Multi-Institutional Assessment and Crowdsourcing Evaluation of Deep Learning for Automated Classification of Breast Density

OBJECTIVE

We developed deep learning algorithms to automatically assess BI-RADS breast density.

METHODS

Using a large multi-institution patient cohort of 108,230 digital screening mammograms from the Digital Mammographic Imaging Screening Trial, we investigated the effect of data, model, and training parameters on overall model performance and provided crowdsourcing evaluation from the attendees of the ACR 2019 Annual Meeting.

RESULTS

Our best-performing algorithm achieved good agreement with radiologists who were qualified interpreters of mammograms, with a four-class κ of 0.667. When training was performed with randomly sampled images from the data set versus sampling equal number of images from each density category, the model predictions were biased away from the low-prevalence categories such as extremely dense breasts. The net result was an increase in sensitivity and a decrease in specificity for predicting dense breasts for equal class compared with random sampling. We also found that the performance of the model degrades when we evaluate on digital mammography data formats that differ from the one that we trained on, emphasizing the importance of multi-institutional training sets. Lastly, we showed that crowdsourced annotations, including those from attendees who routinely read mammograms, had higher agreement with our algorithm than with the original interpreting radiologists.

CONCLUSION

We demonstrated the possible parameters that can influence the performance of the model and how crowdsourcing can be used for evaluation. This study was performed in tandem with the development of the ACR AI-LAB, a platform for democratizing artificial intelligence.

Investigation of Synthetic Relaxometry and Diffusion Measures in the Differentiation of Benign and Malignant Breast Lesions as Compared to BI-RADS

BACKGROUND

Breast cancer is the most common malignant tumor in women and a quantitative contrast-free method is highly desirable for its diagnosis.

PURPOSE

To investigate the performance of quantitative MRI in differentiating malignant from benign breast lesions and to compare with the Breast Imaging Reporting and Data System (BI-RADS).

STUDY TYPE

Retrospective.

SUBJECTS

Eighty patients (56 with malignant lesions and 24 with benign lesions).

FIELD STRENGTH/SEQUENCE

Diffusion-weighted imaging (DWI) with a single-shot echo planar sequence and synthetic MRI with magnetic resonance image compilation (MAGiC) were performed at 3T.

ASSESSMENT

T₁ relaxation time (T₁ ), T₂ relaxation time (T₂ ), and proton density (PD) from synthetic MRI and apparent diffusion coefficient (ADC) from DWI were analyzed by two radiologists (Reader A, Reader B). Univariable and multivariable models were developed to optimize differentiation between malignant and benign lesions and their performances compared to BI-RADS.

STATISTICAL TESTS

The diagnostic performance was evaluated using multivariate logistic regression analysis and area under the receiver operating characteristic (ROC) curves (AUC).

RESULTS

T₂ , PD, and ADC values for malignant lesions were significantly lower than those in benign breast lesions for both radiologists (all P < 0.05). The combined T₂ , PD, and ADC model had the best performance for differentiating malignant and benign lesions with AUC, sensitivity, specificity, positive predictive value, and negative predictive values of 0.904, 94.6%, 87.5%, 94.6%, and 87.5%, respectively. The corresponding results for BI-RADS were no AUC, 94.6%, 75.0%, 89.8%, and 85.7%, respectively.

DATA CONCLUSION

The approach that combined synthetic MRI and DWI outperformed BI-RADS in the differential diagnosis of malignant and benign breast lesions and was achieved without contrast agents. This approach may serve as an alternative and effective strategy for the improvement of breast lesion differentiation.

LEVEL OF EVIDENCE

3.

TECHNICAL EFFICACY STAGE

3.

A novel approach with dual-sampling convolutional neural network for ultrasound image classification of breast tumors

Breast cancer is one of the leading causes of female cancer deaths. Early diagnosis with prophylactic may improve the patients' prognosis. So far ultrasound (US) imaging is a popular method in breast cancer diagnosis. However, its accuracy is bounded to traditional handcrafted feature methods and expertise. A novel method named Dual-Sampling Convolutional Neural Networks (DSCNN) was proposed in this paper for the differential diagnosis of breast tumors based on US images. Combining traditional convolutional and residual networks, DSCNN prevented gradient disappearance and degradation. The prediction accuracy was increased by the parallel dual-sampling structure, which can effectively extract potential features from US images. Compared with other advanced deep learning methods and traditional handcraftedfeaturemethods,DSCNNreachedthebestperformance withanaccuracyof91.67%andan AUC of 0.939. The robustness of the proposed method was also verified by using a public dataset. Moreover, DSCNN was compared with evaluation from three radiologists utilizing US-BI-RADS lexicon categories for overall breast tumors assessment. The result demonstrated that the prediction sensitivity, specificity and accuracy of the DSCNN were higher than those of the radiologist with 10- year experience, suggesting that the DSCNN has the potential to help doctors make judgement in clinic.

