Breast ultrasonography has evolved considerably over the last few

Transcription

1 Diagn Interv Radiol 2009; 15: Turkish Soiety of Radiology 2009 BREAST IMAGING ORIGINAL ARTICLE New elastographi lassifiation of reast lesions during and after ompression Eduardo de Faria Castro Fleury, Jose Carlos Vendramini Fleury, Seastiao Piato, Deio Roveda Jr. PURPOSE Proposal for the lassifiation of reast masses through ultrasound elastography in order to differentiate enign and malignant lesions with histologial orrelation. MATERIALS AND METHODS 188 patients enrolled for perutaneous iopsy of 228 reast lesions. Elastography was performed and interpreted aording to riteria reated y the authors, with sores varying from 1 to 4 ased on elastiity of images otained upon release of ompression. These results were ompared with the histologial results; elastiity sores of 1 and 2 were onsidered enign, a sore of 3 as proaly enign, and 4 as suspiious for malignany. Positive preditive value, speifiity, and diagnosti auray have een alulated. The results were evaluated using Fisher s exat test and the analysis of the reeiver operating harateristi (ROC) urve to determine the assoiation with the histologial results, and diagnosti auray of the proposed lassifiation. RESULTS The positive preditive value, speifiity, and diagnosti auray of the sores were 76.5%, 95.9%, and 94.7%, respetively. Of 228 lesions tested, 26 tests yielded true positive results; 8 yielded false positive results; 190 true negative results; and 4 false negative results. There was assoiation with the histologial results y the Fisher method (P < 0.05) and an exellent area elow the ROC urve of (onfidene range of 95%, ). CONCLUSION The lassifiation y elastography proposed y the authors an e used as an important tool omined with ultrasonographi studies for differentiating enign and malignant lesions of the reast. Key words: reast ultrasonography US strain imaging elastography From the Department of Radiology (E.F.C.F. edufleury@ hotmail.om, S.P., D.R.J.), Santa Casa de São Paulo, Brazil; and the Department of Radiology (J.C.V.F.), Centro de Tomografia Computadorizada, São Paulo, Brazil. Reeived 18 Septemer 2008; revision requested 17 Otoer 2008; revision reeived 26 Otoer 2008; aepted 14 Novemer Breast ultrasonography has evolved onsideraly over the last few deades. From the limited diagnosti method that allowed only the disrimination of solid and ysti lesions, it has eome a sophistiated tool for evaluating and lassifying masses and ysts (1, 2). Ultrasonographi riteria that help differentiate enign from malignant lesions inlude shape, margin, orientation, eho pattern, posterior aousti features, lesion oundary, and vasularization, as proposed in the latest edition of Breast Imaging Reporting and Data System (BI-RADS TM ) (3). In this ontext, new researh is eing onduted on methods that may improve the diagnosti speifiity and auray of ultrasonography without ompromising its sensitivity. In 1991, Ophir et al. (4) introdued a new tehnique alled elastography, whih allowed the evaluation of soft tissue y elasti deformation, through the appliation of ompression to the area of interest, earing in mind that the mehanial properties of the tissue are important indiators for the diagnosis of malignant lesions, given that most reast arinomas are harder than the adjaent tissue (5). Currently there are two main lines of researh to determine the linial appliaility of ultrasound elastography. The first line of researh is ased on the assessment of the mass size efore and after ompressing the target areas. This assessment uses software with whih soft lesions appear lighter and firm lesions appear darker, and malignant lesions tend to e more evident than enign lesions (6 8). The other line of researh is ased on the use of a software that applies a different olor spetrum to tissues aording to their hardness, ranging from red to soft tissues, green to intermediate tissues and dark lue to hard tissues (9 11). There is no onsensus as to whih is the est tehnique or lassifiation defining its linial appliation. The main limitations are the interoserver variaility desried in previous studies and the lak of a standard for how ompression should e performed during the study. This study proposes an elastographi lassifiation system using not only the image of the lesion efore and during ompression, ut also after parenhymal deompression and its orrelation with the respetive histologial result. The introdution of the time point (deompression) into the study shows that the patient s reast tissue type has less influene on the results, allowing evaluation riteria to e adopted with sores ranging from 1 to 4, ased on the olor variation oserved in the areas of interest. Materials and methods Patients This is a prospetive trial. Images from 207 patients, with a mean age of 44 years (range, 17 83), were evaluated. They presented with

