Women s Imaging Original Research

Transcription

1 Women s Imaging Original Research Waldherr et al. One-View Breast Tomosynthesis Versus Two-View Mammography Women s Imaging Original Research Christian Waldherr 1 Peter Cerny 1 Hans J. Altermatt 2 Gilles Berclaz 3 Michele Ciriolo 4 Katharina Buser 4 Martin J. Sonnenschein 1 Waldherr C, Cerny P, Altermatt HJ, et al. Keywords: breast cancer, mammography, screening, senology, tomosynthesis DOI: /AJR Received November 4, 2011; accepted after revision June 27, The authors disclose any financial relationship with a commercial organization. 1 Division of Radiology, Breast Center Bern, Klinik Engeried, Sonnenhof AG, Riedweg 15, 3012 Bern, Switzerland. Address correspondence to C. Waldherr (christianwaldherr@sonnenhof.ch). 2 Division of Histopathology, Institute of Pathology Länggasse, Bern, Switzerland. 3 Division of Gynecology and Surgical Oncology, Breast Center Bern, Klinik Engeried, Bern, Switzerland. 4 Division of Oncology, Breast Center Bern, Klinik Engeried, Bern, Switzerland. AJR 2013; 200: X/13/ American Roentgen Ray Society Value of One-View Breast Tomosynthesis Versus Two-View Mammography in Diagnostic Workup of Women With Clinical Signs and Symptoms and in Women Recalled From Screening OBJECTIVE. The purpose of this study is to compare the diagnostic value of one-view digital breast tomosynthesis versus two-view full-field digital mammography (FFDM) alone, and versus a combined reading of both modalities. MATERIALS AND METHODS. The datasets of one-view digital breast tomosynthesis and two-view FFDM of abnormal mammograms in 144 consecutive women admitted for diagnostic workup with clinical signs and symptoms (n = 78) or recalled from screening (n = 66) were read alone and in a combined setting. The malignant or benign nature of the lesions was established by histologic analysis of biopsied lesions or by month follow-up. RESULTS. Eighty-six of the 144 patients were found to have breast cancer. The BI-RADS categories for one-view digital breast tomosynthesis were significantly better than those for two-view FFDM (p < 0.001) and were equal to those of the combined reading in both women admitted for diagnostic workup and women recalled from screening. The sensitivity and negative predictive values of digital breast tomosynthesis were superior to those of FFDM in fatty and dense breasts overall and in women admitted for diagnostic workup and in women recalled from screening. Only 11% of digital breast tomosynthesis examinations required additional imaging, compared with 23% of FFDMs. CONCLUSION. In patients with abnormal mammograms, one-view digital breast tomosynthesis had better sensitivity and negative predictive value than did FFDM in patients with fatty and dense breasts. They also suggest that digital breast tomosynthesis would likely increase the predictive values if incorporated in routine screening. M ammography is the reference standard for screening and diagnostic imaging of breast cancer despite its limitations. The false-negative rate of mammography ranges from 8% to 66% in symptomatic women, depending on breast density and tumor type [1 6]. This limited sensitivity is mainly attributed to fibroglandular tissue overlapping the tumor [7]. If an abnormality is found, supplemental mammographic views and ultrasound should be used for further characterization. The main benefit of breast ultrasound is improved specificity when used in a targeted manner. MRI is more sensitive than mammography, ultrasound, or physical examination. However, a major disadvantage of MRI is limited specificity resulting from its enhancement of benign breast lesions and the need for an IV contrast agent [8, 9]. In recent years, new techniques aimed at improving detection of breast carcinomas, including full-field digital mammography (FFDM) [10 12], computer-aided detection [13 15], and, more recently, digital breast tomosynthesis [16 21], have been introduced. FFDM has been shown to significantly improve the performance of radiologists, particularly when interpreting examinations of women with dense breast tissue [11]. Digital breast tomosynthesis is a 3D radiographic technique that obtains 3D information from projection images obtained as the x-ray source moves in an arc above the breast. These images are then reconstructed into a set of images spaced at about 1-mm thickness but may also be recalculated using any slice separation desired. These digital breast tomosynthesis image sets can be obtained instead of a single projection image with FFDM. A digital breast tomosynthesis dataset may include images depending on the compressed breast thickness. Tissue overlap is reduced with digital breast tomosynthesis compared with FFDM because the breast is imaged with a series of images at different planes rather than with a single projection. In recent studies, two-view digital breast tomosynthesis was shown to be subjectively bet- 226 AJR:200, January 2013