Association Between Breast and Thyroid Lesions: A Cross-Sectional Study Based on Ultrasonography Screening in China

Introduction: Prior research has shown an association between breast and thyroid cancers, although their relationship is unclear. In China, asymptomatic women undergoing regular health checkups usually undergo breast and thyroid ultrasonography screening. The present cross-sectional ultrasound-based study estimated the prevalence of breast masses (BM) and thyroid nodules (TN) and their relationship among a population-based cohort of Chinese women. Methods: This study included 34,184 consecutive asymptomatic Chinese women who underwent both breast and thyroid ultrasound evaluation during one health care examination. Detected lesions were assigned into categories of different malignant risks according to the Breast and Thyroid Imaging Reporting and Data System (BI-/TI-RADS). Binomial logistic regression was used to determine the association between occurrence of BM and TN, and multinomial logistic regression was used to analyze the correlation of BM and TN in different BI-/TI-RADS categories. Associations between BM and TN, as well as anthropometric and biochemical markers, were also explored. Results: Of those enrolled, 6371 (18.6%) had BM, 12,153 (35.6%) had TN, and 2279 (6.7%) had both. After adjusting for age, body mass index (BMI), and height, females with TN had a higher risk of BM (odds ratio [OR] = 1.151, 95% confidence interval [CI 1.081-1.225], p < 0.0001) than those with normal thyroids, and females with BM had a higher risk of TN (OR = 1.165 [CI 1.096-1.238], p < 0.0001) than those without BM. Women with a TN >10 mm (OR = 1.249 [CI 1.104-1.413], p = 0.0004) and those with a TN ≤10 mm (OR = 1.134 [CI 1.062-1.211], p = 0.0002) were at higher risk of BM compared with those with normal thyroids. As RADS categories increased, so did the correlation between BM and TN. The increased risk of TN was associated with a higher BMI, height, systolic blood pressure, and a lower plasma albumin level. The increased risk of BM was associated with a lower BMI, plasma albumin levels, and higher height. Conclusions: A high prevalence of BM and TN was detected by ultrasonography screening in this cohort of Chinese women. These lesions occurred frequently and simultaneously, particularly in women with lesions in higher RADS categories.

Added value of strain elastography in the characterisation of breast lesions: A prospective study

BACKGROUND

Elastography is a new promising ultrasonographic technique which is used to differentiate benign and malignant breast lesions based on the stiffness of the lesion.

PURPOSE

To determine the role of strain elastography in characterisation of breast lesions and to compare the diagnostic performances of strain elastography and conventional ultrasound (US).

METHODS

In total, 113 breast lesions in 100 women were prospectively evaluated by US and strain elastography followed by the histopathological examination. Elastography score based on the Tsukuba colour scale and strain ratio were determined for each lesion. The sensitivity, specificity, accuracy, positive predictive value and negative predictive value were calculated for each modality and the diagnostic performances were compared. The best cut-off point was calculated for each of the elastography parameters using the receiver operator curve analysis.

RESULTS

Out of the 113 lesions, 40 were malignant (35.4%) and 73 were benign (64.6%). The area under the curve for elastography score showed significant difference with that of US: 0.98 versus 0.90 (Difference = 0.08, p =0.02). The elastography parameters were more specific as compared to US (ES-95 and SR-93% vs. 63%, p < 0.05) with a high negative predictive value. The combined use of elastography and US gave better results with 95% sensitivity, 94% specificity, 94% accuracy and negative predictive value reaching 97%.

CONCLUSION

Strain elastography is a useful adjunct to conventional ultrasonography. The combined use of strain elastography and ultrasound improves the characterisation of breast lesions and helps in down-staging of assigned BI-RADS category, thereby avoiding unnecessary biopsies. ES is the most useful elastography parameter to differentiate between benign and malignant breast lesions.

Can strain US-elastography with strain ratio (SRE) improve the diagnostic accuracy in the assessment of breast lesions? Preliminary results

PURPOSE

To assess the diagnostic value of strain ratio elastography (SRE), a semiquantitative elastosonographic method based on the displacement of the tissue from an external source (manual compression with the transducer), as compared and in combination with conventional ultrasound for the differentiation of breast lesions.

METHODS

One hundred and eighty-two patients with breast lesions consecutively underwent B-mode, color Doppler US, and strain US-elastography. Each lesion was classified according to the BI-RADS lexicon by evaluating the size, the B-mode, and color Doppler features and then evaluated by SRE. Histology proven by biopsy was used as the gold standard and the patients with malignant lesions subsequently underwent operations. The diagnostic performance of each method was assessed with 2 × 2 contingency tables and ROC curve analysis. To maximize the SRE sensitivity and specificity, the SRE cut-off value was calculated using the Youden test.

RESULTS

Histological examination revealed 66 benign and 116 malignant breast lesions. The conventional ultrasound showed sensitivity and specificity for the differentiation of benign and malignant lesions of 86.2% and 75.8%, respectively. Similar results were found for strain US-elastography with a cut-off of 2.49, with sensitivity and specificity of 89.7% and 72.7%, respectively. The association of conventional ultrasound with the SRE value increased the sensitivity (98.3%) but decreased the specificity compared with conventional US alone (63.6%).