2 Figure 1. Classifiation of reast masses using sores and the assessment of olor variation during ompression and after deompression of the reast tissue. Sore 1, lesions that presented the same spetrum of olors as the adjaent reast tissue; Sore 2, lesions whih, after deompression, had soft tissue olor variations overing over 50% of the lesion; Sore 3, lesions whih, after deompression, presented a olor variation on less than 50% of the area (etween 10% and 50%), usually around the margins, varying etween yellow and green on the olor sale; Sore 4, lesions with no signifiant olor variation during ompression and deompression. lesions during the ultrasonographi study, and were referred for perutaneous reast iopsy during the period from May 1 st to Otoer 30 th The mean diameter of the lesions was 1.5 m (median of 1.3 m, ranging from 0.5 m to 3.2 m). The first 10 patients were exluded from the trial and were regarded as having ontriuted to the learning urve for the method. The results of the perutaneous iopsy were enign for these patients. Nine additional patients with nine lesions were exluded due to the presene of non-mass lesions with asymmetry and arhitetural distortions on the ultrasound. All mass lesions aording to the BI-RADS TM lexion were inluded in the study. Two-hundred and eleven (92.5%) ore iopsies and 17 (7.5%) preoperative loalizations were arried out after performing and doumenting the elastographi study; there was histologial onfirmation in all ases. The trial was approved y the institutional review oard. Informed onsent was otained from all the partiipating patients. Equipment Both the onventional and the elastographi studies were performed y two radiologists with 6 and 17 years of experiene in reast imaging, respetively. Examinations were performed using a Sonix SP (Ultrasonix Medial Corporation, Vanouver, Canada) US system and a 5 14 MHz multifrequeny linear proe. For the elastography study, a speial software was used for the Ultrasonix system, version (Beta 1) updated to the ommerial version 2.6. Elastographi study The elastographi study was performed with the patient lying in the same position used for onventional US examination (B-mode) and with the transduer positioned perpendiular to the region of interest (ROI). The target lesion was repeatedly ompressed efore examination to ensure that there was no lateral shift. After the ativation of elastography, ontinuous manual ompression was applied to the target region, perpendiular to the petoral musle, until tissue resistane was deteted. When resistane was felt, manual pressure was interrupted, allowing spontaneous deompression of the reast parenhyma. The study area omprised the region from the suutaneous tissue to the petoral musle and also the margins of the mass up to 0.5 m. The elastographi imaging tehnique used omprises three phases: Step 1: Tissue is imaged with and without light ompression, and RF (radiofrequeny) data lines are aquired in the ROI. Step 2: A displaement is estimated etween every two lines of RF data. Step 3: A strain value for every point in the ROI is estimated ased on RF data deformation. Eah pixel of the elastiity image was assigned one of 256 speifi olors, depending on the magnitude of strain. The sale ranged from red for omponents with greatest strain (i.e., softest omponents) to lue for those with no strain (i.e., firmest omponents). Green indiated average strain in the ROI. The images otained through elastography were superimposed to the B-mode images. The images aquired were assessed in real time through inememory, whih allows retrospetive evaluation of the ehavior of the mass during ompression and after deompression. Overall, study duration did not exeed two minutes per lesion. Elastographi lassifiation The elastographi lassifiation used a four-point sale aording to the olor variation during ompression and after deompression of the ROI (Fig. 1). A sore of 1 was assigned to lesions presenting the same olor spetrum as the peripheral reast tissue (Fig. 2). A sore of 2 was assigned to lesions that after deompression presented variation to lighter strains of more than 50% of the mass area when ompared with the image aquired during ompression (Fig. 3). A sore of 3 was assigned to lesions presenting olor variation of less than 50% of the lesion area (etween 10% and 50%) after deompression (Fig. 4). Finally, a sore of 4 was assigned to the lesions presenting no relevant olor variation Volume 15 Issue 2 New elastographi lassifiation of reast lesions 97

