Luca Alessandro Carbonaro

I.R.C.C.S. Policlinico San Donato, Milano, Lombardy, Italy

Are you Luca Alessandro Carbonaro?

Claim your profile

Publications (32)43.42 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Background and Objective. Vascularity evaluation on breast dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has a potential diagnostic value, but it represents a time consuming procedure, affected by intra- and inter-observer variability. This study tests the application of a recently published method to reproducibly quantify breast vascularity, and evaluates if the vascular volume of cancer-bearing breast, calculated from automatic vascular maps (AVMs), may correlate with pathologic tumour response after neoadjuvant chemotherapy (NAC). Methods Twenty-four patients with unilateral locally advanced breast cancer underwent DCE-MRI before and after NAC, 8 responders and 16 non-responders. A validated algorithm, based on multiscale 3D Hessian matrix analysis, provided AVMs and allowed the calculation of vessel volume before the initiation and after the last NAC cycle for each breast. For cancer bearing breast, the difference in vascular volume before and after NAC was compared in responders and non-responders using the Wilcoxon two-sample test. A radiologist evaluated the vascularity on the subtracted images (first enhanced minus unenhanced), before and after treatment, assigning a vascular score for each breast, according to the number of vessels with length ≥ 30 mm and maximal transverse diameter ≥ 2 mm. The same evaluation was repeated with the support of the simultaneous visualization of the AVMs. The two evaluations were compared in terms of mean number of vessels and mean vascular score per breast, in responders and non-responders, by use of Wilcoxon two sample test. For all the analysis, the statistical significance level was set at 0.05. Results For breasts harboring the cancer, evidence of a difference in vascular volume before and after NAC for responders (median = 1.71 cc) and non-responders (median = 0.41 cc) was found (p = 0.003). A significant difference was also found in the number of vessels (p = 0.03) and vascular score (p = 0.02) before or after NAC, according to the evaluation supported by the AVMs. Conclusions The encouraging, although preliminary, results of this study suggest the use of AVMs as new biomarker to evaluate the pathologic response after NAC, but also support their application in other breast DCE-MRI vessel analysis that are waiting for a reliable quantification method.
    Computer Methods and Programs in Biomedicine. 09/2014;
  • [Show abstract] [Hide abstract]
    ABSTRACT: OBJECTIVE. The purpose of this study was to investigate the diagnostic performance of unenhanced MRI in detecting breast cancer and to assess the impact of double reading. MATERIALS AND METHODS. A total of 116 breasts of 67 women who were 36-89 years old were studied at 1.5 T using an unenhanced protocol including axial T1-weighted gradient-echo, T2-weighted STIR, and echo-planar diffusion-weighted imaging (DWI). Two blinded readers (R1 and R2) independently evaluated unenhanced images using the BIRADS scale. A combination of pathology and negative follow-up served as the reference standard. McNemar and kappa statistics were used. RESULTS. Per-breast cancer prevalence was 37 of 116 (32%): 30 of 37 (81%) invasive ductal carcinoma, five of 37 (13%) ductal carcinoma in situ, and two of 37 (6%) invasive lobular carcinoma. Per-breast sensitivity of unenhanced MRI was 29 of 37 (78%) for R1, 28 of 37 (76%) for R2, and 29 of 37 (78%) for double reading. Specificity was 71 of 79 (90%) for both R1 and R2 and 69 of 79 (87%) for double reading. Double reading did not provide a significant increase in sensitivity. Interobserver agreement was almost perfect (Cohen κ = 0.873). CONCLUSION. An unenhanced breast MRI protocol composed of T1-weighted gradient echo, T2-weighted STIR, and echo-planar DWI enabled breast cancer detection with sensitivity of 76-78% and specificity of 90% without a gain in sensitivity from double reading.
    AJR. American journal of roentgenology. 09/2014; 203(3):674-81.
