tangwei12@fudan.edu.cn

To compare the diagnostic efficiency of contrast-enhanced digital mammography (CEDM) plus digital mammography (DM) and magnetic resonance imaging (MRI) plus DM in symptomatic women.

Between June and December 2015, 196 patients with 240 histologically proven lesions all underwent DM, CEDM and MRI. Two radiologists were responsible for interpreting all images according to the Breast Imaging Reporting and Data System (BI-RADS). The diagnostic performance of each method was assessed by receiver-operating characteristic (ROC) curve. The sensitivity, specificity, accuracy, positive predictive value (PPV) and negative predictive value (NPV) were compared using McNemar's test and Fisher's exact test. A Kappa test was used to assess the interobserver agreement.

The diagnostic performance of CEDM and MRI combined with DM is superior to that of DM alone in symptomatic women; MRI plus DM is slightly better than that of CEDM plus DM, but this difference was not statistically significant

Besides MRI, CEDM is a good tool for breast cancer diagnosing.

jochelsm@mskcc.org

To determine whether CEDM may replace the standard practice of mammography plus targeted ultrasound in the work-up of patients with palpable breast abnormalities.

Patients presenting for evaluation of a palpable mass were consented for this prospective HIPAA-compliant and IRB-approved study between April 2015 and April 2017. 183 women with 200 palpable lesions were evaluable. The region of the palpable abnormality was marked. CEDM was performed with standard MLO and CC views plus a spot film over the palpable abnormality. Low energy images were interpreted first after which contrast enhanced images were provided. Targeted ultrasound was performed. Suspicious lesions were biopsied with appropriate imaging guidance. Any suspicious finding seen only on CEDM was evaluated by MRI for possible biopsy. If not seen on MRI, a 6-month follow up CEDM was performed. Detection rates and PPV3 for CEDM and ultrasound were calculated. Pathology and/or one-year follow up were the gold standard. Patient characteristics including age, menstrual status, breast density, breast parenchymal enhancement and risk factors were recorded.

Mean patient age was 52 (30-80 years). Eighty-three palpable lesions had no imaging correlate and 100 were cysts, fat necrosis or biopsied yielding other benign abnormalities. Seventeen malignant lesions were detected: 14/17 (82%) by CEDM and 16/17 (94%) by ultrasound. Ultrasound and CEDM missed the same cancer which was detected by MRI. PPV3 of CEDM was 42% and ultrasound 37% (p<.0001).

The cancer detection rate of CEDM in this population suggests that CEDM cannot be used to replace mammography plus targeted ultrasound in the work up of palpable abnormalities, although confirmation in a larger study is required for a definitive conclusion.

A negative CEDM did not reliably exclude breast cancer. Mammography plus targeted ultrasound should remain the standard work-up of women with palpable abnormalities.

To evaluate basic performance metrics of contrast enhanced digital mammography (CEDM) when used for breast cancer screening.

IRB approved retrospective review was performed to identify CEDMs performed for routine breast cancer screening between December 2012- April 2016. Medical records were reviewed for risk factors (family history, personal history, BRCA status, prior high risk lesion). The number of biopsies recommended and cancers detected were recorded as well as the tumor histopathologies.

A total of 1197 screening CEDMs were performed during the study period. Median age was 52 (range: 25-82). 319 (27%) women had a family history of breast cancer in a 1st degree relative <50, 470 (39%) a personal history of breast cancer, 103 (9%) a known BRCA mutation, and 371 (31%) a history of a high risk lesion. 1073 (89%) studies were given a BI-RADS 1 or 2 assessment, 28 (2%) a BI-RADS 3, and biopsy recommended in 71(6%). MRI was recommended for further evaluation of a contrast only finding in 31 (3%) women. 22 cancers were detected (12 invasive ductal cancers, 2 invasive lobular cancers, 1 invasive mammary cancer, 7 cases of DCIS) for a PPV3 of 31% and a cancer detection rate of 18/1000.

CEDM is a promising technology that may be used for breast cancer screening of intermediate risk women. The BI-RADS 3 rate, PPV3, and relative proportion of DCIS to invasive cancers are comparable to screening mammography.

CEDM may be a potential alternative screening technique for women at intermediate risk for breast cancer.

rashaakamal@hotmail.com

TThe purpose of this study is to assess the impact of including contrast mammography in the characterization of mammography identified breast asymmetries.