2 One-View Breast Tomosynthesis Versus Two-View Mammography ter than two-view FFDM at highlighting masses and areas of architectural distortion with only minimal added radiation [16 21]. Conventional mammography, in contrast, is reported to be better at delineating the morphology of calcifications [17]. Still, digital breast tomosynthesis is thought to have the potential to lower the recall rate if used in combination with two-view FFDM [16 21]. Despite some promising studies, the relative utility of digital breast tomosynthesis versus two-view FFDM, as well as its general utility, is still poorly known. To determine the value of digital breast tomosynthesis in daily practice, we compared one-view digital breast tomosynthesis alone with two-view FFDM alone and with a combined reading of both modalities. The study design has a bias in favor of mammography because only women with an abnormal 2D mammogram were recruited. Women with lesions that were visible on digital breast tomosynthesis only were excluded. In our opinion, if the predictive values of one-view digital breast tomosynthesis could prevail in a study design such as this, one-view digital breast tomosynthesis has the potential to be a valuable tool in routine screening for breast cancer. Materials and Methods Review Board Approval This study was approved by the institute s review board. All patients volunteered to undergo digital breast tomosynthesis with one view (mediolateral oblique [MLO]) when recalled from screening or during workup for clinical signs or symptoms. Patients Included in this study were all 144 patients referred between December 2008 and June 2009 to the Breast Center with clinical signs and symptoms of breast disease (e.g., lumps, pain, or retracted nipple [n = 78]) or recalled from screening (suspicion of lesions or architectural distortion) for diagnostic workup (n = 66) and who had an abnormal two-view (craniocaudal and MLO) FFDM examination. Oneview (MLO) digital breast tomosynthesis of the abnormal side was obtained immediately after the twoview FFDM if the radiologist on duty observed an abnormality on the FFDM images. Two-View FFDM All two-view (craniocaudal and MLO) FFDM imaging was performed in standard compression by experienced technologists trained in breast radiology. One-View Digital Breast Tomosynthesis If an abnormality was found in the two-view FFDM, a one-view (MLO) digital breast tomosynthesis of the abnormal breast was performed that same day by specialized technologists using a prototype digital breast tomosynthesis system (Selenia, Hologic). The detector area of the prototype digital breast tomosynthesis system was cm, the same as that for the standard FFDM. The digital breast tomosynthesis acquisition parameters were as follows: 11 projection images acquired in increments of approximately 1.4, starting at 7.5 and ending at 7.5 ; imaging time of 4 seconds; and radiation dose equivalent to a single-view FFDM image. After acquisition, the 11 projection images were reconstructed into a series of planar images at 1-mm intervals from the breast support to the compression paddle. Reviewing Two-View FFDM and One-View Digital Breast Tomosynthesis Datasets At least 1 year after image acquisition, all twoview FFDM and one-view digital breast tomosynthesis datasets were reviewed either randomized separately (two-view FFDM datasets alone and one-view digital breast tomosynthesis datasets alone) or in a combined reading (two-view FFDM datasets and one-view digital breast tomosynthesis datasets together) by two blinded specialized breast radiologists in consensus using flat-panel screens (RadiForce GS510, EIZO) for two-view FFDM and a flat-panel display (Coronis 5MP Mammo, Barco) for one-view digital breast tomosynthesis. To minimize the bias, the interval between the two reading methods was at least 2 months. The reviewers categorized the likelihood of malignancy according to the American College of Radiology (ACR) BI-RADS categories (0 5) and categorized breast density by applying the 4-point ACR scale [22]. Qualitative