  • [Show abstract] [Hide abstract]
    ABSTRACT: To evaluate the performance of the first years since the beginning of a mammographic population-based screening program. Women aged 49-69 were invited biennially for two-view film-screen mammography and double reading without arbitration was performed. Interval cancers (ICs) from 2001 to 2006 were identified using screening archives, local pathology archives, and hospital discharge records. The proportional incidence of IC was determined considering breast cancers expected without screening. Three offsite radiologists experienced in breast cancer screening blindly evaluated mammograms prior to diagnosis, randomly mixed with negative mammograms (1:2 ratio). Cases unrecalled at review were considered as true ICs, those recalled by only one reviewer as minimal signs, and those recalled by two or three reviewers as missed cancers. T and N stage of the reviewed ICs were evaluated and compared. A total of 86,276 first level mammograms were performed. Mean recall rate was 6.8% at first and 4.6% at repeat screening. We had 476 screen-detected cancers and 145 ICs (10 of them ductal carcinomas in situ). Absolute incidence was 17 per 10,000 screening examinations. Invasive proportional incidence was 19% (44/234) in the first year, 39% (91/234) in the second year, and 29% (135/468) in the two-year interval. Of 145 ICs, 130 (90%) were reviewed mixed with 287 negative controls: 55% (71/130) resulted to be true ICs, 24% (31/130) minimal signs, and 22% (28/130) missed cancers. The rate of ICs diagnosed in the first year interval was 21% (15/71) for true ICs, 46% (13/28) for missed cancers, and 39% (12/31) for minimal signs, with a significant difference of true ICs rate compared to missed cancers rate (p=0.012). A higher rate of T3 and T4 stages was found for missed cancers (18%, 5/28) compared to minimal signs (6%, 2/31) or true ICs (8%, 6/71), while the rate of N2 and N3 stage for both minimal signs (19%, 6/31) or missed cancers (25%, 7/28) was higher than that for true ICs (10%, 7/71), although all these differences were not significant (p≥0.480). These results showed the possibility to comply with European Community standards in the first years of a screening program implementation.
    European journal of radiology 12/2013; · 2.65 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To evaluate the performance of a commercial computer aided detection (CAD) system (CAD BREAST DTS, im3D S.p.A.) for detecting lesions at digital breast tomosynthesis (DBT) on an independent testing set. METHOD AND MATERIALS The CAD system was retrospectively tested on a set of 143 patients. Craniocaudal (CC) and mediolateral oblique (MLO) DBT projections were acquired with a Hologic Selenia Dimensions system and reconstructed with the Briona library (Real Time Tomography LLC). All patients signed an informed consent form. A total of 80 histologically confirmed malignant lesions (57 masses, 18 microcalcification clusters and 6 masses with associated microcalcifications) were detected and annotated by experienced radiologists who drew a 3D bounding box around each lesion view. CAD BREAST DTS yields both masses and microcalcification clusters candidates. For masses, a CAD true positive was registered when the CAD marking overlapped by at least 20% the radiologists marking; for microcalcification clusters, when at least two of the microcalcifications identified by CAD fell within the radiologists marking. A CAD false positive was registered in all other cases, to avoid chance matchings. Masses with associated microcalcifications were considered correctly identified if CAD marked at least a mass or a microcalcification cluster. RESULTS At the selected operating point, per-lesion sensitivity was 89% (95% C.I. 80-94%). The system detected 48/56 masses, 17/18 microcalcification clusters and 6/6 masses with microcalcifications. Mean number of false positives per view was 2.8 ± 1.9 (mean ±standard deviation), of which 2 were marked as masses and 0.8 as microcalcification clusters. CONCLUSION The DBT CAD sensitivity is comparable to that observed for 2D digital mammography CAD systems, with a fairly low number of false positives per view. Further work, especially on difficult cases such as screening interval cancers, and comparing reading with and without CAD, is needed to understand its role in clinical practice. CLINICAL RELEVANCE/APPLICATION A commercial CAD system for masses and microcalcification clusters detection is evaluated on an independent testing set.
    Radiological Society of North America 2013 Scientific Assembly and Annual Meeting; 12/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To use morphologic and dynamic MRI features of the index lesion to predict ANS. METHOD AND MATERIALS Seventy patients (median age 60 yrs) were studied at 1.5 T with axial STIR and axial or coronal T1w gradient-echo dynamic sequences (gadobenate dimeglumine, 0.1 mmol/kg). All index lesions were evaluated using morphologic and dynamic descriptors by one reader with 5 years of experience in breast MRI; lesion size at maximum intensity projection of subtracted images was recorded. ANS was defined on the basis of sentinel node biopsy (SNB) or axillary dissection. Correlation between lesion features and ANS was estimated using χ2 and Mann-Whitney U tests. Subsequently, considering only variables significantly correlated to ANS, we arbitrarily fixed the positive predictive value (PPV) and the negative predictive value (NPV) at 100% and retrospectively reconstruct the corresponding diagnostic tests (respectively SPIN and SNOUT test). Ultrasonographic (US) and MRI suspicious lymph nodes were registered and compared with the new diagnostic criteria obtained at 100% predictive values. RESULTS At pathology after surgical excision we had: 38 invasive ductal carcinomas (IDC), 7 invasive lobular carcinomas, 16 IDC associated with ductal carcinoma in situ (DCIS), and 9 pure DCIS. SNB was negative (N-) in 56 patients while 24 patients had axillary nodal metastasis (N+) at histopathology. The ANS was significantly correlated to lesion size (P=.002) and to dynamic curve type (P=.029). To a fixed 100% PPV, corresponded a diagnostic combined criteria of a lesions size > 30 mm and a type3 dynamic curve (4/24 lesions, 17%); all these patients had N+ with no suspicious features at US or MRI. To a fixed 100% NPV corresponded a diagnostic combined criteria of a lesions size ≤ 14 mm and a type-1 or 2 dynamic curve (10/46 lesions, 22%). CONCLUSION Using size and dynamic enhancement descriptors of the index lesion, 17% of N+ could be recognized and treated directly with axillary dissection saving cost and time for SNB while 22% of N- could be assessed as true negative without any surgical procedure. CLINICAL RELEVANCE/APPLICATION Lesion size and dynamic enhancement descriptors can be used to predict ANS and avoid unnecessary sentinel node biopsy for both N+ and N- women in about 40% of patients.