CONCLUSION

Strain US-elastography can be associated with BI-RADS US examination. According to our preliminary results, it helped increase the sensitivity although it decreased the specificity. However, further multicenter studies on a larger population are warranted.

Image evaluation and breast density categories as a function of mammary positioning in full-field digital mammography

BACKGROUND

Appropriate mammary positioning is an important factor in optimizing image quality in mammography (MMG).

PURPOSE

To study the correlation of quality criteria and breast density classification proposed by the American College of Radiology (ACR) and European Guidelines and its influence to achieve a proper positioning, therefore an adequate MMG.

MATERIAL AND METHODS

A total of 128 routine MMG examinations were reviewed for the definition of breast composition parenchyma and assessment of several quality criteria proposed by the ACR and European Guidelines to achieve an adequate MMG. Adequate MMG was defined as a difference between the posterior nipple line (PNL), difference of the mediolateral oblique (MLO) and craniocaudal (CC) incidences > 1 cm. The quality criteria were analyzed as a function of correlation coefficient in order to evaluate the individual impact of each factor and analysis of variance (ANOVA) for all criteria. Receiver operating characteristic (ROC) curves were plotted to evaluate the performance of the criteria on each type of parenchyma.

RESULTS

Negative correlation of fatty breasts and visibility of the mammary angle, a greater number of skin folds and PNL > 1 cm (r < 0). Dense MMG presented less visibility of the lateral tissue compared with other categories. Area under the curve of ROC analysis revealed values of 53.1% and 54.7% for the right and left breasts, respectively.

CONCLUSION

Several factors influenced in the MMG process, but we find that breast parenchyma has a substantial role in affecting these criteria and therefore a correct position for diagnosis, which could compromise MMG diagnostic performance.

S-Detect characterization of focal solid breast lesions: a prospective analysis of inter-reader agreement for US BI-RADS descriptors

BACKGROUND

To assess inter-reader agreement for US BI-RADS descriptors using S-Detect: a computer-guided decision-making software assisting in US morphologic analysis.

METHODS

73 solid focal breast lesions (FBLs) (mean size: 15.9 mm) in 73 consecutive women (mean age: 51 years) detected at US were randomly and independently assessed according to the BI-RADS US lexicon, without and with S-Detect, by five independent reviewers. US-guided core-biopsy and 24-month follow-up were considered as standard of reference. Kappa statistics were calculated to assess inter-operator agreement, between the baseline and after S-Detect evaluation. Agreement was graded as poor (≤ 0.20), moderate (0.21-0.40), fair (0.41-0.60), good (0.61-0.80), or very good (0.81-1.00).

RESULTS

33/73 (45.2%) FBLs were malignant and 40/73 (54.8%) FBLs were benign. A statistically significant improvement of inter-reader agreement from fair to good with the use of S-Detect was observed for shape (from 0.421 to 0.612) and orientation (from 0.417 to 0.7) (p < 0.0001) and from moderate to fair for margin (from 0.204 to 0.482) and posterior features (from 0.286 to 0.522) (p < 0.0001). At baseline analysis isoechoic (0.0485) and heterogeneous (0.1978) echo pattern, microlobulated (0.1161) angular (0.1204) and spiculated (0.1692) margins and combined pattern (0.1549) for posterior features showed the worst agreement rate (poor). After S-Detect evaluation, all variables but isoechoic pattern showed an agreement class upgrade with a statistically significant improvement of inter-reader agreement (p < 0.0001).

CONCLUSIONS

S-Detect significantly improved inter-reader agreement in the assessment of FBLs according to the BI-RADS US lexicon but evaluation of margin and echo pattern needs to be further improved, particularly isoechoic pattern.

Subjective Versus Quantitative Methods of Assessing Breast Density

In order to find a consistent, simple and time-efficient method of assessing mammographic breast density (MBD), different methods of assessing density comparing subjective, quantitative, semi-subjective and semi-quantitative methods were investigated. Subjective MBD of anonymized mammographic cases (n = 250) from a national breast-screening programme was rated by 49 radiologists from two countries (UK and USA) who were voluntarily recruited. Quantitatively, three measurement methods, namely VOLPARA, Hand Delineation (HD) and ImageJ (IJ) were used to calculate breast density using the same set of cases, however, for VOLPARA only mammographic cases (n = 122) with full raw digital data were included. The agreement level between methods was analysed using weighted kappa test. Agreement between UK and USA radiologists and VOLPARA varied from moderate (κw = 0.589) to substantial (κw = 0.639), respectively. The levels of agreement between USA, UK radiologists, VOLPARA with IJ were substantial (κw = 0.752, 0.768, 0.603), and with HD the levels of agreement varied from moderate to substantial (κw = 0.632, 0.680, 0.597), respectively. This study found that there is variability between subjective and objective MBD assessment methods, internationally. These results will add to the evidence base, emphasising the need for consistent, simple and time-efficient MBD assessment methods. Additionally, the quickest method to assess density is the subjective assessment, followed by VOLPARA, which is compatible with a busy clinical setting. Moreover, the use of a more limited two-scale system improves agreement levels and could help minimise any potential country bias.

