test result (L+: the accuracy rate for diagnosis of disease) is found to be considerably high for this test, while the odds ratio for a negative test result (L–: the accuracy rate for diagnosis of health) is found to be significantly low. According to the results, this test is significantly capable of differentiating actual patients from actual healthy subjects. This test is defined for the first time in the literature.

Diffusion Tensor Imaging was performed and fractional anisotropy (FA) values were calculated in five nodules. The histopathological findings were benign in four patients and mean FA value was 192×10–3 mm2/s. However, the FA value in the other patient with a papillary cancer (85×10–3 mm2/s) was considerably lower than the results in benign nodules. Statistical analyses were not performed on the case defined, due to obvious areas with hyperintense signal in the T1-weighted images. These areas with increased signal were thought to be due to the apparent proteineous cystic ingredient of the tumor. On the other hand, the nodule ADC values are significantly high (ADC=2.4) and suggest that tumor does not contain marked hemorrhage. Hemorrhagic nodules may lead to low ADC values and give the impression of a malignant nodule. Hemorrhagic nodules that lead to errors in the images were detected and confirmed by repeat biopsies in two cases and they were excluded from the statistical analyses. The other tests are performed on the malignant nodule excluded from the statistical analyses, but they were not found to be concordant with malignancy.

There are some limitations of our study. The first limitation is the small number of malignant nodules. More reliable results could be obtained by increasing this number. As the second limitation, the smallest nodule in this study has a 9-mm diameter. Smaller nodules may not be localized exactly with DW-MRI and this may lead to errors.

As a conclusion, we have discussed the value of DW-MRI, a noninvasive modality, in differentiation of malignant from benign thyroid nodules. We have found that nodule-cord SI/nodule ADC ratio has the highest values for sensitivity and specificity among the tests defined for characterization of nodules. Novel studies with large series are warranted to ensure routine usage of DW-MRI in thyroid nodules.

Back to Article Outline

References 

  1. Rojeski MT, Gharib H. Nodular thyroid disease: evaluation and management. N Engl J Med. 1985;313:428–436
  2. Van Herle AJ, Rich P, Ljung BME. The thyroid nodule. Ann Intern Med. 1992;96:221–232
  3. Grebe SKG, Hay ID. Follicular cell-derived thyroid carcinoma. Cancer Treat Res. 1997;89:91–140
  4. DeGroot LJ, Reilly M, Pinnameneni K, Refetoff S. Retrospective and prospective study of radiation induced thyroid disease. Am J Med. 1983;74:852–862
  5. Tan G, Gharib H, Reading C. Solitary thyroid nodule: comparison between palpation and ultrasonography. Arch Intern Med. 1995;155:2418–2423
  6. Brown CL. Pathology of the cold nodule. Clin Endocrinol Metab. 1981;10:235–245
  7. Frates MC, Benson CB, Charboneau JW, Cibas ES, Clark OH, Coleman BG, et al. Management of thyroid nodules detected at US: society of radiologists in ultrasound consensus conference statement. Radiology. 2005;237:794–800
  8. Okamoto T, Yamashita T, Harasawa A. Test performance of three diagnostic procedures in evaluating thyroid nodules: physical examination, ultrasonography and fine needle aspiration cytology. Endocr J. 1994;41:243–247
  9. Razek A, Sadek A, Kombar OR. Role of apparent diffusion coefficient values in differentiation between malignant and benign solitary thyroid nodules. Am J Neuroradiol. 2008;29:563–568
  10. Ramacciotti CE, Pretorius HT, Chu EW, Diagno Barsky SH, Brennan MF, Robbins J. Diagnostic accuracy and use of aspiration biopsy in the management of thyroid nodules. Arch Intern Med. 1984;144(6):1169–1173
  11. Solbiati L, Osti V, Cova L, Cova L, Tonolini M. Ultrasound of thyroid, parathyroid glands and neck lymph nodes. Eur Radiol. 2001;11:2411–2424[Epub 2001 Oct 25]
  12. Schlumberger MJ, Filetti S, Hay ID. Nontoxic goiter and thyroid neoplasia. In:  Larsen PR,  Kronenberg HM,  Melmed S, et al. editor. Williams textbook of endocrinology. 10th ed.. Philadelphia: WB Saunders; 2003;p. 457–490
  13. Solbiati L, Livraghi T, Ballarati E, Ierace T, Crespi L. Thyroid gland. In:  Solbiati L,  Rizzatto G editor. Ultrasound of superficial structures. Edinburgh: Churchill Livingstone; 1995;p. 49–85
  14. Rich P. The thyroid nodule. Ann Intern Med. 1982;6:221–232
  15. Reading C, Charboneau JW, Hay I, Sebo TJ. Sonography of thyroid nodules: a “classic pattern” diagnostic approach. Ultrasound Q. 2005;21:157–165
  16. Koh DM, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR. 2007;188(6):1622–1635
  17. Qayyum A. Diffusion-weighted Imaging in the abdomen and pelvis: concepts and applications. RadioGraphics. 2009;29:1797–1810

PII: S0899-7071(11)00080-5

doi:10.1016/j.clinimag.2011.04.001

Clinical Imaging
Volume 36, Issue 1 , Pages 1-7, January 2012

Article Tools

* Image Usage & Resolution