    Radiological Society of North America 2012 Scientific Assembly and Annual Meeting; 11/2012
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE: This study was done to estimate the diagnostic performance of an asymmetric increase in breast vascularity (AIBV) for ipsilateral cancer. MATERIALS AND METHODS: A total of 197 patients without previous breast interventions underwent bilateral contrastenhanced (gadoterate meglumine, 0.1 mmol/kg) magnetic resonance (MR) imaging. Vessels >-2 mm in diameter and ≥3 cm in length were counted on maximum intensity projections: a difference ≥2 in number between the two breasts was considered AIBV. Pathology or ≥1 year follow-up served as a reference standard. The difference in sensitivity of AIBV between invasive and ductal carcinoma in situ (DCIS) as well as the association between AIBV and the diameter of invasive lesions or the histological grade were evaluated using χ(2) test. RESULTS: Pathology revealed 82 malignancies and 20 benign lesions: 70 invasive carcinomas (57 ductal, nine lobular, three mucinous, one papillary) and 12 DCIS: 10 fibroadenomas, two papillomas, two atypical ductal hyperplasias and six other benign lesions. The remaining 95 patients were negative at follow-up. Sensitivity of AIBV was 74% (61/82), specificity 94% (108/115), accuracy 86% (169/197), positive predictive value 90% (61/68) and negative predictive value 84% (108/129). Sensitivity for invasive cancers (80%; 56/70) was significantly higher than that for DCIS (42%; 5/12) (p<0.001). For invasive cancers, sensitivity was 40% (2/5) for lesions ≤9 mm in diameter, 69% (9/13) for those 10-14 mm, 79% (15/19) for those 15-19 mm and 91% (30/33) for those ≥20 mm (p<0.001). The G3 lesion rate was 49% (27/55) among true positives and only 7% (1/14) among false negatives (p=0.009). CONCLUSIONS: An association between AIBV and ipsilateral cancer exists, particularly for invasive cancers ≥20 mm or with high pathologic grade.
    La radiologia medica 08/2012; · 1.46 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: This study aimed to estimate the frequency and timing of washout in a series of pathologically proven benign mass-like breast lesions at dynamic magnetic resonance imaging. Institutional review board approval was obtained for this retrospective study. We evaluated enhancement kinetics of 33 pathologically confirmed benign breast lesions: fibroadenomas (n = 22), adenosis (n = 6), typical ductal hyperplasia (n = 2), fibroadenoma with ductal hyperplasia (n = 1), fibrosclerosis (n = 1), and inflammatory lesion (n = 1). Coronal 3-dimensional T1-weighted gradient-echo sequences were acquired before/after intravenous injection of 0.1 mmol/ kg gadoterate meglumine (time resolution, 111 seconds), 1 before and 5 after contrast injection. The time point at which the kinetic curve demonstrated a washout was recorded. Cumulative distribution of lesions showing washout was built. Paired comparisons of specificity for washout kinetics were performed using the McNemar test. Of 33 lesions, washout was never observed in 20 (61%), whereas 13 (39%) showed washout during the study. Of these 13 lesions, only 1 (inflammatory mass) exhibited washout within the first 3 minutes (specificity, 97%), 9 within 6 minutes (specificity, 73%), and 13 within 8 minutes (specificity, 61%). Specificity of washout kinetics within 3 minutes (97%) was significantly larger than that from the sixth minute (73%) and thereafter (P < 0.016). A prolonged observation for dynamic breast magnetic resonance imaging may result in false-positive washout, especially after 6 minutes. Late washout should not be considered a reliable marker of malignancy.