A phased approach to implementing the Breast Imaging Reporting and Data System (BI-RADS) in low-income and middle-income countries

BACKGROUND

Successful breast cancer detection programs rely on standardized reporting and interpreting systems, such as the Breast Imaging Reporting and Data System (BI-RADS), to improve system performance. In low-income and middle-income countries, evolving diagnostic programs have insufficient resources to either fully implement BI-RADS or to periodically evaluate the program's performance, which is a necessary component of BI-RADS. This leads to inconsistent breast ultrasound interpretation and a failure to improve performance.

METHODS

The authors applied the Breast Health Global Initiative's phased implementation strategy to implement diagnostic ultrasound and BI-RADS within the context of a limited-resource setting.

RESULTS

The authors recommended starting with triage ultrasound to distinguish suspicious masses from normal breast tissue and benign masses such as cysts because the majority of health workers performing ultrasounds at this level have minimal breast imaging experience. Transitioning to full diagnostic ultrasound with condensed or full BI-RADS should occur after performance and quality metrics have been met.

CONCLUSIONS

Transitioning through these phases across facilities likely will occur at different times, particularly in rural versus urban settings.

Value of Hand-held Ultrasound in the Differential Diagnosis and Accurate Breast Imaging Reporting and Data System Subclassification of Complex Cystic and Solid Breast Lesions

To explore the value of hand-held ultrasound (HHUS) for diagnosing complex cystic and solid breast lesions, 472 pathologically proven lesions were analyzed. The lesions were divided into four types based on ultrasound features. Positive predictive values (PPVs) for lesion types and risk factor performances were assessed. Furthermore, HHUS and mammography (MAM) performances were compared: 27 lesions missed in MAM were detected in HHUS. Ultrasound feature analysis revealed higher PPVs for type III and IV lesions than for type I and II lesions. In patients older than 51 y, a type III or IV lesion with a diameter greater than 18 mm, an irregular shape, a non-parallel orientation, an uncircumscribed margin, calcification, vascularity and abnormal axillary lymph nodes were suggestive of malignancy; the area under the curve reached 0.869. Thus, ultrasound is useful in diagnosing complex cystic and solid breast lesions, which should be categorized as Breast Imaging Reporting and Data System (BI-RADS) 4B or 4C.

Comparison of Qualitative and Volumetric Assessments of Breast Density and Analyses of Breast Compression Parameters and Breast Volume of Women in Bahcesehir Mammography Screening Project

Objective

We aimed to compare visual and quantitative measurements of breast density and to reveal the density profile with compression characteristics.

Materials and Methods

Screening mammograms of 1399 women between May 2014 and May 2015 were evaluated by using Volpara 4^th and 5^th version. First 379 mammograms were assessed according to ACR BI-RADS 4^th edition and compared to Volpara. We categorized the breast density in two subgroups as dens or non-dens. Two radiologists reviewed the images in consensus. Agreement level between visual and volumetric methods and volumetric methods between themselves assessed using weighted kappa statistics. Volpara data such as fibroglandular volume (FGV), breast volume (BV), compression thickness (CT), compression force (CF), compression pressure (CP) were also analyzed with relation to the age.

Results

1399 mammograms were distributed as follows: 12.7% VDG1, 39.3% VDG2, 34.1% VDG3, 13.9% VDG4 according to the 4th edition of Volpara; 1.2% VDG1, 46% VDG2, 36.8% VDG3, 15.9% VDG4 according to the 5^th edition of Volpara. The difference between two editions was 4.7% increase in dense category. 379 mammograms, according to ACR BI-RADS 4^th edition, were distributed as follows: 25.9% category A, 50.9% category B, 19.8% category C, 3.4% category D. The strength of agreement between the Volpara 4^th and 5^th editions was found substantial (k=0.726). The agreements between visual assessment and both Volpara editions were poor (k=-0.413, k=-0.399 respectively). There was a 142% increase in dense group with the VDG 4^th edition and 162% with the VDG 5^th edition when compared to visual assessment. Compression force decreased while compression pressure increased with increasing Volpara Density Grade (VDG) (p for trend <0.001 for both). Compression thickness and breast volume decreased with increasing VDG (p for trend <0.001 for both). The FGV decreases with age and the breast volume increases with increasing age (p<0.001).

Conclusion

Visual assessment of breast density doesn't correlate well with volumetric assessments. Obtaining additional information about physical parameters and breast profile by the results of quantified methods is important for breast cancer risk assessments and prevention strategies.

Can second-look ultrasound downgrade MRI-detected lesions? A retrospective study

PURPOSE

To determine whether MRI-detected suspicious (BIRADS 4 & 5) breast lesions can be downgraded using second-look ultrasound (SLU) and thus reduce unnecessarily performed breast biopsies.