3 a Figure 2. a. Example of sore 1. Mass in B-mode (a), during ompression () and after deompression (). There is no definition of the tissue mass when the elastographi study is assoiated with B-mode, with histology showing firoysti hange in the reast of a 26-year-old woman. Biopsy samples were otained y ore iopsy. a Figure 3. a. Example of sore 2. Mass in B-mode (a), during ompression () and after deompression (). The tissue mass presents an intense posterior aousti shadow, and is poorly haraterized in B-mode (a) during ompression; the mass is etter defined () than after deompression () and there is a hange in olor overing more than 50% of its area, whih suggests it is enign. The histologial study has onfirmed that the tissue mass is enign, ompatile with a firoadenoma in involution in a 46-year-old woman. Fragments were otained y ore iopsy. a Figure 4. a. Example of sore 3. Mass in B-mode (a), during ompression (), and after deompression (). The ovoid mass limited to B- mode (a). During ompression the mass is distinguished from normal tissue, more solid than the tissue (), and after deompression () there is a hange in olor overing less than 50% of the tissue mass, suggesting that it is enign. The histologial study has onfirmed that the mass is enign, ompatile with firoadenoma in a 33-year-old woman. Biopsy samples were otained y ore iopsy. 98 June 2009 Diagnosti and Interventional Radiology Fleury et al.

4 a Figure 5. a. Example of sore 4. Mass in B-mode (a), during ompression (), and after deompression (). The ovoid mass with irumsried margins limited to B-mode (a). During ompression the mass is distinguished from the normal tissue (), and after deompression () there is no hange in olor of the mass, suggesting malignany. Histologial study has onfirmed an invasive loular arinoma in a 45-year-old woman. during ompression and after deompression of the parenhyma, appearing lue on oth images (Fig. 5) (12). The lassifiation system used y the authors is similar to the one proposed y Saperrotta et al. (11); however, this proposed system relies on the assessment of the images during deompression periods, where the influene of the manner in whih the ompression is applied in the region of interest is smaller, thus simplifying the study systematization. Pathologial diagnosis All samples otained were sent for histologial study, and were analyzed y a speialized reast pathologist with 17 years of experiene. The lesions were divided into two groups: Group 1, enign lesions; and Group 2, malignant lesions. Group 1 was divided into three sugroups: Group 1a with firoysti alterations, Group 1 with firoadenomas, and Group 1 with low malignant potential, inluding papillomas, radial sars, myoepitheliomas and slerosing Tale 1. Histologial results. Distriution of the histologial results aording to group division Results Group Lesions Perentage (%) Firoysti hanges 1a Firoadenoma Low malignant potential Malignany Total Tale 2. Mean, median, and standard deviation of the elastography sores aording to histologial groups Histology Group Mean Median Standard deviation Firoysti hanges 1a Firoadenoma Low malignant potential Malignant lesions, aording to the lassifiation proposed y Ellis et al. (13, 14). Statistial analysis Sensitivity, speifiity, positive preditive value, and negative preditive value y elastography were evaluated in omparison with the histologial results of the samples. Sores 1, 2, and 3 were onsidered negative, and sore 4 was onsidered positive. For omparison purposes, these values were alulated for onventional ultrasound (B-Mode), with the lesions lassified as BI-RADS TM ategories 1, 2, and 3 onsidered negative, and 4 and 5 positive. The iopsied lesions were from patients with indiation for iopsy referred from other servies, and were lassified as follows: 118 (51.8%) BI-RADS TM ategory 3, 104 (45.6%) BI- RADS TM ategory 4, and 6 (2.6%) BI- RADS TM ategory 5. Before iopsies were performed, these lesions were relassified in our servie as follows: three (1.3%) BI- RADS TM ategory 1, 20 (8.8%) BI-RAD- S TM ategory 2, 138 (60.5%) BI-RAD- S TM ategory 3, 57 (25%) BI-RADS TM ategory 4, and 10 (4.4%) BI-RADS TM ategory 5. The lesions relassified as BI-RADS TM ategory 1 were interpreted in our servie as areas of firoysti hanges interspersed in the heterogeneous reast tissue. However, iopsies were performed to onfirm these findings. Fisher s exat test was used to test the assoiation etween the elastogram and the histologial result, with Volume 15 Issue 2 New elastographi lassifiation of reast lesions 99