    Journal of computer assisted tomography 05/2012; 36(3):301-5. · 1.38 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To estimate the spatial displacement of breast lesions and nipples in MR images when the patient is moved from the standard prone to a supine position close to ultrasound (US) or surgical setting. Eleven patients underwent breast MRI in prone position with dynamic 3D T1-weighted sequences using 0.1 mmol/kg gadobenate dimeglumine. Subsequently, the patient was repositioned in supine position and a 3D volumetric interpolated breathhold examination sequence was acquired using a thoracic surface coil. For both positions we measured the following minimal distances: (A) from lesion margin to the coronal plane passing through the anterior surface of the sternum, antero-posterior, on native axial images; (B) from lesion margin to the medial sagittal plane, on native axial images, latero-medial; (C) from lesion margin to the axial plane passing through the tracheal bifurcation, cranio-caudal; (D) from lesion margin to the thoracic wall/pectoral muscle, on native axial images; (E) from lesion margin to the skin, on native axial images; (F) from lesion margin to the base of the nipple, on oblique reconstructions. Measurements from A to D were also obtained for each nipple. The prone-to-supine spatial displacement was calculated as the absolute difference between the measurement obtained in supine position and the same measurement obtained in prone position. Displacements were presented as mean ± standard deviation and median in parenthesis. Lesion displacements were (mm): A = 60 ± 38 (55); B = 40 ± 26 (41); C = 41 ± 33 (34); D = 32 ± 31 (27); E = 6 ± 5 (7); and F = 8 ± 6 (7). Nipple displacements were (mm): A = 84 ± 44 (91); B = 54 ± 24 (56); C = 27 ± 15 (24); and D = 48 ± 20 (48). These preliminary results show that preoperative breast MRI in prone position implies a median lesion displacement of about 3-6 cm along the three orthogonal directions in comparison with supine MRI. Conversely, median lesion-to-skin and lesion-to-nipple displacements were less than 1cm, even though nipple displacements were similar to or larger than those of lesions. The lesion-to-nipple distance may be the most reliable measure to be used for second look breast US. Larger studies are warranted in order to define an optimized breast MRI protocol in the preoperative setting.
    European journal of radiology 04/2012; 81(6):e771-4. · 2.65 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a radiological tool for the detection and discrimination of breast lesions. The aim of this study is to evaluate a computer-aided diagnosis (CAD) system for discriminating malignant from benign breast lesions at DCE-MRI by the combined use of morphological, kinetic, and spatiotemporal lesion features. Fifty-four malignant and 19 benign breast lesions in 51 patients were retrospectively evaluated. Images were acquired at two centers at 1.5 T. Mass-like lesions were automatically segmented after image normalization and elastic coregistration of contrast-enhanced frames. For each lesion, a set of 28 3D features were extracted: ten morphological (related to shape, margins, and internal enhancement distribution); nine kinetic (computed from signal-to-time curves); and nine spatiotemporal (related to the variation of the signal between adjacent frames). A support vector machine (SVM) was trained with feature subsets selected by a genetic search. Best subsets were composed of the most frequent features selected by majority rule. The performance was measured by receiver operator characteristics analysis with a stratified tenfold cross-validation and bootstrap method for confidence intervals. SVM training by the three separated classes of features resulted in an area under the curve (AUC) of 0.90 ± 0.04 (mean ± standard deviation), 0.87 ± 0.06, and 0.86 ± 0.06 for morphological, kinetic, and spatiotemporal feature, respectively. Combined training with all 28 features resulted in AUC of 0.96 ± 0.02 obtained with a selected feature subset composed by two morphological, one kinetic, and two spatiotemporal features. Quantitative combination of morphological, kinetic, and spatiotemporal features is feasible and provides a higher discriminating power than using the three different classes of features separately.
    Medical Physics 04/2012; 39(4):1704-15. · 2.91 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE Our aim was to estimate the prevalence of extramammary incidental findings (EMIFs) in patients undergoing breast MRI. METHOD AND MATERIALS We retrospectively reviewed 74 breast MRI studies of 73 patients, aged from 24 to 87 years (mean 55 ± 14 years) performed from June 1st, 2010 to March 15th, 2011. Studies were evaluated by a radiologist with 4-year experience mainly dedicate to breast imaging. Reported EMIFs were extracted from reports. An independent observer with 5-year experience in body CT and MRI reviewed all images in order to detect EMIFs (focused assessment). Major and minor EMIFs were distinguished. The prevalence of initially reported EMIFs was compared with that obtained with focused assessment, considered as reference standard. McNemar test was used. RESULTS Out of 74 exams, 8 EMIFs were initially reported in 6 patients. The focused assessment resulted in 34 EMIFs in 29 patients: 2 major EMIFs (1 vertebral metastasis and 1 ascending aorta ectasia) and 32 minor EMIFs (10 liver cysts, 6 disventilatory bands, 3 hiatal hernias, 2 small pleural effusions, 2 accessory spleens, 2 pulmonary atelectasias, 1 simple renal cysts, 1 liver hemangioma, 1 focal liver enhancement, 1 rib abnormality, 1 vertebral abnormality, 1 azygous ectasia, 1 abnormal lung density). All of initially reported EMIFs were found at focused assessment. Of 74 exams, 29 presented at least one EMIF with an overall prevalence of 39% (95%CI 28%-51%). The mean frequency of EMIFs per examination was 1:9.2 at initial reporting, while it was 1:2.2 at focused assessment (P<.001). Of the total 34 EMIFs, 26 (76%) were lost at initial reporting. CONCLUSION There is a highly significant difference between the rate of EMIFs initially reported in routine breast imaging practice and those found at a focused assessment by a body radiologist. CLINICAL RELEVANCE/APPLICATION Breast radiologists should pay more attention to EMIFs during routine clinical practice.