MATERIALS METHODS

A retrospective single-center review of consecutive patients, who underwent breast MRI studies during a 12-month time period was performed. 94 patients with 103 lesions undergoing SLU of incidentally detected MRI BI-RADS 4&5 lesions which were not identified on previous ultrasound were included in the study. The SLU detection rate and SLU features of the lesions were assessed. Histology (91/103) or two year follow up (n = 12) were defined as the reference standard for lesion diagnosis.

RESULTS

57 (55.3 %) of the 103 lesions were identified on SLU. 17 of the identified lesions were malignant (29.8 %). Lesions detected on ultrasound presented on MRI as masses in 66.7 % (38/57) and non-mass in 33.3 % (19/57). Our findings showed that it is possible to distinguish between malignant and benign lesions with SLU. The results were significant (p < 0.05) for the following morphological features: shape, orientation, margins, architectural distortion, hyperechoic rim/ edema. All lesions classified as SLU BI-RADS 2 in our study were benign and thus, 30 % of all unnecessary biopsies could potentially have been avoided. Including SLU BI-RADS 3 lesions, this rate increased to 60 %, while yielding one (of 17, 5.8 %) false negative result. All three BI-RADS 5 lesions detected by SLU presented as malignant on ultrasound.

CONCLUSION

SLU can potentially downgrade incidental MRI BIRADS 4 lesions. This may reduce the number of unnecessarily performed biopsies by 30-60 %, thus simplifying patient management.

Both a biopsy method and a therapeutic procedure in BI-RADS 4A and 4B lesions: Ultrasound-guided vacuum-assisted breast biopsy

Objectives

This study aimed to evaluate outcomes, complications, and follow-up results of ultrasound-guided vacuum-assisted breast biopsy (UG-VABB) in BI-RADS 4 A and B lesions.

Material and Methods

Between Agust 2014 to January 2018, fifty BI-RADS 4A and BI-RADS 4B lesions of 41 patients biopsied with 10G vacuum needle by a single radiologist were retrospectively evaluated.

Results

All patients were females and mean age of the 41 patients was 50.12 ± 8.63. Of all lesions, 84% was benign, 6% was ADH, 4% was in-situ cancer, and 6% was diagnosed as malign. Follow-up duration after VABB was 0-51 months and mean was 20.92 months. Complications were as vasovagal-induced seizure in 3 patients (7.3%) and intramammary hematoma in 16 patients (39%). Hematoma was diagnosed in 3 patients (7.3%) at the 6th month follow-up and it was resolved in all patients at the 12th month follow-up. Higher breast density resulted in higher hematoma rates. There was no relationship between lesion BI-RADS subgroups, lesion size or sample number and hematoma development. During the follow-up, residue lesion in 1 (2.4%) patient and scar tissue in 2 (4.9%) patients was detected.

Conclusion

US-guided VABB, with low complication rates and low scar development, is also a therapeutic excision method without remaining residue, which should be primarily preferred in smaller than 2 cm BI-RADS 4A and 4B lesions whose malignancy rates are relatively low. Hematoma, which is the most frequent complication, resorbed entirely in the 12th month in all patients.

Image Quality and Performance Benchmarks in Vehicle and Hospital Mammography

BACKGROUND

Whether the quality and clinical performance of mammograms obtained in vehicles and those obtained in fixed facilities are equal remains unknown. We compared the characteristics of examinees screened in hospital and vehicle settings.

PATIENTS AND METHODS

Data from women who had undergone mammography at Shuang Ho Hospital from January 1, 2013, to December 31, 2016, were obtained from the Women's Breast Screening Database and used for analysis. The records revealed that 43,807 and 11,955 women had undergone mammography in vehicle and hospital settings, respectively. The performance benchmarks, including recall rate, cancer detection rate, and positive predictive value, in the 2 settings were compared. In addition, the image quality was compared by reviewing 110 records from each setting.

RESULTS

The hospital mammograms had greater subtotal mean scores (189.2 ± 5.9) compared with the vehicle mammograms (185.5 ± 7.7; P < .0001) in the mediolateral oblique view. Mobile mammography contributed to a lower odds ratio of classification in the Breast Imaging Reporting and Data System categories of 0, 4, and 5. In general, all performance benchmarks, including the cancer detection rate and positive predictive value of mobile and hospital mammography, were satisfactory. However, the recall rate with the hospital mammography service was slightly greater than the acceptable benchmark.

CONCLUSION

Mobile mammography services should be continued with improvements in image quality. The reduction in the number of patients with a category of 0 in the classification system in both mammography service settings and the enhancement of data linking to previous mammograms warrants additional attention.

S-Detect characterization of focal breast lesions according to the US BI RADS lexicon: a pictorial essay

High-resolution ultrasonography (US) is a valuable tool in breast imaging. Nevertheless, US is an operator-dependent technique: to overcome this issue, the American College of Radiology (ACR) has developed the breast imaging-reporting and data system (BI-RADS) US lexicon. Despite this effort, the variability in the assessment of focal breast lesions (FBLs) with the use of BI-RADS US lexicon is still an issue. Within this framework, evidence shows that computer-aided image analysis may be effective in improving the radiologist's assessment of FBLs. In particular, S-Detect is a newly developed image-analytic computer program that provides assistance in morphologic analysis of FBLs seen on US according to the BI-RADS US lexicon. This pictorial essay describes state-of-the-art of sonographic characterization of FBLs by using S-Detect.