This prospective study included 380 patients with mammography identified breast asymmetries: single view, focal, global and developing asymmetries. Contrast enhanced spectral mammography (CESM) was performed using the same machine. A pair of low and high energy images wastaken in each position. Lesions were classified into enhancing and non-enhancing lesions. Enhancing lesions (focus, mass and non-mass) were further categorized according to the ACR MRI BIRADS lexicon. Core and surgical biopsy was taken and histopathology results were correlated with the corresponding imaging findings and accordingly the diagnostic indices of mammography alone and when adding CESM were calculated and compared.

The study included 380 mammography identified breast asymmetries: 230/380 (60.5%), focal, 128/380 (33.7%) global 20/380, (5.3%) single view and 2/380 (0.5%) developing asymmetries. Additional mammography signs (microcalcifications, distortion and skin thickening) were identified in 56/380 cases and they strongly correlated with an underlying malignant pathology (p<0.05). After contrast injection, no contrast uptake was seen in 88 cases (74 benign and 14 malignant lesions). Enhancing lesions were classified as focus (16/380, ), mass (121/380) and non-mass(155/380). The calculated sensitivity, specificity positive and negative predictive values of mammography were 82.8%, 46.8%, 66.8% and 75.7% as compared to 94.4%, 76.4%, 88.5% and 87.4% after adding CESM. False negative lesions on CESM were 14/380 non enhancing malignant lesions (mainly DCIS) while false positive cases were 30/380 inflammatory breast lesions and atypical fibroadenomas showing malignant morphology descriptors.

CESM should be incorporated in the diagnostic workup of breast asymmetries to highlight or exclude underlying malignant lesions with the exception of when an underlying inflammatory process is considered due to the high incidence of false positive results in this case.

Mammography identified breast asymmetries pose a diagnostic challenge and they often entail un necessary diagnostic and interventional procedures. CESM when added to Mammography can help in the characterization of the underlying pathology.

This work presents a method for automatic breast lesion classification in dual energy contrast enhanced spectral mammography (CESM). The aim is to assess its feasibility and to evaluate the potential for biopsy sparing in benign breast lesions.

A dataset of 130 CESM breast lesions was generated retrospectively from 65 benign and 65 malignant lesions following biopsy. Each lesion was manually contoured using standard PACS viewer drawing tools. Based on this data set, a supervised learning algorithm was developed to tell apart benign from malignant lesions. A supervised classifier was constructed using deep learning methods (DL) in conjunction with transfer learning. The transfer learning approach begins with a pre-trained neural network (AlexNet), trained on the ImageNet database, comprised of 14 million natural images from 1000 different classes. This network was further fine-tuned for our CESM data set, in order to extract strong features representing the characteristic patterns of benign and malignant lesions. A subset of 82 lesions was used to train the network (back-propagation) while 13 other provided inter-epoch validation. Eventually, the trained network was tested on the remaining set of 35 lesions (which were not used for training). A classification score was generated by the network for each lesion in the test set (see the Figure; benign -blue and malignant-red). A higher score signifies a higher probability for that lesion to be malignant. Malignant/benign classification is performed by imposing a selected threshold value on the classification score. The value of this threshold determines the sensitivity and specificity of the corresponding working point.

Setting a conservative score threshold of 0.2 (Figure), leads to Sensitivity = 0.95 and Specificity = 0.47, corresponding to a potential 47% sparing in biopsies at the cost of 1 false negative.

This work demonstrates the feasibility of automatic lesion classification in dual energy CESM images. The method has the potential to reduce the number of benign breast biopsies while preserving high sensitivity.

Automatic, or computer assisted lesion classification in CESM images has the potential to reduce the number of benign breast biopsies, thereby reducing both the patient anxiety and the medical costs.

To evaluate the diagnostic accuracy of contrast-enhanced cone-beam breast-CT (CE-CBCT) in dense breast tissue and compare it to non-contrast CBCT, mammography (MG) and magnet resonance imaging (MRI).

In this prospective ethics-board approved study, 41 women (52 breasts, 100 lesions), median age of 57.9 years (IQR 48.9-64.9, range 41.6-78.6 years) with ACR density type c or d and BI-RADS 4 or 5 assessment in MG and/ or ultrasound were included. Based on amended ACR BI-RADS criteria, MG, non-contrast CBCT, CE-CBCT and MRI were independently evaluated by two blinded readers. The area under the receiver operating curve (AUC), sensitivity and specificity were compared between the different imaging modalities. All data were evaluated by means of descriptive statistics. ANOVA-type statistics were used for comparison.