5 signifiane at P < 0.05, using the ommerial software SPSS 12.0 (SPSS In., Chiago, USA). The auray of the method was also determined using the parametri estimate of the area under the reeiver operating harateristi (ROC) urve, using the ommerial Stata 8.0 software (StataCorp, College Park, Texas, USA). In order to assess the agreement etween oservers, a kappa test aording to the riteria desried y Landis and Koh was used (15). All signifiane proailities (P values) shown were two-sided. Values <0.05 were onsidered statistially signifiant. The software SAS (Statistial Analysis System, Cary, North Carolina, USA) was used for the alulations. For this purpose, all lesions were assessed again and relassified y eah of the oservers using inememory. Tale 3. Frequeny and perentage of histologial groups aording to elastography sores Histology Group Sore 1 Sore 2 Sore 3 Sore 4 Total Firoysti hanges 1a 6 (9.2%) 49 (75.4%) 7 (10.8%) 3 (4.6%) 65 (100%) Firoadenomas 1 2 (1.8%) 56 (50%) 52 (46.4%) 2 (1.8%) 112 (100%) Low malignant potential 100 June 2009 Diagnosti and Interventional Radiology In addition, the optimal ut-off point was determined aording to the Youden index (J) (16), J = max[se i + SP i - 1] for the proposed sores, where SE i and SP i are the values for sensiility and speifiity for all possile ut-off points. Results Pathologial diagnosis Of the 228 lesions evaluated, 65 (28.5%) were inluded in Group 1a; 112 (49.1%) in Group 1; 21 (9.2%) in Group 1; and 30 (13.2%) in Group 2 (Tale 1). Of the 30 malignant results (Group 2), 19 (63.4%) were invasive dutal arinomas; nine (30.0%) were invasive loular arinomas; one (3.3%) was a papillary arinoma; and one (3.3%) was a arinoid tumor. Of the 21 lesions lassified in Group 1, 13 (61.9%) were otained during (38.1%) 10 (47.6%) 3 (14.3%) 21 (100%) Malignant (13.3%) 26 (86.7%) 10 (100%) Tale 4. True-positive results (TP), true-negative results (TN), false-positive results (FP), and false-negative results (FN) for the elastography sores Sore 1 Sore 2 Sore 3 Sore 4 Total TP TN FP FN Total Tale 5. True-positive results (TP), true-negative results (TN), false-positive results (FP), and false-negative results (FN) for the B-mode study BI-RADS 1 BI-RADS 2 BI-RADS 3 BI-RADS 4 BI-RADS 5 Total TP TN FP FN Total surgial exisional iopsy. Of the remaining eight lesions, five (23.8%) underwent exisional iopsy after the diagnosis from the perutaneous iopsy, and were onsidered enign. Imaging follow-up for one year was performed for the other three lesions (14.3%). Malignany was not oserved in any of these ases. Elastography sores The mean and median sores for the histologial lassifiation of the lesions were 2.1 and 2.0, respetively, for Group 1a; 2.5 and 2.0 for Group 1; 2.8 and 3.0 for Group 1; and 3.9 and 4.0 for Group 2 (Tale 2). Tale 3 shows the frequeny of the histologial groups, aording to the eletrographi sores (Tale 3). The four (1.7%) false-negative results otained y the elastogram were lassified as sore 3, with the following findings: two loular arinomas with diameters of 0.9 m and 0.8 m, respetively, one papillary arinoma of 2.3 m, and one arinoid tumor of 1.2 m. The positive and negative preditive values, sensitivity, speifiity, and diagnosti auray of the elastographi sores were 76.47%, 97.94%, 86.67%, 95.96%, and 94.74%, respetively (Tale 4). When the sores were analyzed separately, we otained an NPV of 100% for sore 1, an NPV of 100% for sore 2, an NPV of 94.52% for sore 3, and a PPV of 76.47% for sore 4. For the onventional study, we otained 90% sensitivity, 79.80% speifiity, and 81.14% diagnosti auray (Tale 5). For the elastogram, the ROC urves, revealed an area under the urve of 0.954, a onfidene interval etween and 0.982, and error of (Fig. 6). For the onventional study, ROC urves revealed an area under the urve of 0.888, a onfidene interval etween and 0.946, and error of (Fig. 6). The optimal ut-off point identified for the proposed lassifiation was sore 3, whih is the ut-off point orresponding to the maximum value of the Youden index aounting for sensitivity, speifiity, and diagnosti auray of 88%, 98.2%, and 96.5%, respetively, for oserver 1, and 86.2%, 95.5%, and 94.2%, respetively, for oserver 2. No interoserver statistially signifiant differene was oserved at a signifiane level of 5%. Fleury et al.

