Tissue Harmonic Imaging - Wiley Online Library

Article

Tissue Harmonic Imaging Utility in Breast Sonography

Kazimierz T. Szopinski, MD, Anna M. Pajk, MD, Maciej Wysocki, MD, Dominique Amy, MD, Malgorzata Szopinska, MD, Wieslaw Jakubowski, MD

Objective. To determine the impact of tissue harmonic imaging on visualization of focal breast lesions and to compare gray scale contrast between focal breast lesions and fatty tissue of the breast between tissue harmonic imaging and fundamental frequency sonography. Methods. A prospective study was performed on 219 female patients (254 lesions) undergoing sonographically guided fine-needle biopsy. The fundamental frequency and tissue harmonic images of all lesions were obtained on a scanner with a wideband 7.5-MHz linear probe. Twenty-three breast carcinomas, 6 suspect lesions, 9 fibroadenomas, 1 papilloma, 1 phyllodes tumor, 162 unspecified solid benign lesions, and 40 cysts were found. In 12 cases the fine-needle aspiration did not yield sufficient material. The gray scale intensity of the lesions and adjacent fatty tissue was measured with graphics software, and the gray scale contrast between lesions and adjacent fatty tissue was calculated. Results. Tissue harmonic imaging improved the gray scale contrast between the fatty tissue and breast lesions in 230 lesions (90.6%; P < .001) compared with fundamental frequency images. The contrast improvement was bigger in breasts with predominantly fatty or mixed (fatty/glandular) composition than in predominantly glandular breasts. The overall conspicuity, lesion border definition, lesion content definition, and acoustic shadow conspicuity were improved or equal in the harmonic mode for all lesions. Conclusions. The tissue harmonic imaging technique used as an adjunct to conventional breast sonography may improve lesion detectability and characterization. Key words: breast carcinoma; breast lesions; breast sonography; needle biopsy, breast tissue; harmonic imaging.

Abbreviations THI, tissue harmonic imaging

Received September 20, 2002, from the Department of Diagnostic Imaging, Second Faculty of Medicine, Medical University of Warsaw, Warsaw, Poland (K.T.S., A.M.P., M.S., W.J.); Department of Pathology, Center of Medical Postgraduate Education, Warsaw, Poland (M.W.); and Centre d’Imagerie Medicale du Pays d'Aix, Aixen-Provence, France (D.A.). Revision requested October 31, 2002. Revised manuscript accepted for publication January 7, 2003. We thank Peter B. Barker, DPhil, for reviewing the manuscript and Anna Karwanska for help with statistical management of the data. Address correspondence and reprint requests to Kazimierz T Szopinski, MD, Department of Diagnostic Imaging, Second Faculty of Medicine, Medical University of Warsaw, ulica Kondratowicza 8, 03-242 Warsaw, Poland; e-mail: kazszo@poczta. onet.pl.

T

issue harmonic imaging (THI) is an increasingly used sonographic technique, which offers an improved signal-noise ratio, reduced side lobe artifacts, and improved lateral resolution compared with fundamental frequency sonography. It uses the effect of nonlinear propagation of the acoustic signal and generation of the higher-frequency signals in the insonated tissue.1,2 There are 2 basic harmonic imaging methods. In conventional frequency-based second-harmonic imaging, a narrowband pulse is emitted, and then high-pass or narrow-bandpass filtering is applied to the received echoes to filter out the fundamental echo components. This results in reduction of both spatial and contrast resolution. In wideband harmonic imaging (also called pulse inversion or phase inversion harmonic imaging) a train of 2 pulses is emitted, with the phase of the second inverted relative to the phase of the first.

© 2003 by the American Institute of Ultrasound in Medicine • J Ultrasound Med 22:479–487, 2003 • 0278-4297/03/$3.50

Tissue Harmonic Imaging in Breast Sonography

When the echoes from the transmitted pulses are added, the linear components of the echoes cancel each other, whereas the nonlinear components are amplified. This results in superior contrast and spatial resolution.3 Originally used in abdominal and pelvic studies, especially in obese patients, THI has also proved its efficacy in carotid, cardiac, and thyroid sonography.4–11 Recently, a few papers have reported subjective improvement of image features in THI applied to breast examination12,13 (C. L. Rapp, BS, RDMS, RDCS, and A. T. Stavros, MD, case study, GE Medical Systems, Waukesha, WI; available at: http://www.gemedicalsystems. com/rad/us/education/edu_rapparticle.html; accessed May 10, 2002). The purpose of this study was to determine whether use of THI might improve the gray scale contrast between breast lesions and adjacent fatty tissue and the visibility of breast lesions compared with fundamental frequency imaging.