    Radiological Society of North America 2011 Scientific Assembly and Annual Meeting; 11/2011
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE p { margin-bottom: 0.21cm; } Digital breast tomosynthesis (DBT) is increasingly attracting attention as a possible adjunct in breast cancer diagnosis, yet little is still known on the level of performance of computer aided-detection systems on these kinds of images. In this work, we present the evaluation of a commercial CAD system (CAD BREAST DTS, im3D S.p.A.) on a DBT multicenter independent testing set. METHOD AND MATERIALS p { margin-bottom: 0.21cm; } In this study, a commercial CAD system was retrospectively applied to breast tomosynthesis images in 154 female patients evaluated in 3 different centers. Craniocaudal and mediolateral oblique images were acquired with a Hologic Selenia Dimensions DBT. The dataset included a total of 57 microcalcification clusters (24 of malignant histology and 33 benign). All malignant lesions were histologically confirmed, whereas benign lesions were confirmed by either histology or follow-up. Lesions for which neither was available were excluded from the analyses. Cases were annotated by an experienced mammographist by tracing a bounding box around each cluster. A CAD true positive was registered when at least two of the microcalcifications identified by CAD fell within the radiologist markings. A CAD false positive was registered in all other cases, to avoid chance matchings. CAD performance was evaluated by computing per-lesion detection rate and mean number of false positives per projection at the operating point. RESULTS p { margin-bottom: 0.21cm; } A per-lesion detection rate of 89% (51/57) at 1.7 mean false positives per projection (median: 1, IQR=[0,2]) was obtained. Sensitivity on malignant clusters was equal to 83% (20/24). CONCLUSION p { margin-bottom: 0.21cm; } The evaluated CAD system presented a good stand-alone detection rate at a fairly low false-positive rate. Further work -particularly on difficult cases, such as screening interval cancers, and comparing reading with and without CAD- is needed to understand its role in clinical practice. CLINICAL RELEVANCE/APPLICATION p { margin-bottom: 0.21cm; } A first evaluation of a commercial CAD for the detection of microcalcification clusters in DBT is presented on a multicenter independent testing set.
    Radiological Society of North America 2011 Scientific Assembly and Annual Meeting; 11/2011
  • [Show abstract] [Hide abstract]
    ABSTRACT: To describe and test a new fully automatic lesion detection system for breast DCE-MRI. Studies were collected from two institutions adopting different DCE-MRI sequences, one with and the other one without fat-saturation. The detection pipeline consists of (i) breast segmentation, to identify breast size and location; (ii) registration, to correct for patient movements; (iii) lesion detection, to extract contrast-enhanced regions using a new normalization technique based on the contrast-uptake of mammary vessels; (iv) false positive (FP) reduction, to exclude contrast-enhanced regions other than lesions. Detection rate (number of system-detected malignant and benign lesions over the total number of lesions) and sensitivity (system-detected malignant lesions over the total number of malignant lesions) were assessed. The number of FPs was also assessed. Forty-eight studies with 12 benign and 53 malignant lesions were evaluated. Median lesion diameter was 6 mm (range, 5-15 mm) for benign and 26 mm (range, 5-75 mm) for malignant lesions. Detection rate was 58/65 (89%; 95% confidence interval [CI] 79%-95%) and sensitivity was 52/53 (98%; 95% CI 90%-99%). Mammary median FPs per breast was 4 (1st-3rd quartiles 3-7.25). The system showed promising results on MR datasets obtained from different scanners producing fat-sat or non-fat-sat images with variable temporal and spatial resolution and could potentially be used for early diagnosis and staging of breast cancer to reduce reading time and to improve lesion detection. Further evaluation is needed before it may be used in clinical practice.