6 Figure 6. Reeiver operating harateristi (ROC) urves for elastography and onventional US. The assessment of the risk of malignany improved when elastography imaging was availale. Disussion Over the last deade, elastography has eome an important tool for the study of soft tissues, with the linial perspetive of deteting lesions and determining pathologial tissue alterations, enaling the adequate treatment of lesions (17 20). The information aquired is similar to that otained through manual palpation; however, the data from elastography studies is more sensitive and less sujetive (21 25). Studies onduted to evaluate lesions aording to the size, using software with whih soft lesions appear lighter, and hard lesions appear darker, and malignant lesions tend to e more evident than enign lesions, showed good diagnosti auray; however, the main limitation of the method was the interoserver variaility. Moon et al. (8) arried out a study of 100 reast masses y ontinual elastography exam using a omputer assisted diagnosti program through segmented images of the masses, evaluating the margins, hanges in anteroposterior distane, and differenes etween the total area and hardness. These authors otained a sensitivity of 85%, speifiity of 88%, diagnosti auray of 87%, and negative preditive value of 90%. Based on a similar onept, ut also onsidering the interoserver variaility, Regner et al. (7) onluded that elastography has good potential for differentiating reast tissue lesions, ut that one of its main limitations was low speifiity for some readers (24%). Burnside et al. (6), presented similar results, and onluded that interoserver variaility and image quality interfere with the oserver performane. In the present study, we did not onsider tumor size measurements using elastography to e an adequate riterion for lesion analysis, mainly eause during the proess of the elastographi study, the definition of the margins of the lesion is poor, and this may influene the final result of the analysis. This may e one of the fators that ontriuted to the inter-oserver variation reported in these series. The other researh line ased on the use of software that applies a different olor spetrum to tissues aording to their hardness showed good diagnosti auray, although there were no studies evaluating interoserver variaility. Itoh et al. (9) proposed a lassifiation for the lesions aording to the olor spetrum variation otained in a dynami form. A total of 111 lesions were assessed using a five-point lassifiation system, where sores of 1, 2, and 3 were onsidered enign, and sores 4 and 5 were onsidered malignant. These authors otained a sensitivity of 86.5%, speifiity of 89.8% and diagnosti auray of 88.3%. Saperrotta et al. (11), using a simplified threepoint lassifiation system ased on the lassifiation y Itoh et al. showed a sensitivity of 80% and a speifiity of 80.9%, whih was similar to findings of the ultrasound study. They onluded that elastography may e a useful aid to US for less experiened radiologists in the assessment of solid non-palpale reast lesions, espeially BI-RADS 3, for whih speifiity was higher (88.7%). The main limitation reported y the authors is that elastography is operator-dependent, and there may e an interoserver variaility; however, all studies onluded that sonoelastography requires training and pratie to learn the appropriate tehnique. There are no reports on elastography studies using the images aquired after deompression, ut only pre- and post-ompression images in the region of interest. We elieve that studies arried out during ompression and after deompression, with ompressions performed aording to the proposed systematization, an not only standardize the study, ut also an provide parameters for omparison etween these two time points. We onsider that the diffiulty in quantifying the strength to e applied for parenhymal ompression an e overome y using gradual ompression, until resistane is felt in the reast on whih fore was not applied, and taking into aount the moment of spontaneous deompression. Aording to Hooke s law (26), when elasti deformity is reated in a material (strain) y an external fore (stress), the aumulated energy (elasti potential energy) allows the material to restore its original shape after deompression. Beause it is umulative, the elasti potential energy, whih is the fore used during the deompression period, does not depend on the manner (aeleration) with whih ompression is performed. This ould minimize the main limitation of the method, whih is the lak of a standard for the way in whih the ompression fore should e applied. The soring system used in this study is similar to those used y Saperrotta et al. (11), although in this study, images in the ompression and deompression Volume 15 Issue 2 New elastographi lassifiation of reast lesions 101