categoric items, such as mass shape (round or oval, lobulated, or irregular), margins (sharp, microlobulated, indistinct, or spiculated), mass density (low, isodense, or high), architectural distortion, and calcifications (benign monomorphic, intermediate amorphic, or probably malignant polymorphic), were also scored. The radiologists were blinded to the findings of other modalities, to clinical reports, patient history, histology, and clinical follow-up, as well as to the reasons for referral. If the two readers could not reach consensus, datasets were forwarded to a third reviewer. Follow-Up Medical records and new diagnostic imaging of patients with BI-RADS categories 1 3 lesions were reviewed after months to determine whether breast cancer had occurred during this interval. Histologic Analysis Histologic verification of biopsied lesions (core biopsy, vacuum-assisted biopsy, or surgery) or month follow-up with ultrasound and FFDM or MRI was used to establish whether a lesion was malignant or benign. Statistical Analysis Descriptive values, such as BI-RADS category, mass density, margins, ACR density, and calcifications, in each modality were compared applying the Bowker chi-square test [23]. The kappa coefficient was used to determine interreader agreement. Correct BI-RADS categories, as verified by histology or month follow-up, were compared applying the McNemar chi-square test [23], and the interreader agreement was compared using the kappa coefficient. The predictive values of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were compared using the Cochran global test [23] and the McNemar chisquare test. The interreader agreement was assessed using the kappa coefficient. The relationship of correct BI-RADS category and 4-point ACR breast density score was compared calculating the McNemar chi-square test. All tests were two sided, with a p value of 0.05 considered to be statistically significant. SPSS (version 19, SPSS) and BiAS for Windows (version 9.06, Epsilon) were used for statistical analysis. Finally, the detection rate, correct BI-RADS up- and downgrading, the need for additional imaging, and the reader confidence in both two-view FFDM and one-view digital breast tomosynthesis were determined. Results Patients Histopathologic proof was available for 116 lesions or breasts, including all breasts with carcinoma (n = 86). The mean tumor size was 2.1 cm (range, cm). Scoring of Lesion Descriptors One-view digital breast tomosynthesis correlated significantly better with the combined reading for describing lesion margins than with the two-view FFDM reading alone. For all other lesion descriptors, the differences were not statistically significant (Table 1). Comparison of Correct BI-RADS Categories As shown in Tables 2 and 3, according to BI-RADS category, one-view digital breast tomosynthesis correctly scored scans in 121 of the 144 patients (84%), the combined reading correctly scored 122 (85%) scans, and two-view FFDM correctly scored 107 (74.3%) scans. In the event of a correct BI- RADS category in a patient with breast cancer, one-view digital breast tomosynthesis and the combined reading performed signifi- AJR:200, January

3 Waldherr et al. TABLE 1: Comparison of Interreader Agreement for Descriptive Values of One-View Digital Breast Tomosynthesis Alone Versus Two-View Full-Field Digital Mammography (FFDM) Alone, and Versus a Combined Reading of Both Modalities Variable FFDM vs Digital Breast Tomosynthesis FFDM vs Combined Reading Digital Breast Tomosynthesis vs Combined Reading ACR density score Masses Margins Mass density Calcifications Note Data are kappa coefficients. Comparisons were calculated for all 144 women admitted for diagnostic workup with clinical signs and symptoms or recalled from screening. ACR = American College of Radiology. TABLE 2: BI-RADS Categories for Full-Field Digital Mammography (FFDM), Digital Breast Tomosynthesis, and Combined Reading (CR) in All 144 Patients BI-RADS Category FFDM Tomosynthesis Digital Breast Combined Reading Benign Malignant Benign Malignant Benign Malignant 0 34 (57.8) 19 (22.1) 10 (17.5) 9 (10.5) 9 (15.8) 6 (7.0) 1 1 (1.8) 2 (2.3) 16 (28.1) 1 (1.2) 14 (24.6) 0 (0.0) 2 4 (7.0) 0 (0.0) 18 (31) 0 (0.0) 17 (29.3) 0 (0.0) 3 6 (10.5) 1 (1.2) 2 (3.5) 0 (0.0) 4 (7.0) 1 (1.2) 4 12 (21.1) 52 (60.5) 11 (19.3) 42 (48.8) 12 (21.1) 54 (62.8) 5 1 (1.8) 12 (14.0) 1 (1.8) 34 (39.5) 2 (3.5) 25 (29.1) Missing 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) Total 58 (100.0) 86 (100.0) 