    Journal of Magnetic Resonance Imaging 09/2011; 34(6):1341-51. · 2.57 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To describe MRI features of fat necrosis of the breast. Twenty-five lesions in 16 patients were retrospectively analyzed. MRI was performed due to equivocal findings at conventional imaging after surgical treatment of cancer (n=14) or during anticoagulant therapy (n=1), after focal mastitis treated with ductal resection (n=1). In the 15 patients with previous surgery MRI was performed after a median interval of 24 months, using short tau inversion recovery (STIR) and contrast-enhanced dynamic T1-weighted sequences. Signal-to-noise ratio (SNR) inside the lesion and surrounding healthy fat was calculated on both STIR and unenhanced T1-weighted images. Maximal lesion diameter was measured on STIR images. All lesions had final clinical and imaging assessment in favor of fat necrosis and negative clinical and imaging follow-up (21-40 months; median 24 months). At STIR sequence, fat necrosis appeared as a "black hole", being markedly hypointense (median SNR=29) compared with surrounding fat (median SNR=95) (P<0.001), while no significant difference was found at unenhanced T1-weighted sequence. No significant correlation with time from treatment was found. Of 25 lesions, 15 showed ring enhancement, with continuous increase (n=10), plateau (n=2), or wash-out curve (n=3). The 11 enhancing lesions in the 8 patients with previous radiation therapy showed an initial enhancement higher than that of the 4 enhancing lesions in the 2 patients who did not, although the difference was not significant (P=0.104). Fat necrosis of the breast exhibits a "black hole" sign on STIR images, allowing for an easier diagnosis in clinical practice.
    European journal of radiology 07/2011; 81(4):e573-9. · 2.65 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The purpose of this article is to assess changes in breast vascular maps on dynamic contrast-enhanced MRI (DCE-MRI) after primary chemotherapy in patients with locally advanced breast cancer (LABC). Thirty-four patients with unilateral LABC underwent DCE-MRI before and after anthracycline- and taxane-based primary chemotherapy. The number of vessels 30 mm or longer in length and 2 mm or larger in maximum transverse diameter were counted on maximum intensity projections of the first subtracted phase for each of the two breasts. Patients achieving pathologic response or small clusters of residual cancer cells after primary chemotherapy were considered as responders, and those with an inferior pathologic response were considered as nonresponders. The mean (± SD) number of vessels in the breast harboring the cancer and in the contralateral breast was 2.7 ± 1.3 and 1.1 ± 1.0 (p < 0.001), respectively, before primary chemotherapy and 1.3 ± 1.1 and 1.1 ± 1.1 (p = 0.147), respectively, after primary chemotherapy. Overall, primary chemotherapy was associated with a significant reduction in DCE-MRI vascular maps in the breast harboring the cancer only (p < 0.001). Of the 34 patients, 10 were considered responders and 24 were nonresponders. The mean number of vessels in the breast harboring the cancer changed from 2.7 ± 1.1 to 0.6 ± 0.8 for the 10 responders and from 2.7 ± 1.4 to only 1.6 ± 0.9 for the 24 nonresponders. The mean reduction of vascular map in the breast harboring the cancer was significantly higher in responders compared with nonresponders (p = 0.017). Before primary chemotherapy, DCE-MRI vascular maps were asymmetrically increased ipsilaterally to the LABC. After primary chemotherapy, vascular maps significantly changed only in the breast harboring the cancer, with significant differences between responders and nonresponders.
    American Journal of Roentgenology 05/2011; 196(5):1214-8. · 2.90 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: OBJECTIVE: The purpose of this article is to review the use of gadobenate dimeglumine, a high-relaxivity gadolinium-based contrast material, for breast MRI. CONCLUSION: Thanks to its high relaxivity, gadobenate dimeglumine offers valuable advantages in terms of lesion conspicuity, detection rate, and sensitivity for malignant breast lesions. However, a higher enhancement of benign lesions should be taken into account to avoid reduced specificity.
    American Journal of Roentgenology 04/2011; 196(4):942-55. · 2.90 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: : To prospectively compare clinical breast examination, mammography, ultrasonography, and contrast-enhanced magnetic resonance imaging (MRI) in a multicenter surveillance of high-risk women. : We enrolled asymptomatic women aged ≥ 25: BRCA mutation carriers; first-degree relatives of BRCA mutation carriers, and women with strong family history of breast/ovarian cancer, including those with previous personal breast cancer. : A total of 18 centers enrolled 501 women and performed 1592 rounds (3.2 rounds/woman). Forty-nine screen-detected and 3 interval cancers were diagnosed: 44 invasive, 8 ductal carcinoma in situ; only 4 pT2 stage; 32 G3 grade. Of 39 patients explored for nodal status, 28 (72%) were negative. Incidence per year-woman resulted 3.3% overall, 2.1% <50, and 5.4% ≥ 50 years (P < 0.001), 4.3% in women with previous personal breast cancer and 2.5% in those without (P = 0.045). MRI was more sensitive (91%) than clinical breast examination (18%), mammography (50%), ultrasonography (52%), or mammography plus ultrasonography (63%) (P < 0.001). Specificity ranged 96% to 99%, positive predictive value 53% to 71%, positive likelihood ratio 24 to 52 (P not significant). MRI showed significantly better negative predictive value (99.6) and negative likelihood ratio (0.09) than those of the other modalities. At receiver operating characteristic analysis, the area under the curve of MRI (0.97) was significantly higher than that of mammography (0.83) or ultrasonography (0.82) and not significantly increased when MRI was combined with mammography and/or ultrasonography. Of 52 cancers, 16 (31%) were diagnosed only by MRI, 8 of 21 (38%) in women <50, and 8 of 31 (26%) in women ≥ 50 years of age. : MRI largely outperformed mammography, ultrasonography, and their combination for screening high-risk women below and over 50.