Materials and Methods Subjects A prospective study was performed on 219 consecutive female patients (mean age, 48.7 years; range, 20–82 years) immediately before fine-needle biopsy of breast masses. A total of 254 focal breast lesions were examined. Before each investigation, the nature of the study had been explained to the patients, and their informed consent was obtained. Imaging Technique Sonographic images were obtained on a Sonoline Elegra scanner (Siemens AG, Erlangen, Germany) with a wideband 7.5-MHz linear probe. The default settings for breast imaging, as supplied by the manufacturer, were used. The harmonic images on the Sonoline Elegra machine were obtained with a phase inversion wideband harmonic technique. The actual emitter frequencies were 7.2 MHz in the fundamental mode and 3.4 MHz in the harmonic mode. All machine settings remained unchanged during imaging. If necessary, receiver gain was occasionally changed in the harmonic mode to obtain equal gray scale intensity of fatty tissue on the fundamental and harmonic frequency images. The mechanical index changed automatically according to the focal zone position and varied between 0.5 and 0.8 in the fundamental mode and between 1.1 and 1.6 in the harmonic mode. 480

A 180° turn of the probe was done on each lesion to find an area of fatty tissue located adjacent to the lesion at the same depth. The probe position was also adjusted to avoid overlap of the lesion with shadows from Cooper ligaments. Once the optimal imaging plane was chosen, a fundamental mode image of the lesion was obtained. Then the display was switched to the dual mode, and images in fundamental and harmonic modes were obtained without any change of the machine settings. After registration of this image, the receiver gain on the harmonic mode image was adjusted, if necessary, to obtain the same gray scale intensity of the fatty tissue as on the fundamental frequency image. This dualmode image was registered. At the end, a single harmonic mode image was obtained. The sonographic plane of the section was maintained constant while the images were obtained. The fine-needle aspirations were performed with 22- or 23-gauge needles under sonographic control. The material from each solid lesion was aspirated twice. In all cases, the tip of the needle was shown inside the sampled mass. The material was evaluated by a specialized cytologist (M.W.). Sonographic Image Analysis All images were saved on a magneto-optical disk. The images were transferred to a personal computer and saved in a bit-map format. The gray scale intensities of the lesions and adjacent fatty tissue were measured using Photoshop 5.0 software (Adobe Systems Incorporated, Mountain View, CA). The regions of interest were rectangular, covering the maximal portion of the lesion or fatty tissue cross section without crossing its borders. The relative gray scale contrast between lesions and adjacent fatty tissue was calculated according to the formula C = |(IL – IF)|/IF, where C was relative contrast; IL, gray scale intensity of the lesion; and IF, gray scale intensity of the fatty tissue. The difference in contrast between the harmonic (CH) and fundamental (CF) mode images was defined as CC = CH – CF. Three radiologists reviewed images independently and used a subjective scale (2 indicated better; 1, similar or the same; and 0, worse) to compare the tissue harmonic and fundamental frequency images. Overall conspicuity, border definition, internal structure definition, and acoustic shadow intensity were evaluated. A subjective scale score was established in each cateJ Ultrasound Med 22:479–487, 2003

Szopinski et al

gory by summing the observers’ results for each patient (the highest score was 6 when all observers rated the THI image as better than the fundamental frequency image, and the lowest score was 0 when all observers rated the THI image as worse). The readers were blinded to the final diagnosis; however, they were not blinded to the technique in which the images were obtained. Statistical Analyses The statistical methods used included the following: (1) 2-tailed paired Student t tests to assess the statistical significance of gray scale contrast difference between fundamental frequency and harmonic frequency images; (2) 2way analysis of variance (Duncan test) to assess the statistical significance of differences in contrast improvement between various types of lesions and dependence of contrast change on various mechanical index values and to determine whether the improvement of subjective scores was associated with the gray scale contrast improvement; (3) Pearson correlation methods to assess the dependence of gray scale contrast in harmonic images on the gray scale contrast in fundamental frequency images, dependence of gray scale contrast change on the original fundamental frequency gray scale contrast, and dependence of contrast change on the dimensions of lesions; (4) a κ statistic to determine the interobserver agreement between readers in subjective evaluation of overall visibility, border definition, internal structure definition, and shadow intensity; (5) a χ2 test to determine the dependence of subjective visibility scores on the breast tissue composition, on the lesion type, and on the mechanical index; and (6) a Kruskal-Wallis test to assess the dependence of gray scale contrast change on the tissue composition and dependence of subjective scores on the dimensions of the lesions. P < .05 was considered statistically significant. κ > 0.75 was interpreted as excellent reproducibility; κ = 0.4 to 0.75 as good reproducibility; and κ < 0.4 as marginal reproducibility.14 All tests were performed with SAS statistical software (SAS Institute Inc, Cary, NC).