58 (100.0) 86 (100.0) 58 (100.0) 86 (100.0) Note Data are no. (%) of patients. TABLE 3: Correct BI-RADS Categories Breast Cancer Status, Imaging Modality BI-RADS Categories 0 3 BI-RADS Categories 4 and 5 With breast cancer (n = 86) FFDM 22 (25.6) 64 (74.4) Digital breast tomosynthesis 10 (11.6) 76 (88.4) Combined reading 7 (8.1) 79 (91.9) Without breast cancer (n = 58) FFDM 44 (76.8) 14 (23.2) Digital breast tomosynthesis 46 (78.9) 12 (21.1) Combined reading 44 (75.4) 14 (24.6) Note Data are no. (%) of patients. In cases of malignancy, only BI-RADS categories 4 and 5 were scored correctly. In cases of benignity, only BI-RADS categories 0 3 were scored correctly. FFDM = full-field digital mammography. cantly better than did two-view FFDM (p < 0.001, Cochran global test). In 19 patients (13.2%), the scores of both modalities were incorrect, in 104 patients (72.2%) the scores of both were correct, in 16 patients (11.1%) one-view digital breast tomosynthesis was correct and two-view FFDM was incorrect, and in three patients (2.1%) two-view FFDM was correct and one-view digital breast tomosynthesis was incorrect. The difference between one-view digital breast tomosynthesis and two-view FFDM was significant (p < 0.001, McNemar test). Predictive Values of All 144 Patients One-view digital breast tomosynthesis alone and the combined reading showed significantly better sensitivity and NPV than did twoview FFDM alone (Table 4). The three reading methods did not differ significantly for specificity or PPV. Predictive Values in Women Admitted With Clinical Signs and Symptoms and in Women Recalled From Screening As expected, the prevalence of carcinoma was much lower in the screening group (36.4%) than in the group with clinical signs and symptoms (80.8%). One-view digital breast tomosynthesis and the combined reading had significantly better sensitivity and NPV than did two-view FFDM in women with clinical signs and symptoms (Table 5) and in women recalled from screening (Table 6). The differences among the three reading methods were not significant for specificity or PPV. Sensitivity, Specificity, PPV, and NPV for Fatty and Dense Breasts Predictive values of the datasets of oneview digital breast tomosynthesis and two-view FFDM in breasts with low and high density are shown in Table 7. Sensitivity was significantly higher for one-view digital breast tomosynthesis than for two-view FFDM in both fatty (ACR density scores 1 and 2) and dense (ACR density scores 3 and 4) breasts. The sensitivity of one-view digital breast tomosynthesis for dense breasts was even higher than that of two-view FFDM for fatty breasts. The differences between the two modalities in specificity, PPV, and NPV were not significant for fatty or dense breasts. Correct BI-RADS Up- and Downgrading by One-View Digital Breast Tomosynthesis Versus Two-View FFDM One-view digital breast tomosynthesis performed a correct BI-RADS category upgrading in 33 of the 144 patients (22.9%) (Figs. 1 and 2) and a correct BI-RADS category downgrading in 18 patients, thus lowering the need for additional imaging from 23% for two-view FFDM to 11% for oneview digital breast tomosynthesis. Detection Rate One-view digital breast tomosynthesis detected two additional breast carcinomas that were scored as normal by FFDM. An additional four breast carcinomas with benign morphology on ultrasound were biopsied because of suspicious findings on digital breast tomosynthesis. Discussion The few studies examining the sensitivity and specificity of digital breast tomosynthesis 228 AJR:200, January 2013

4 One-View Breast Tomosynthesis Versus Two-View Mammography have lacked a relevant number of patients. Furthermore, to our knowledge, no previous studies have separately evaluated patients undergoing diagnostic workup and patients recalled TABLE 4: Sensitivity, Specificity, Positive Predictive Value (PPV), and Negative Predictive Value (NPV) of Full-Field Digital Mammography (FFDM), Digital Breast Tomosynthesis, and Combined Reading in All 144 Patients Value FFDM Digital Breast Tomosynthesis Combined Reading Sensitivity, % (95% CI) 74.4 ( ) 88.4 ( ) 91.9 ( ) Specificity, % (95% CI) 76.8 ( ) 78.9 ( ) 75.4 ( ) PPV, % (95% CI) 83.1 ( ) 86.4 ( ) 84.9 ( ) NPV, % (95% CI) 66.2 ( ) 81.8 ( ) 86.0 ( ) True-positive (%) True-negative (%) False-positive (%) False-negative (%) Note True-positive findings refer to BI-RADS categories 4 and 5 in case of malignancy, true-negative findings refer to BI-RADS categories 1 and 2 in case of