Histological examination was performed on 63 breast lesions (6 benign, 6 high-risk, 51 malignant). Follow-up imaging was performed for 37 lesions. The AUCs for breast cancer diagnosis for reader 1 and 2 were: 0.79/ 0.69 (CE-CBCT), 0.78/ 0.76 (MRI), 0.70/ 0.62 (non-contrast CBCT) and both 0.69 (MG). CE-CBCT improved breast cancer diagnosis sensitivity by 30-37% in comparison to MG, and was comparable to MRI (MG: 0.84/ 0.93 vs. MRI: 0.70/ 0.86). Associated ANOVA-type statistics for differences in AUC and sensitivity across imaging modalities were p=0.0443 and p<0.001, respectively.

This study showed that CE-CBCT can accurately identify malignant breast lesions in a diagnostic setting. CE-CBCT improved lesion detection in comparison to MG and non-contrast CBCT and was comparable to MRI in density type c and d breasts.

The results show that CE-CBCT is a promising new method that may be a suitable alternative to MRI in patients with contraindications to MRI or in regions with limited MRI availability.

dr.mahahelal@yahoo.com

to compare contrast enhanced spectral mammography and automated breast ultrasound with regards to cancer size, extension and multiplicity in reported cases with breast cancer

Institutional review board approval and patient informed consent were obtained from 70 patients with proved breast cancer (64% invasive ductal grade II, 16% invasive ductal grade III and 20% invasive lobular). 3D ABUS were done for anteroposterior ;lateral and medial acquisitions and CESM performed using low and high energy exposures in craniocaudal and mediolateral oblique views after IV injection of contrast agent. Included breast cancers were analyzed regarding disease extension, size, and multiplicity.Operative data was the gold standard reference

CESM and 3D ABUS showed comparable measurements of the cancer size and their accuracy was 95% and 93% respectively. ABUS was superior in evaluating cancer extension to the surroundings (parenchyma +/- skin) with an accuracy of 90% compared to 87% for CESM. Multiplicity was better demonstrated by CESM that showed an accuracy of 96% compared to 90% by ABUS. The overall performance of CESM in staging of breast cancer was 95.7% sensitivity, 88% specificity and 93% total accuracy versus 95%, 76.5% and 91.3% respectively for ABUS.

3D ABUS is a non-invasive alternative tool to CESM in the staging of breast cancer. ABUS was non-inferior to CESM in the evaluation of cancer size. 3D ABUS is more accurate in determining parenchymal infiltration and CESM is better in detection of multiplicity.

ABUS is considered a revolution in breast scanning by ultrasound imaging that can be used as a non-invasive, fast and easy tool of breast imaging in early detection (in other words; screening) and staging of breast carcinomas

jsumkin@upmc.edu

To compare MRI, MBI and CEM imaging assessments of index lesions and extent of disease with pathology in patients with newly diagnosed breast cancer.

We compared results from 79 women with biopsy-proven breast cancers in 80 breasts who underwent, under an IRB-approved protocol, MRI, MBI and CEM within one week. Six specifically trained, breast radiologists participated in the study. Each examination was independently interpreted prospectively by one radiologist with full access to all prior breast imaging, but blinded to the interpretations of the two other study modalities. Findings for the 80 cancers were recorded and compared in a side by side review/correlation with surgical pathology (59 cases) and in neo-adjuvant treated cases (21) with image-guided biopsy results.

The primary cancers consisted of 40 IDC, 17 IDC/DCIS, 18 ILC, and 4 DCIS and 1 IDC/ILC only. Segmental mastectomy or mastectomy was performed on 57 women, (58 breasts) without neo-adjuvant therapy and 51 of these were detected by all three modalities. The average size by pathology of the primary index cancers was 31 mm and were measured by the radiologists on imaging as 33 mm, 29 mm and 32 mm for MRI, CEM and MBI, respectively (p=0.17). The range in imaging sizes for primary index cancers was 3-95 mm by pathology. Four verified index lesions were missed on CEM, four on MBI and three on MRI. 7 cases were correctly upstaged from an original diagnosis of unifocal to multi-focal (3 MRI, 2 CEM, and 3 MBI) and multi-centric (2 MRI, 2 CEM, and 1 MBI). MRI depicted 42 false positive enhancing regions, CESM 16, and MBI 16 (p<0.01). A radiologist rendered a report considering all imaging studies that led in 6 cases to changes in surgical management based on the additional information provided. Of these, 2 were affected by all three modalities, while 1 primarily by MRI, 3 primarily by CEM, and none primarily by MBI.

All three modalities performed comparably in terms of sensitivity and in estimating tumor size, as compared with surgical pathology. MRI was slightly more sensitive than CEM and MBI, but had significantly more false positives than CEM and MBI.

Breast MRI is extremely sensitive for cancer staging, but has more false positive findings. Access and expense issues remain. MBI and CEM are possible alternatives to MRI for this purpose.