Results Twenty-three lesions (9.1%) were cancers; 9 lesions (3.5%) were fibroadenomas; 1 lesion (0.4%) was a papilloma; 1 lesion (0.4%) was a J Ultrasound Med 22:479–487, 2003

phyllodes tumor; 162 lesions (63.8%) were rated as unspecified benign; and 40 lesions (15.7%) were cysts. Six lesions (2.36%) were rated as indeterminate, and in 12 cases (4.7%), no cellular material was obtained. Maximal lesion diameters ranged from 3 to 100 mm (mean, 12.1 mm). Objective Evaluation The gray scale contrast between a lesion and adjacent fat was better on harmonic imaging compared with the fundamental mode in 230 lesions (90.6%; Fig. 1). In 24 lesions (9.4%; 17 solid benign, 2 carcinomas, 1 fibroadenoma, 2 solid suspect/indeterminate lesions, and 2 cysts) the gray scale contrast was more reduced in the harmonic mode than in the fundamental mode. The changes in gray scale contrast were compared by a 2-tailed Student t test and were statistically significant (P < .0001). The results of gray scale contrast measurements in the fundamental and harmonic modes and gray scale contrast changes are summarized in Table 1. There was a positive correlation of gray scale contrast between fundamental and harmonic frequency images (r = 0.83; P < .0001). A negative correlation was found between gray scale contrast in fundamental frequency images and the gray scale contrast change (r = –0.37; P < .0001); that is, greater contrast improvement could be observed in lesions with less contrast in the original fundamental frequency images. No significant differences in gray scale contrast improvement were found between different lesion types. The change in gray scale contrast was not dependent on the dimensions of the lesions or the mechanical index value. The gray scale contrast change increased with the depth of the lesion position; however, this dependence was not statistically significant. The tissue composition of breasts was evaluated on the obtained images: 56 breasts were predominantly glandular; 65 were predominantly fatty; and 133 had mixed composition without obvious predominance of either tissue. The tissue composition was evaluated only in the plane containing the imaged lesion and did not reflect the predominant tissue composition of the whole breast. The mean gray scale contrast improvement was greater in predominantly fatty and mixed (glandular/fatty) breasts than in predominantly glandular breasts (P < .01). The results of gray scale contrast measurements in the fundamental and harmonic modes in breasts 481


of different tissue composition are summarized in Table 2. Subjective Evaluation The results of subjective assessment of the overall conspicuity, border definition, internal structure definition, and acoustic shadow intensity are summarized in Table 3. In all cases, the overall conspicuity, border definition (Fig. 2), internal structure definition (Fig. 3), and acoustic shadow intensity (Fig. 4) were better or similar in the harmonic mode. The interobserver reproducibility was good in the assessment of overall conspicuity, border definition, and internal structure definition and excellent in assessment of acoustic shadow intensity. The subjective scores are summarized in Table 4. The proportion of the highest subjective scores was higher in carcinomas than the mean proportion of the highest scores in all lesions. The lowest proportion of maximal subjective scores was observed in indetermi-

nate lesions. However, the statistical significance of the dependence of subjective scores on lesion type could not be shown. The subjective scores did not depend on the dimensions of the lesions. There was no statistically significant dependence of the scores for conspicuity, content, and border definition on breast tissue composition. Statistically significant dependence of shadow enhancement on breast tissue composition and dependence of the subjective scores on the type of lesion or the mechanical index could not be shown. The gray scale contrast change was significantly greater in the lesions with the highest scores of conspicuity and border and content definition improvement (score = 6) than in the lesions with the lowest scores (score = 3). This may be evidence that improved gray scale contrast reflects enhanced lesion visibility. There was no statistically significant dependence of the change in the intensity of acoustic shadows on the gray scale contrast change in the lesions.