benignity, false-positive findings refer to BI-RADS categories 4 and 5 in case of benignity, and false-negative findings refer to BI-RADS categories 1 and 2 in case of malignancy. TABLE 5: Predictive Values Sensitivity, Specificity, Positive Predictive Value (PPV), and Negative Predictive Value (NPV) for Full-Field Digital Mammography (FFDM), Digital Breast Tomosynthesis, and Combined Reading in 78 Women Admitted With Clinical Signs and Symptoms Imaging Modality Sensitivity Specificity PPV NPV FFDM Digital breast tomosynthesis Combined reading Note Data are percentages. The total prevalence of breast cancer among diagnostic examinations was 80.8% (63/78). TABLE 6: Predictive Values Sensitivity, Specificity, Positive Predictive Value (PPV), and Negative Predictive Value (NPV) of Full-Field Digital Mammography (FFDM), Digital Breast Tomosynthesis, and Combined Reading in the 66 Women Recalled From Screening Imaging Modality Sensitivity Specificity PPV NPV FFDM Digital breast tomosynthesis Combined reading Note Data are percentages. The total prevalence of breast cancer for recall screening examinations was 36.4% (24/66). signs and symptoms (n = 78) or recalled from screening (n = 66) were read alone and in a combined setting. The results reveal a substantial 14% improvement in sensitivity in breasts with high and low density using one-view digital breast tomosynthesis versus two-view FFDM. The results were the same for both patients with clinical signs and symptoms and patients recalled from screening. Our present findings are supported, at least, in part by the literature. In the study by Noroozian et al. [24] comparing digital breast tomosynthesis to mammographic spot views for characterizing breast masses, four blinded radiologists individually evaluated digital breast tomosynthesis images and mammographic spot views of 67 masses in 67 women. The mean mass visibility ratings were slightly better with digital breast tomosynthesis than with mammographic spot view for all four readers. Gur et al. [25] conducted a free-response receiver operating characteristic analysis of a previously reported study in which eight experienced radiologists interpreted 125 examinations, 35 of them with verified cancers. On average, the radiologists achieved a 16% improvement in performance using a combined FFDM and digital breast tomosynthesis display mode than with the FFDM alone mode (95% CI, 7% 26%; p < 0.01). Tingberg et al. [26] reviewed digital breast tomosynthesis and FFDM examinations in over 2000 women. Initial experience showed that digital breast tomosynthesis had greater diagnostic precision and superior accuracy of size measurements and stage determination than did FFDM. In a study by Hakim et al. [27], four experienced radiologists reviewed the imaging studies, including FFDM, additional diagnostic views, and digital breast tomosynthesis, of 25 women with known masses. FFDM and digital breast tomosynthesis combined were perceived to be better for diagnosis in 50% (50/100) of the ratings (25 cases four readers = 100 ratings) compared with FFDM alone and additional views. Andersson et al. [20] conducted an unblinded study of 36 women to compare breast cancer visibility in TABLE 7: Sensitivity, Specificity, Positive Predictive Value (PPV), and Negative Predictive Value (NPV) Using One-View Digital Breast Tomosynthesis and Two-View Full-Field Digital Mammography (FFDM) for Fatty and Dense Breasts Breast Density, Imaging Modality Sensitivity Specificity PPV NPV Low density (ACR density scores 1 and 2) Digital breast tomosynthesis 94.4 ( ) 74.1 ( ) 82.9 ( ) 90.5 ( ) FFDM 78.6 ( ) 73.3 ( ) 80.5 ( ) 70.9 ( ) High density (ACR density scores 3 and 4) Digital breast tomosynthesis 84.0 ( ) 83.9 ( ) 89.4 ( ) 76.5 ( ) FFDM 70.5 ( ) 80.8 ( ) 86.1 ( ) 61.8 ( ) Note Data are percentage (95% CI). ACR = American College of Radiology. from screening. In the current study, the datasets of one-view digital breast tomosynthesis and two-view FFDM of 144 consecutive women admitted for diagnostic workup with clinical AJR:200, January