7 a d Figure 7. a d. Example of reast arinoma. B- mode study (a), histology (), elastography during ompression (), and after deompression (d). Large irregular mass (4.2 m) lassified as BI-RADS TM 5. In the elastographi study, this lesion was lassified as sore 4 with no olor variation oserved during or after deompression. Histology of the iopsy speimen showed invasive dutal arinoma. the papillary arinoma, are lesions that are generally softer on manual palpation (27, 28). Our results differ from the ones reported in the literature, whih showed that elastography has etter auray for lesions smaller than 2.0 m. Our study demonstrated that firoadenomas larger than 2.0 m were lassified with malignant sores, and onversely, that small malignant tumors tended to present enign sores. Figure 7 shows a dutal arinoma of approximately 4.2 m lassified as sore 4 y elastography, showing a etter orrelation with histologial type than with size for the diagnosis of malignany. Further studies are needed to onfirm whether there is a higher orrelation of elastographi sores with the histology of the lesion or with its size. One of the limitations of our study was the small sample of malignant lesions ompared to the enign lesions. However, this ratio is similar to the ratio oserved in linial pratie. There was also a higher prevalene of loular arinomas than is desried in the literature, and this may have influened our results. The lassifiation y elastography proposed here, through the evaluation of tissue after ompression and deompression of the reast parenhyma, an e an important tool, omined with ultrasonographi studies, for differentiating enign and malignant lesions among lesions lassified as true masses aording to the BI-RADS TM lexion. periods were ompared. Sores 1 and 2 were onsidered enign, sore 3 proaly enign, and 4 suggestive of malignany. We elieve that the simpler the lassifiation, the easier its appliaility and reproduiility. Analyzing the results otained in this study, it is lear that there is strong statistial evidene of an assoiation etween the histologial diagnosis and the sores proposed y the authors for elastography, wherey sores 1, 2, and 3 were onsidered negative for malignany, and sore 4 was onsidered suggestive of malignany (P < 0.001). The ROC urve showed an area under the urve of , demonstrating the exellent diagnosti auray of the method. Comparing this with the area otained in the onventional study, it was found that the investigator was etter ale to assess the risk of malignany when elastographi imaging was availale. We elieve that the sensitivity (76.46%), speifiity (97.49%) and diagnosti auray (86.67%) oserved in this study were higher than those oserved in other studies due to the exlusion of nonmass lesions with asymmetry and distortions. Suh lesions are omposed of healthy tissue interspersed within pathologi tissue, whih may lead to false-negative results. All of the false-negative results had een lassified as sore 3. Dimensions of the lesions that were mislassified varied in a great range (etween 0.8 and 2.3 m) whih showed the histologial type of the lesions influened the results more than did their dimensions. In the previous reports the tumors initially lassified as enign, partiularly the arinoid tumor and Referenes 1. Kaplan SS. Clinial utility of ilateral whole-reast US in the evaluation of women with dense reast tissue. Radiology 2001; 221: Stavros AT, Thikman D, Rapp CL, Dennis MA, Parker SH, Sisney GA. Solid reast nodules: use of sonography to distinguish etween enign and malignant lesions. Radiology 1995; 196: Amerian College of Radiology. Breast imaging reporting and data system (BI-RADS), ultrasound. 4th ed. Reston, Va: Amerian College of Radiology, Ophir J, Céspedes I, Ponnekanti H, Yazdi Y, Li X. Elastography: a quantitative method for imaging the elastiity of iologial tissues. Ultrason Imaging 1991; 13: Cho N, Moon WK, Park JS, Cha JH, Jang M, Seong MH. Nonpalpale reast masses: evaluation y US elastography. Korean J Radiol 2008; 9: Burnside ES, Hall TJ, Sommer AM, et al. Differentiating enign from malignant solid reast masses with US strain imaging. Radiology 2007; 245: June 2009 Diagnosti and Interventional Radiology Fleury et al.