Figure 1. Images from a 57-year-old woman with an 11-mm benign solid breast lesion. A, Sonogram in the fundamental mode showing a slightly hypoechoic mass, barely distinguishable from the fat, with a weak posterior acoustic shadow. B, Sonogram in the harmonic mode clearly showing an improvement in gray scale contrast and margin definition.

A

482

B

J Ultrasound Med 22:479–487, 2003

Szopinski et al

Discussion Objective Evaluation To our knowledge, this was the first attempt to quantify the gray scale contrast improvement in THI of the breast. The results of our study show that THI significantly improves the gray scale contrast between most focal breast lesions and the neighboring fatty tissue of the breast, which is reflected in improved conspicuity and better border and content definition of the lesions. The negative correlation between the original gray scale contrast in fundamental frequency images and the gray scale contrast change shows the benefit of THI in assessment of lesions with originally low contrast on fundamental frequency images. Our data confirm the opinion of C. L. Rapp, BS, RDMS, RDCS, and A. T. Stavros, MD (case study, GE Medical Systems) that solid nodules isoechoic to the surrounding tissue appear more hypoechoic on harmonic imaging. The fatty tissue gray scale intensity is the reference with which the echogenicity of breast lesions is usually compared.15 We chose for comparison the area of fatty tissue closest to the lesion, located, if possible, at the same distance from the surface of the probe as the lesion. In many cases (the predominantly glandular composition of a breast), however, an ideally positioned piece of fat was not available despite a 180° turn of the probe. The gray scale contrast

improvement was significantly bigger in the mixed composition and predominantly fatty breasts than in the predominantly glandular breasts. This supports the opinion that THI is especially beneficial in the presence of fatty tissue, which adversely affects the quality of the ultrasound beam.4 As could be expected, the gray scale contrast improvement depended on how deep the lesion was located in the breast. However, this dependence was not statistically significant, probably because of the small differences of distance. In our study, there was no significant difference in gray scale contrast change between malignant and benign lesions. In 2 cases of breast cancer (8.7% of all cancers) the gray scale contrast was diminished in harmonic images. The contrast decrease, however, was minimal (from 0.50 to 0.48 and from 0.89 to 0.84, respectively), and the overall conspicuity and border definition were rated equal (score = 3) in both cases by all observers. The content definition in 1 cancer was rated equal by all observers (score = 3), and in the second case, the content definition was rated better by 2 observers (score = 5). In both cases, the acoustic shadows were rated by 2 observers as stronger in the harmonic images (score = 5). Subjective Evaluation Our results concerning the subjective evaluation of the lesions are in agreement with the data of

Table 1. Gray Scale Contrast Between Lesions and Adjacent Fatty Tissue Lesion Type

Mean Contrast (Range) Fundamental Mode Harmonic Mode

Mean Contrast Change (Range)

Cancer (n = 23) Fibroadenoma (n = 9) Papilloma (n = 1) Phyllodes tumor (n = 1) Unspecified benign (n = 162) Suspect, indeterminate (n = 18) Cysts (n = 40) All lesions (n = 254)

0.41 0.31 0.57 0.22 0.33 0.40 0.62 0.39

0.16 0.16 0.15 0.16 0.14 0.10 0.13 0.14

(0.14–0.89) (0.05–0.73)

(0.0003–1.46) (0.13–0.79) (0.07–0.997) (0.0003–1.46)

0.56 0.47 0.73 0.39 0.46 0.52 0.75 0.52

(0.24–0.84) (0.08–0.84)

(0.01–1.38) (0.02–0.84) (0.32–1.0) (0.01–1.38)

(–0.05–0.42) (–0.16–0.37)

(–0.88–0.61) (–0.24–0.35) (–0.50–0.40) (–0.88–0.61)

Table 2. Comparison of THI Versus Fundamental Frequency Sonography for Gray Scale Contrast in Different Types of Breast Tissue Composition Breast Composition

Mean Contrast (Range) Fundamental Mode Harmonic Mode

Mean Contrast Change (Range)