5 Waldherr et al. one-view digital breast tomosynthesis versus one- or two-view FFDM. Forty breast cancers were found in 37 breasts. The authors report that cancer visibility on digital breast tomosynthesis was greater in 22 of the 40 cancers compared with single-view FFDM. In a fully crossed mode-balanced observer performance study by Good et al. [19], 30 selected cases were read by nine experienced radiologists under three different display conditions. Several radiologists did perceive the digital breast tomosynthesis image set to be better than FFDM (p < 0.05) for diagnosing this specific cohort. In the current study, digital breast tomosynthesis was expected to be superior to FFDM in evaluating dense breasts, where tissue overlap is the major problem. In fact, not only was digital breast tomosynthesis found to be 15% more sensitive than FFDM in dense breasts, A Fig year-old woman with invasive ductal carcinoma with spiculated mass in right breast. A, Full-field digital mammogram shows mass in upper right breast. RMLO = right mediolateral oblique. B, Digital breast tomosynthesis image provides better depiction of spiculated border of mass and coexisting architectural distortion, finding highly suggestive of malignancy. but it was 20% more sensitive in fatty breasts. This can be explained in some patients by the ability of digital breast tomosynthesis to better delineate radial distortions of low-density lesions, better demarcate small lesions, and better assess lesion margins. The lesion margin is one of the key factors in determining whether a lesion is scored as benign or malignant. In addition to the superior sensitivity of digital breast tomosynthesis, our study reveals a significantly (15 20%) better NPV for digital breast tomosynthesis compared with FFDM, resulting from the reduced tissue overlap (especially in small lesions), better delineation of lesion margins, and improved demarcation of radial distortions. NPV is an essential value, especially in a screening situation, where a patient with a negative finding will not be recalled for 1 or even as long as 2 years. B The high specificity and PPV of FFDM are associated with the high number of BI- RADS category 0 lesions requiring additional imaging, a rate up to 50% higher than that of digital breast tomosynthesis. One-view digital breast tomosynthesis reduced the recall rate by more than 50% compared with two-view FFDM. Breast ultrasound is highly valuable in the diagnostic workup of women admitted with clinical signs and symptoms and in women with an abnormal mammogram. Its value as a screening tool, however, is compromised by issues related to its reproducibility, its high false-positive rates, its low PPV for biopsy recommendations, the operator dependency of its examinations, and its inability to image most cases of ductal carcinoma in situ. In the current study, we found four patients with abnormal digital breast tomosynthesis examinations, and occult breast cancer was found on ultrasound in four women admitted for diagnostic workup. Stereotactic or MRIguided biopsies were performed in these patients. This finding disconfirms the proposition that ultrasound could compensate for digital breast tomosynthesis. The present results are limited by the lack of compressions, magnification, or other diagnostic views that could have improved the diagnostic performance of the two-view FFDM. Despite the random readings and standardized time gaps between the readings, a certain bias by the three readers cannot be ruled out. The advantages of one-view digital breast tomosynthesis, such as its significant superior predictive values (sensitivity and NPV) and its significant reduction of the need for additional imaging in breasts of both low and high density, strongly support its routine use at least in patients with an abnormal mammography. In 2012, one manufacturer introduced a prototype one-view digital breast tomosynthesis system that is able to recalculate highquality 2D mammograms from the 3D dataset. This means that it is possible to get 2D and 3D images at the same time from a single scan by applying the same low dose used in a normal digital mammogram. Given this capability, plus the high sensitivity and specificity of digital breast tomosynthesis in breasts of low and high density, we can recommend digital breast tomosynthesis also for routine screening. Further studies are needed to examine the potential of digital breast tomosynthesis in a screening setting as well as its potential as a single diagnostic tool. 230 AJR:200, January 2013