    Investigative radiology 02/2011; 46(2):94-105. · 4.85 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To evaluate the requests for bedside chest radiography (BCR) by clinicians. After IRB approval, we retrieved the patients' clinical records for demographics, unit, and duration for recovery in a two weeks period. For each BCR, exposure data and the reason for BCR were registered. The BCR results were categorized as one or more of the followings: negative, regular/irregular device position, known finding(s) unmodified/modified in respect with the previous BCR, new expected finding(s) (pleural effusion, low ventilation), or new unexpected findings (pulmonary edema, pneumothorax). As a utility indicator of BCR, we considered the rate of chest CT performed in these patients during the study period and the following week. We have estimated the effective radiation dose. A total of 337 BCRs (126 patients) entered the analysis, 74% of them being performed in post-surgery intensive care unit. Seventy-3 patients (58%) performed 1 or 2 BCRs, 53 (42%) 3 or more BCRs with a maximum of 13 BCRs performed on a newborn. The mean total effective dose was 0.2 mSv/patient (maximum 1 mSv). In post-surgery intensive care unit the mean daily BCR rate was 0.8/day (maximum 2/day). On 337 BCR requests, 49% showed no motivations at all, 42% reported the word "check" and in 9% a well-defined clinical query was specified. The rate of incorrect catheter position and new unexpected findings was 4%. One chest CT have been requested. BCR is often requested as a routine examination with a little rate of unexpected findings.
    European journal of radiology 02/2011; 81(3):e260-3. · 2.65 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To evaluate a computer aided diagnosis (CAD) system for discriminating malignant from benign breast lesions at dynamic contrast-enhanced(DCE) MRI METHOD AND MATERIALS We retrospectively used 73 patologically confirmed lesions (54 malignant and 19 benign) detected in 51 exams acquired at two centers at 1.5 T. The first center, acquired 26 exams with a GE scanner (Signa Excite Hdx), using a fat-sat SPGR three-dimensional (3D) sequence and 0.1 mmol/kg of Gd-DPTA (maximum 7 scans: one unenhanced; 5-6 contrast-enhanced; time resolution = 50-90 s). The second center, acquired 25 exams with a Siemens scanner (Sonata Maestro Class), using a non-fat-sat 3D FLASH sequence and 0.1 mmol/kg of Gd-BOPTA (5 scans: one unenhanced, 4 contrast-enhanced; time resolution = 120 s). Mass-like lesions were selected by two experienced radiologists and automatically segmented after image normalization and elastic co-registration of contrast-enhanced frames. For each lesion, a set of 28 3D features were extracted: 10 morphological (related to shape, margins, and internal contrast-enhancement distribution); 9 kinetic (computed from signal-to-time curves); and 9 spatio-temporal (related to the variation of the signal between adjacent frames). A support vector machine (SVM) was trained with feature subsets selected by a genetic search. Best subsets were composed of the most frequent features selected by majority rule.The performance was measured by receiver operator characteristics ROC analysis with 10 experiments of 10-fold cross-validation. RESULTS SVM training by the three separated classes of features gives area under ROC (AUROC) of 0.78±0.02(mean±standard deviation), 0.81±0.03, 0.74±0.02 for morfological, kinetic and spatio-temporal feature, respectively. Combined training with all 28 features resulted in AUROC of 0.90±0.02 obtained with a feature subset composed by 3 morphological, 1 kinetic, 1 spatio-temporal feature CONCLUSION Results suggest that the combination of selected morphological, kinetic and spatio-temporal features in CAD systems results in an accurate malignant/benign differentiation of breast mass-like lesions at DCE-MRI. CLINICAL RELEVANCE/APPLICATION The combined use of morphological, kinetic and spatio-temporal features can improve the accuracy in classifying mass-like breast lesions at dynamic-contrast enhanced MRI
    Radiological Society of North America 2010 Scientific Assembly and Annual Meeting; 11/2010