Predominantly glandular Mixed Predominantly fatty

0.381 (0.029–0.997) 0.363 (0.0003–1.457) 0.433 (0.044–0.997)

0.1 (–0.235–0.370) 0.145 (–0.878–0.606) 0.145 (–0.179–0.372)


0.482 (0.048–0.9997) 0.508 (0.010–1.377) 0.578 (0.018–1.0)

483


Table 3. Comparison of THI Versus Fundamental Frequency Sonography for Overall Conspicuity, Border Definition, Internal Structure Definition, and Acoustic Shadow Intensity Observer 1, n (%) Better Similar Worse

Characteristic

Overall conspicuity Border definition Internal structure definition Acoustic shadow intensity

111 (43.7) 143 (56.3) 120 (47.2) 134 (52.8) 130 (51.2) 124 (48.8) 168 (66.1) 86 (32.9)

0 0 0 0

Observer 2, n (%) Better Similar Worse

114 (44.9) 140 (55.1) 126 (49.6) 128 (50.4) 138 (54.3) 116 (45.7) 161 (63.4) 93 (36.6)

others12,13 (C.L. Rapp, BS, RDMS, RDCS, A.T. Stavros, MD, case study, GE Medical Systems) who reported improved conspicuity, wall and margin definition, internal echo texture and the presence of internal echoes, extent of throughtransmission and clarity of the posterior echo pattern, and overall quality in most lesions compared with fundamental frequency images. In our data, however, only the intensity of acoustic echoes was rated as stronger in harmonic images in most cases by all observers. The inter-

0 0 0 0

Observer 3, n (%) Better Similar Worse

112 (44.1) 113 (44.5) 127 (50) 154 (60.6)

142 (55.9) 141 (55.5) 127 (50) 100 (39.4)

0 0 0 0

κ

0.69 0.65 0.65 0.81

nal echo definition was rated as better by 2 observers in most cases and by 1 observer in half of the cases. The overall conspicuity and border definition were rated by all observers as the same or similar in most cases. The discrepancy may be due to differences in applied evaluation criteria. In this study, a 3-level rating was used (better, same or similar, and worse), Rosen and Soo12 used a 5-level scale, and in a study by Seo et al,13 the images could be rated only as better or worse.

Figure 2. Images from a 64-year-old woman with an 11-mm breast carcinoma. A, Sonogram in the fundamental mode showing an irregularly shaped mass, slightly hypoechoic to the adjacent fatty tissue, with a weak posterior acoustic shadow. B, Sonogram in the harmonic mode showing a frankly hypoechoic mass with a strong acoustic shadow. The extensions in the upper portion of the tumor and microlobulations are clearly shown.

A

484

B


Szopinski et al

A

B

Figure 3. Images from a 53-year-old woman with an 8-mm complicated cyst. A, Sonogram in the fundamental mode showing an oval hypoechoic lesion. B, Sonogram in the harmonic mode clearly showing a fluid-debris level.

Figure 4. Images from a 58-year-old woman with a 14-mm breast carcinoma. A, Sonogram in the fundamental mode showing an irregularly shaped mass slightly hypoechoic to the adjacent fatty tissue, with a weak posterior acoustic shadow. B, Sonogram in the harmonic mode showing a frankly hypoechoic mass with a strong acoustic shadow.

A


B

485


Table 4. Subjective Scores of Change of Overall Conspicuity, Border Definition, Internal Structure Definition, and Acoustic Shadow Intensity Characteristic

3

Conspicuity Border definition Internal structure definition Acoustic shadows

112 (44.1) 104 (40.9) 91 (35.8) 77 (30.3)

Score, n (%) 4 5

31 (12.2) 25 (9.8) 27 (10.6) 15 (5.9)

27 (10.6) 41 (16.1) 40 (15.7) 18 (7.1)

1.

Ward B, Baker AC, Humphrey VF. Nonlinear propagation applied to the improvement of resolution in diagnostic medical ultrasound. J Acoust Soc Am 1997; 101:143–154.

2.

Whittingham TA. Tissue harmonic imaging. Eur Radiol 1999; 9:S323–S326.

3.

Haerten R, Lowery C, Becker G, Gebel M, Rosenthal S, Sauerbrei E. Ensemble(tm) tissue harmonic imaging: the technology and clinical utility. Electromedica 1999; 61:50–56.

4.