Density of breast: An independent risk factor for developing breast cancer, a prospective study at two premium breast centers

Background

Breast cancer is the leading cause of death among women worldwide. Studies have identified breast density as a controversial risk factor of breast cancer. Moreover, studies found that breast density reduction through Tamoxifen could reduce risk of breast cancer significantly. To date, no study on the association between breast density and breast cancer has been carried out in Malaysia. If breast density is proven to be a risk factor of breast cancer, intervention could be carried out to reduce breast cancer risk through breast density reduction.

Purpose

To determine if density of breast is an independent risk factor which will contribute to development of breast cancer.

Materials and Methods

A prospective cohort study is carried out in two hospitals targeting adult female patients who presented to the Breast Clinic with symptoms suspicious of breast cancer. Participants recruited were investigated for breast cancer based on their symptoms. Breast density assessed from mammogram was correlated with tissue biopsy results and final diagnosis of benign or malignant breast disease.

Results

Participants with dense breasts showed 29% increased risk of breast cancer when compared to those with almost entirely fatty breasts (odds ratio [OR] 1.29, 95% CI 0.38‐4.44, P = .683). Among the postmenopausal women, those with dense breasts were 3.1 times more likely to develop breast cancer compared with those with fatty breasts (OR 3.125, 95% CI 0.72‐13.64, P = .13). Moreover, the chance of developing breast cancer increases with age (OR 1.046, 95% CI 1.003‐1.090, P < .05). In contrast, the density of breast decreases with increasing age (P < .05) and body mass index (P = .051). The proportion of high breast density whether in the whole sample size, premenopausal, or postmenopausal group was consistently high.

Conclusion

Although results were not statistically significant, important association between breast density and risk of breast cancer cannot be ruled out. The study is limited by a small sample size and subjective assessment of breast density. More studies are required to reconcile the differences between studies of contrasting evidence.

Complex Solid and Cystic Breast Cancer: A Series of Six Case Reports

A cyst in the breast containing a thick wall, internal septations, or a solid intracystic component is defined as a complex solid and cystic breast mass. These lesions carry a malignant potential between 23-31% and thus require further evaluation with biopsy [1]. We report six cases in which patients were found to have a complex solid and cystic mass, all of which were proven to be malignant breast cancers of varying etiologies. We also review the literature on malignant etiologies of complex solid and cystic breast masses, including their clinical presentation, work-up, histopathologic and immunochemistry findings, treatment, and prognosis.

The Utility of Short-Interval Follow-Up for Baseline High-Risk Screening Breast MRI

OBJECTIVE

Patients with a high lifetime risk of developing breast cancer undergo annual screening magnetic resonance imaging (MRI) starting at age 30. A proportion of these patients are subsequently required to undergo short-interval follow-up MRI 6 months after their baseline examination. The purpose of this study was to evaluate the utility and outcome of these short-interval follow-up examinations in this population.

METHODS

A retrospective review was performed of 523 patients who received their baseline high-risk breast screening MRI at our institution between 2013 and 2017. The proportion of patients who received a short-interval follow-up MRI at 6 months was recorded. The findings at baseline and follow-up MRI were recorded, as well as the outcomes and results of any interventions performed.

RESULTS

Ninety-six (17.6%) patients (age range: 25-67, mean age: 41) received a short-interval follow-up MRI following their baseline screening examination. Indications for follow-up included moderate to marked background parenchymal enhancement, nonmass enhancement, and likely benign enhancing mass. Of the 92 patients, 5 (5.4%) went on to have a biopsy, with none revealing a malignant pathology. The remainder either returned to routine screening (91.3%) or received further imaging in the form of ultrasound or additional follow-up MRI (3.3%).

CONCLUSION

Short-interval follow-up breast MRI in high-risk patients after a baseline screening study with likely benign findings is unlikely to yield clinically significant findings. This retrospective study can be considered a starting point for additional future work looking at the rate, indications, and yield of short-interval follow-up following baseline high-risk screening breast MRI studies.

The Diagnostic Value of 3D Power Doppler Ultrasound Combined With VOCAL in the Vascular Distribution of Breast Masses

RATIONALE AND OBJECTIVES

This study uses a three-dimensional energy Doppler technique combined with the Virtual Organ Computer-aided Analysis (VOCAL) method in order to determine the diagnostic threshold of blood flow index in breast tumors to provide a reference for evaluation and treatment options.

MATERIALS AND METHODS

We collected 322 solid lesions which had been operated. Each lesion met the definite pathological diagnosis; collected lesions included 262 cases of benign lesions and 60 cases of malignant lesions. All examinations were performed by using GE LOGIQ E9 with VOCAL software. Volume and four distinct vascular indices of gray mean (MG), power mean, ratio (R), and vascular flow index (VFI) were calculated by using the VOCAL software. Sampling and calculation were repeated three times and the mean value was calculated.