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE Our aim was to review the electronic database of our institution and verify whether the constantly raising use of breast magnetic resonance imaging (MRI) for preoperative staging of breast cancer was associated with an increasing mastectomy rate. METHOD AND MATERIALS Electronic database of our academic Breast Surgery Unit was accessed. Chi square and Spearman correlation were used. RESULTS From January 2003 to December 2008, 1,522 breast surgical procedures were performed. The clinical indication for 983 interventions was the removal of a needle-biopsy proven malignancy. Of 983 patients, 168 underwent mastectomy for extensive disease diagnosed with imaging modalities or during surgical intervention. Preoperative staging MRI was performed in 105 patients. The mastectomy rate for the total number of performed surgical treatments was 25/121 (20.7%) in 2003, 25/159 (15.7%) in 2004, 22/140 (15.7%) in 2005, 21/161 (13%) in 2006, 26/182 (14.3%) in 2007, and 49/220 (21.4%) in 2008. The preoperative MRI examinations were 3 in 2003, 2 in 2004, 13 in 2005, 14 in 2006, 31 in 2007, and 42 in 2008. The overall rate of mastectomies for the 5-years period was 168/983 (17.1%), with a stratified mastectomy rate of 37.1% for the patients who underwent preoperative MRI and 14.6% for the patients staged without MRI (P < 0.01). No significant correlation was found between mastectomy rate and number of preoperative MRI examinations performed in the 5-year period (r = 0.21). CONCLUSION We have increasingly performed MRI to improve the surgical planning of breast cancer patients with no significant increase in the total mastectomy rate, this rate remaining under 22% during the whole period. The higher mastectomy rate in patients who performed preoperative MRI can be easily explained by the preferred use of preoperative MRI in patients with a higher suspicion of extensive disease at clinical examination and conventional imaging (mammography and sonography). CLINICAL RELEVANCE/APPLICATION Preoperative MRI can be implemented in clinical practice without increasing the mastectomy rate.
    Radiological Society of North America 2010 Scientific Assembly and Annual Meeting; 11/2010
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To estimate the diagnostic value of vessels feeding breast lesions as a marker of malignancy. METHOD AND MATERIALS We reviewed a series of 94 1.5-T breast MR examinations (patients’ mean age 53±13 years) in which at least a mass-like lesion was detected and pathological examination was available from core-needle or excisional biopsy. The imaging protocol included a dynamic study with 0.1 mmol/kg of gadobenate dimeglumine using a 3D FLASH (TR/TE=11/4.76 ms, matrix 384×384 or 512×512) sequence. Using maximum intensity projection reconstructions, we counted the number of vessels feeding the lesion and the diameter of the largest of them. Spearman correlation coefficient , χ2 test and 95% confidence intervals (95% CIs) were used. RESULTS Out of 94 breast index lesions, 82 (87%) were malignant 64 invasive cancers, 18 DCIS), while the remaining 12 (13%) were fibroadenomas (n=6), papillomas (n=5), and phyllodes (n=1). At least one feeding vessel was detected in 41/82 (50% sensitivity, 95% CI 39%-61%) malignant lesions, while no feeding vessels were associated (P<.001) with the 12 benign lesions (100% specificity, 95% CI 73%-100%). Positive and negative predictive values were 41/41 (100%, 95% CI 91%-100%) and 12/53 (23%, 95% CI 12%-36%), respectively. The pathological grade of malignant lesions correlated with both the number of feeding vessels (r=0.392, P<.001) and the maximal diameter (r=0.364, P<.001). Malignant lesions without feeding vessels had a median diameter lower than those with at least a feeding vessel (15 mm versus 25 mm, P<.001). The maximal diameter of feeding vessels correlated with the lesion diameter (r=0.419, P<.001). There was no significant difference between invasive and in situ cancers in terms of both the presence of feeding vessels (P=.283) and the maximal feeding vessel diameter (P=.342). CONCLUSION While only 50% of cancers were associated with feeding vessels, breast lesions with at least one feeding vessel showed a high probability of being malignant. Moreover, the higher the histological lesion grade, the larger the number and diameter of feeding vessels. CLINICAL RELEVANCE/APPLICATION When assessing breast mass-like lesions with contrast-enhanced MRI, the presence of feeding vessels should be considered as a marker of malignancy to be added to the established BI-RADS descriptors.
    Radiological Society of North America 2010 Scientific Assembly and Annual Meeting; 11/2010

Publication Stats

103 Citations
43.42 Total Impact Points

Institutions

  • 2011–2014
    • I.R.C.C.S. Policlinico San Donato
      Milano, Lombardy, Italy
    • Institute for Cancer Research and Treatment
      Torino, Piedmont, Italy
  • 2010
    • University of Milan
      • Department of Biomedical, Surgical , and Dental Sciences
      Milano, Lombardy, Italy
  • 2009
    • Azienda Ospedaliera Universitaria San Martino di Genova
      Genova, Liguria, Italy