6 One-View Breast Tomosynthesis Versus Two-View Mammography A B Fig year-old woman with invasive ductal carcinoma in right breast. A, Mediolateral oblique (MLO) full-field digital mammography projection shows distortion in upper right breast. B, MLO digital breast tomosynthesis projection shows spiculated mass. References 1. Wallis MG, Walsh MT, Lee JR. A review of false negative mammography in a symptomatic population. Clin Radiol 1991; 44: Coveney EC, Geraghty JG, O Laide R, Hourihane JB, O Higgins NJ. Reasons underlying negative mammography in patients with palpable breast cancer. Clin Radiol 1994; 49: Joensuu H, Asola R, Holli K, Kumpulainen E, Nikkanen V, Parvinen LM. Delayed diagnosis and large size of breast cancer after a false negative mammogram. Eur J Cancer 1994; 30A: Chew SB, Hughes M, Kennedy C, Gillett D, Carmalt H. Mammographically negative breast cancer at the Strathfield Breast Centre. Aust N Z J Surg 1996; 66: Moss HA, Britton PD, Flower CD, Freeman AH, Lomas DJ, Warren RM. How reliable is modern breast imaging in differentiating benign from malignant breast lesions in a symptomatic population? Clin Radiol 1999; 54: Kolb TM, Lichy J, Newhouse JH. Comparison of the performance of screening mammography, physical examination, and breast US and evaluation of factors that influence them: an analysis of 27,825 patient evaluations. Radiology 2002; 225: Murphy IG, Dillon MF, Doherty AO, et al. Analysis of patients with false negative mammography and symptomatic breast carcinoma. J Surg Oncol 2007; 96: Kuhl C. The current status of breast MR imaging. Part I. Choice of technique, image interpretation, diagnostic accuracy, and transfer to clinical practice. Radiology 2007; 244: Yabuuchi H, Matsuo Y, Okafuji T, et al. Enhanced mass on contrast-enhanced breast MR imaging: lesion characterization using combination of dynamic contrast-enhanced and diffusion-weighted MR images. J Magn Reson Imaging 2008; 28: D Orsi CJ, Newell MS. Digital mammography: clinical implementation and clinical trials. Semin Roentgenol 2007; 42: Pisano ED, Gatsonis C, Hendrick E, et al.; Digital Mammographic Imaging Screening Trial (DMIST) Investigators Group. Diagnostic performance of digital versus film mammography for breast-cancer screening. N Engl J Med 2005; 353: Rafferty EA. Digital mammography: novel applications. Radiol Clin North Am 2007; 45: James JJ, Gilbert FJ, Wallis MG, et al. Mammographic features of breast cancers at single reading with computer-aided detection and at double reading in a large multicenter prospective trial of computer-aided detection: CADET II. Radiology 2010; 256: Khoo LA, Taylor P, Given-Wilson RM. Computer-aided detection in the United Kingdom National Breast Screening Programme: prospective study. Radiology 2005; 237: Gilbert FJ, Astley SM, McGee MA, et al. Single reading with computer-aided detection and double reading of screening mammograms in the United Kingdom National Breast Screening Program. Radiology 2006; 241: Gur D, Abrams GS, Chough DM, et al. Digital breast tomosynthesis: observer performance study. AJR 2009; 193: Poplack SP, Tosteson TD, Kogel CA, Nagy HM. Digital breast tomosynthesis: initial experience in 98 women with abnormal digital screening mammography. AJR 2007; 189: Niklason LT, Christian BT, Niklason LE, et al. Digital breast tomosynthesis in breast imaging. Radiology 1997; 205: Good WF, Abrams GS, Catullo VJ, et al. Digital breast tomosynthesis: a pilot observer study. AJR 2008; 190: Andersson I, Ikeda DM, Zackrisson S, et al. Breast tomosynthesis and digital mammography: a comparison of breast cancer visibility and BI- RADS classification in a population of cancers with subtle mammographic findings. Eur Radiol 2008; 18: Teertstra HJ, Loo CE, van den Bosch MA, et al. Breast tomosynthesis in clinical practice: initial results. Eur Radiol 2010; 20: D Orsi CJ, Bassett LW, Berg WA, et al. BI-RADS: mammography, 4th ed. In: D Orsi CJ, Mendelson EB, Ikeda DM, et al., eds. Breast Imaging Reporting and Data System: ACR BI- RADS breast imaging atlas Reston, VA: American College of Radiology, Sachs L, Hedderich J. Angewandte Statistik, Methodensammlung, 12. Berlin, Germany: Springer- Verlag, Noroozian M, Hadjiiski L, Rahnama-Moghadam S, et al. Digital breast tomosynthesis is comparable to mammographic spot views for mass characterization. Radiology 2012; 262: Gur D, Bandos AI, Rockette HE, et al. Localized detection and classification of abnormalities on FFDM and tomosynthesis examinations rated under an FROC paradigm. AJR 2011; 196: Tingberg A, Förnvik D, Mattsson S, Svahn T, Timberg P, Zackrisson S. Breast cancer screening with tomosynthesis: initial experiences. Radiat Prot Dosimetry 2011; 147: Hakim CM, Chough DM, Ganott MA, Sumkin JH, Zuley ML, Gur D. Digital breast tomosynthesis in the diagnostic environment: a subjective side-byside review. AJR 2010; 195:528; [web]w172 W176 AJR:200, January