RESULTS

The average age and power of the malignant group were greater than those of the benign group, ie p < .01 which had significant differences. The gray mean of the malignant group was lower than that of the benign group, ie p > .05 which had no significant differences between benign and malignant groups. The ratio, vascular flow index and volume had significant differences, i.e. p < .01. The area under the receiver operating characteristic curve (AUC) were 0.864, 0.830, 0.800, 0.758, and 0.764 for age, power, ratio, vascular flow index, and volume, respectively. The research indicators were higher than 50% of the curve showing their diagnostic value. The cut-off points of age, power, ratio, vascular flow index, and volume were 37.5, 26.56, 0.031, 0.846, and 1.75, respectively. Their corresponding sensitivity were 93.3%, 75%, 81.7%, 68.3%, 63.3%, and the specificity were 68.7%, 81%, 70.2%, 75.6%, and 81.7%, respectively. Comparison of vascular indices combined with the Breast imaging reporting and data System (BI-RADS) score and simple BI-RADS method, the AUC of power + BI-RADS, ratio + BI-RADS, VFI + BI-RADS, and BI-RADS alone are 0.928, 0.903, 0.895, and 0.796, respectively, which were higher than 50% of the curve. Sensitivity was 81.7%, 80%, 88.3%, 86.7%, and specificity was 88.5%, 85.5%, 77.1%, 69.5%, respectively. The power + BI-RADS method has the highest AUC among these three methods.

CONCLUSIONS

Quantitative measurement of blood flow and blood vessel distribution in breast tumors by three-dimensional power Doppler ultrasound combined with the VOCAL method is more accurate and sensitive than the traditional two-dimensional ultrasound. And this method has potential promising applications in many current active research areas, such as the studies of random distribution of intratumoral blood vessels or the normalization of tumor blood vessels. Three-dimensional power Doppler ultrasound combined with the VOCAL method provides a new approach to achieving accurate judgments and the method evaluates the curative effect in breast cancer patients.

Pediatric and Adolescent Surgical Breast Clinic: Preliminary Experience

STUDY OBJECTIVE

The diagnostic and treatment tools used in breast centers are largely geared to adults, and there is little consideration of the unique characteristics of breast diseases in younger age groups. Herein we report on the preliminary experience of a specialized breast clinic for children and adolescents.

DESIGN

Retrospective, observational.

SETTING

Pediatric Surgical Breast Clinic of Schneider Children's Medical Center of Israel.

PARTICIPANTS

Patients referred to the clinic during the first 18 months of its establishment.

INTERVENTIONS AND MAIN OUTCOME MEASURES

Rate of breast masses, rate of malignancy, and types of evaluation and treatment.

RESULTS

Forty-seven patients aged 0-19 years were referred. Breast masses were suspected in 23/47 (48.9%) and confirmed using ultrasound in 14/47 (29.7%; mean age, 16.4 years), followed by needle core biopsy in 7. Seven patients had multiple masses. Breast Imaging and Reporting Data System scores ranged from 3 to 4b. All cases were treated as fibroadenomas. Four patients underwent surgery because of a large mass (>5 cm; n = 3) or patient's preference to remove the lump rather than follow-up (n = 1). There were no malignancies. The median time from patient identification of the mass until seeking medical help was 5 months.

CONCLUSION

Several findings distinct to this age group were highlighted: high rate of multiple masses, lack of malignancy, and high risk of delayed diagnosis. Care should be taken when using Breast Imaging and Reporting Data System categorization because it was formulated for adults and might be very limited in the adolescent population. Longer-term studies of larger cohorts are planned.

The implementation of natural language processing to extract index lesions from breast magnetic resonance imaging reports

BACKGROUND

There are often multiple lesions in breast magnetic resonance imaging (MRI) reports and radiologists usually focus on describing the index lesion that is most crucial to clinicians in determining the management and prognosis of patients. Natural language processing (NLP) has been used for information extraction from mammography reports. However, few studies have investigated NLP in breast MRI data based on free-form text. The objective of the current study was to assess the validity of our NLP program to accurately extract index lesions and their corresponding imaging features from free-form text of breast MRI reports.

METHODS

This cross-sectional study examined 1633 free-form text reports of breast MRIs from 2014 to 2017. First, the NLP system was used to extract 9 features from all the lesions in the reports according to the Breast Imaging Reporting and Data System (BI-RADS) descriptors. Second, the index lesion was defined as the lesion with the largest number of imaging features. Third, we extracted the values of each imaging feature and the BI-RADS category from each index lesion. To evaluate the accuracy of our system, 478 reports were manually reviewed by two individuals. The time taken to extract data by NLP was compared with that by reviewers.