Desser TS, Jeffrey RB Jr, Lane MJ, Ralls PW. Tissue harmonic imaging: utility in abdominal and pelvic sonography. J Clin Ultrasound 1999; 27:135–141.

5.

Hong HS, Han JK, Kim TK, et al. Ultrasonographic evaluation of the gallbladder: comparison of fundamental, tissue harmonic, and pulse inversion harmonic imaging. J Ultrasound Med 2001; 20:35–41.

6.

Ono M, Asanuma T, Tanabe K, et al. Improved definition of left atrial thrombus by tissue harmonic imaging. J Ultrasound Med 1999; 18:253–255.

7.

Ortega D, Burns PN, Simpson DH, Wilson SR. Tissue harmonic imaging: is it a benefit for bile duct sonography? AJR Am J Roentgenol 2001; 176:653–659.

8.

Saleh A, Cupisti K, Feldkamp J, Grust A, Modder U. Does tissue harmonic imaging allow early determination of post-operative thyroid volume [in German]? Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 2001; 173:325–328.

9.

Shapiro RS, Wagreich J, Parsons RB, Stancato-Pasok A, Yeh HC, Lao R. Tissue harmonic imaging sonography: evaluation of image quality compared with conventional sonography. AJR Am J Roentgenol 1998; 171:1203–1206.

6

84 (33.1) 84 (33.1) 96 (37.8) 144 (56.7)

It is noteworthy that the cancers had the largest proportion of high subjective scores in all categories. That may be due in part to their complicated outlines, which were enhanced by the harmonic technique, and strong acoustic shadows. The fact that the observers were not blinded to the technique in which the images were obtained could have biased the subjective evaluation. However, we think that this kind of blinding would be somewhat artificial, because a radiologist experienced in sonography can readily tell a THI image from a fundamental frequency image of the same structure in virtually all cases. We are aware of possible limitations of fineneedle aspiration (which is the current policy at our institution) compared with core biopsy or open biopsy.16 Possible false-negative results may, to some extent, have influenced our results concerning the dependence of gray scale contrast on the nature of lesions. The fact that all the examinations were performed with 1 manufacturer’s equipment ensures uniformity of the data; however, it is also a limitation. The results we obtained may be corroborated by other investigators using other sonographic machines. Conclusions The results presented here suggest that THI substantially enhances the contrast between focal breast lesions and the surrounding fat. The overall conspicuity, border definition, and content definition are better or equal in the harmonic mode in all lesions. Tissue harmonic imaging enhances acoustic shadows, which are important sonographic signs. However, the strong acoustic shadows from the Cooper ligaments may limit the use of THI in “first-pass” scans. Tissue harmonic imaging may prove a valuable addition to conventional sonographic examinations of the breast, improving lesion visualization, and may be a problem-solving tool when studying suspect lesions or areas in the breast. 486

References

10. Treadwell MC, Seubert DE, Zador I, Goyert GL, Wolfe HM. Benefits associated with harmonic tissue imaging in the obstetric patient. Am J Obstet Gynecol 2000; 182:1620–1622. 11. Willinek WA, von Falkenhausen M, Strunk H, Schild HH. Tissue harmonic imaging in comparison to conventional sonography: influence on image quality and observer variability in the intima-media thickness measurements of the common carotid artery [in German]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 2000; 172:641–645.


Szopinski et al

12. Rosen EL, Soo MS. Tissue harmonic imaging sonography of breast lesions: improved margin analysis, conspicuity, and image quality compared with conventional ultrasound. Clin Imaging 2001; 25:379– 384. 13. Seo BK, Oh YW, Kim HR, et al. Sonographic evaluation of breast nodules: comparison of conventional, real-time compound, and pulse-inversion harmonic images. Korean J Radiol 2002; 3:38–44. 14. Rosner B. Fundamentals of Biostatistics. 4th ed. Belmont CA: Duxbury Press; 1995. 15. Stavros AT, Thickman D, Rapp CL, Dennis MA, Parker SH, Sisney GA. Solid breast nodules: use of sonography to distinguish between benign and malignant lesions. Radiology 1995; 196:123–134. 16. Pisano ED, Fajardo LL, Caudry DJ, et al. Fine-needle aspiration biopsy of nonpalpable breast lesions in a multicenter clinical trial: results from the Radiologic Diagnostic Oncology Group V. Radiology 2001; 219: 785–792.


487