8 7. Regner DM, Hesley GK, Hangiandreou NJ, et al. Breast lesions: evaluation with US strain imaging linial experiene of multiple oservers. Radiology 2006; 238: Moon WK, Chang RF, Chen CJ, Chen DR, Chen WL. Solid reast masses: lassifiation with omputer-aided analysis of ontinuous US images otained with proe ompression. Radiology 2005; 236: Itoh A, Ueno E, Tohno E, et al. Breast disease: linial appliation of US elastography for diagnosis. Radiology 2006; 239: Tohno E, Ueno E. Current improvements in reast ultrasound, with a speial fous on elastography. Breast Caner 2008; 15: Saperrotta G, Ferranti C, Costa C, et al. Role of sonoelastography in non-palpale reast lesions. Eur Radiol 2008; 18: Fleury EFC, Rinaldi JFR, Piato S, Fleury JCV, Roveda Jr D. Features of ysti reast lesions at ultrasound elastography. Radiol Bras 2008; 41: Ellis IO, Humphreys S, Mihell M, Pinder SE, Wells CA, Zakhour HD; UK National Coordinating Committee for Breast Sreening Pathology; European Commission Working Group on Breast Sreening Pathology. Best Pratie No 179. Guidelines for reast needle ore iopsy handling and reporting in reast sreening assessment. J Clin Pathol 2004; 57: Courtillot C, Plu-Bureau G, Binart N, et al. Benign reast diseases. J Mammary Gland Biol Neoplasia 2005; 10: Landis JR, Koh GG. The measurement of oserver agreement for ategorial data. Biometris 1977; 33: Youden WJ. Index for rating diagnosti tests. Caner 1950; 3: Hoyt K, Forserg F, Ophir J. Analysis of a hyrid spetral strain estimation tehnique in elastography. Phys Med Biol 2006; 51: O Donnell M, Skovoroda AR, Shapo BM, Emelianov SY. Internal displaement and strain imaging using ultrasoni spekletraking. IEEE Trans Ultrason Ferroeletr Freq Control 1994; 41: Emelianov SY, Luinski MA, Weitzel WF, Wiggins RC, Skovoroda AR, O Donnell M. Elastiity imaging for early detetion of renal pathology. Ultrasound Med Biol 1995; 21: Doyley MM, Bamer JC, Fuehsel F, Bush NL. A freehand elastographi imaging approah for linial reast imaging: system development and performane evaluation. Ultrasound Med Biol 2001; 27: Krouskop TA, Wheeler F, Kallel F, Garra B, Hall T. Elasti moduli of reast and prostate tissues under ompression. Ultrason Imaging 1998; 20: Konofagou EE, Ophir J, Krouskop TA, Garra BS. Elastography: from theory to linial appliations Summer Bioengineering Conferene, June 25 29, Key Bisayne, Florida, USA. 23. Hall TJ. AAPM/RSNA physis tutorial for residents: topis in US: eyond the asis: elastiity imaging with US. Radiographis 2003; 23: Garra BS, Cespedes EI, Ophir J, et al. Elastography of reast lesions: initial linial results. Radiology 1997; 202: Hall TJ, Zhu Y, Spalding CS. In vivo real-time freehand palpation imaging. Ultrasound Med Biol 2003; 29: Hooke s law. Enylopedia Britannia We site. Aessed Otoer 18, Gupta C, Malani AK, Rangineni S. Breast metastasis of ilial arinoid tumor: ase report and literature review. World J Surg Onol 2006; 4: Soo MS, Willifond ME, Walsh R, Bentley RC, Kornguth PJ. Papillary arinoma of the reast: imaging findings. AJR Am J Roentgenol 1995; 164: Volume 15 Issue 2 New elastographi lassifiation of reast lesions 103