RESULTS

The NLP system extracted 889 lesions from 478 reports. The mean number of imaging features per lesion was 6.5 ± 2.1 (range: 3-9; 95% CI: 6.362-6.638). The mean number of imaging features per index lesion was 8.0 ± 1.1 (range: 5-9; 95% CI: 7.901-8.099). The NLP system demonstrated a recall of 100.0% and a precision of 99.6% for correct identification of the index lesion. The recall and precision of NLP to correctly extract the value of imaging features from the index lesions were 91.0 and 92.6%, respectively. The recall and precision for the correct identification of the BI-RADS categories were 96.6 and 94.8%, respectively. NLP generated the total results in less than 1 s, whereas the manual reviewers averaged 4.47 min and 4.56 min per report.

CONCLUSIONS

Our NLP method successfully extracted the index lesion and its corresponding information from free-form text.

Agreement between dynamic contrast-enhanced magnetic resonance imaging and pathologic tumour size of breast cancer and analysis of the correlation with BI-RADS descriptors

Purpose

The purpose of this study was to evaluate magnetic resonance imaging (MRI)-pathology concordance of tumour size in patients with invasive breast carcinoma, with an emphasis on Breast Imaging Reporting and Data System (BI-RADS) descriptors of dynamic contrast-enhanced MRI (DCE-MRI).

Material and methods

Of patients who had preoperative DCE-MRI, 94 were enrolled. Concordance between MRI and the pathological findings was defined as a difference in tumour size of 5 mm or less. The greatest dimension was measured by two radiologists, and BI-RADS descriptives were described in accordance. The gold standard was chosen as the pathologic assessment.

Results

Tumour measurements determined by MRI and the pathological reports were not statistically different (2.64 ± 1.16 cm, Wilcaxon Z = –1.853, p = 0.064). Tumour sizes were concordant in 72/94 patients (76.6%). The mean difference between the pathological and MRI tumour sizes was –0.1 cm. MRI overestimated the size of 17/94 tumours (18.1%) and underestimated the size of 5/94 tumours (5.3%). Discordance was associated with larger tumour size. Histologic and molecular type of tumours, patient age, histologic grade, lymphovascular invasion or perineural invasion positivity, fibroglandular volume, background parenchymal enhancement, and being mass or non-mass were not associated with concordance. Irregular margin and heterogenous enhancement in DCE-MRI were associated with discordance in logistic regression analysis (p = 0.035, OR: 4.24; p = 0.021, OR: 4.96).

Conclusions

Two BI-RADS descriptors of irregular contour and heterogeneous contrast uptake were found to be associated with tumour size discrepancy. This might be attributed to the dynamic and morphologic specialities of tumours primarily rather than tumour biology.

Environmental Influences on Mammographic Breast Density in California: A Strategy to Reduce Breast Cancer Risk

State legislation in many U.S. states, including California, mandates informing women if they have dense breasts on screening mammography, meaning over half of their breast tissue is comprised of non-adipose tissue. Breast density is important to interpret screening sensitivity and is an established breast cancer risk factor. Environmental chemical exposures may play an important role in this, especially during key windows of susceptibility for breast development: in utero, during puberty, pregnancy, lactation, and the peri-menopause. There is a paucity of research, however, examining whether environmental chemical exposures are associated with mammographic breast density, and even less is known about environmental exposures during windows of susceptibility. Now, with clinical breast density scoring being reported routinely for mammograms, it is possible to find out, especially in California, where there are large study populations that can link environmental exposures during windows of susceptibility to breast density. Density scores are now available throughout the state through electronic medical records. We can link these with environmental chemical exposures via state-wide monitoring. Studying the effects of environmental exposure on breast density may provide valuable monitoring and etiologic data to inform strategies to reduce breast cancer risk.

Added Value of Different Types of Elastography in Evaluating Ultrasonography Detected Breast Lesions: A Compared Study With Mammography

BACKGROUND

The purpose of this study was to compare the diagnostic performance of ultrasonography (US) and mammography in the differential diagnosis of breast lesions after adding different types of elastography to US.

PATIENTS AND METHODS

This institutional review board-approved study included 316 breast lesions in 289 women between July 2016 and July 2018. All these lesions were evaluated with conventional US, elastography, and mammography before biopsy or surgery. Elastography, including elasticity imaging (EI), virtual touch tissue imaging (VTI), and virtual touch imaging quantification (VTIQ), were used to downgrade US Breast Imaging-Reporting and Data System category 4A lesions. Diagnostic performances were calculated for mammography, US elastography, and the combination of US and elastography.

RESULTS

The sensitivity of US (100%) was significantly higher than that of mammography (84.6%; P < .001), but the specificity of US (14.5%) was significantly lower than that of mammography (59.1%; P < .001). After adding EI, VTI, and VTIQ to US, the specificity was significantly increased from 14.5% to 69.4%, 72.6%, and 78.0%, respectively (P < .001), and were significantly higher than that of mammography (P = .043, P = .006, and P < .001, respectively). The sensitivity of US + EI (96.2%) and US + VTI (96.2%) was lower than that of US alone, although not significantly (100%; P = .063 and P = .063, respectively).

CONCLUSION

The addition of different types of elastography to US improved the diagnostic performance in the differential diagnosis of breast lesions when compared with mammography.