Good morning. On behalf of the RSNA Educational Committee and our TRAC Chair, Dr. Diane Strollo, we welcome you to the Essentials of Breast Imaging course, MSCS 33. We have four excellent presentations for you. I am Dr. Ellen Mendelsohn, and I'll be presenting the first on Urgent and Emergent Ultrasound-Guided Breast Interventions. My disclosures are there for you to see. And at the conclusion of this presentation, participants will understand that acute abscesses require emergency intervention, recognize types by clinical history, know that ultrasound is the diagnostic procedure in judging mastitis versus abscess questions, and that initial abscess treatment is no longer surgical incision and drainage. There'll be familiarity with treatment measures, and know when surgical intervention is necessary. At the end, we hope that you'll have awareness of leave-me-alone findings and also of inflammatory breast cancer, which can be an abscess mimic to be considered in patients unresponsive to treatment. Evaluation for abscess is an emergency. Mastitis is a breast inflammation. It can be infectious or noninfectious. It's most commonly related to lactation with local symptoms of pain, redness, and warmth. Occasionally, systemic systems will occur that they're flu-like with fever, chills, and aches. Ultrasound shows tissue that is heterogeneous and with architectural distortion and edema. If a fluid collection is not observed, the suggested management is antibiotic treatment and ultrasound follow-up. For if a fluid collection is found, a percutaneous intervention is indicated. This photograph, very uncomfortable to see, looks at a breast with an abscess bulging the skin near the areola. Nine percent of women who are breastfeeding will have lactational abscesses, and it's usually during the first month after delivery, and again, at highest risk at weaning because of milk stasis. Entry is through a cracked nipple, and I think you can see on the areola a linear crack. A normal appearance of fibroglandular tissue in a 37-year-old patient is seen on the left. And here, the ducts go through an area of echogenic fibroglandular tissue in an even pattern. On the right, you see architectural distortion and hypoechoic heterogeneity, a sense of swelling as well. This is mastitis in a lactating patient who is symptomatic with swelling and redness. The patient with mastitis returns sometime later with an abscess collection you can see in the right panel, about to rupture through the skin. She had failed her antibiotic treatment, which sometimes happens. Abscesses can have variable appearances. In the image on the left, the mast, a palpable one, was thought to be solid despite the fluid collection you see anteriorly. Core biopsy rather than aspiration was performed. The histology was acute abscess. In the image on the right, a common vascular pattern, rim vascularity, that we've seen with abscesses is noted. It is not specific, but in the appropriate setting, can suggest abscess. Imaging evaluation is important. And ways to evaluate the extent of large lesions would include extended field-of-view sweeps, a splicing method whereby two adjacent images are put together, and then trapezoidal display. This is an example of panoramic technique which shows the full extent of a 9-centimeter abscess. And trapezoidal display, where the posterior portion of the field-of-view is widened and deepened for greater visualization of the collection. For the interventional procedure, use at least 10 ml of 1 or 2% lidocaine, anticipating that the procedure will be painful for the patient. Aspirate or drain with a 16 or 18-gauge needle with tubing. If the material is very thick, you can use a 12 or 14-gauge needle and syringe to pull the material out. For large abscesses, a trocar technique can be used to place a 6 or 8 French self-retaining catheter. Serial aspirations every three days to start. The patient should continue to breastfeed from her unaffected breast and pump the affected breast to prevent myelostasis. For antibiotic treatment, which should be administered, there's currently concern for MRSA coverage. And in those cases, clitomycin and vancomycin. Galactoceles, painless cysts filled with milk or milky fluid occurring during or after lactation, are not urgent. Some fluid collections and abscesses that are unassociated with pregnancy and lactation include infected cysts, some subacute and chronic processes like idiopathic granulomatous mastitis, and intraductal ductectasia-related processes. Most of these are not urgent unless the patient is symptomatic, which she can be. Risk factors for these abscesses include smoking, diabetes, immunosuppressed conditions, as well as presence of foreign bodies such as nipple rings and piercings. Other categories include post-procedural complications like pseudoaneurysms after stereotactic biopsy and post-surgical fluid collections, which one can expect to see after lumpectomy and generally unless the patient is symptomatic or leave-me-alone lesions. This is ductectasia and abscess, and the treatment may require surgical excision. Aspiration should be performed initially, but they can recur even after duct excision. In a patient who had left mastectomy and chest wall radiation, an abscess formed around a median sternotomy wire that was placed at the time of a valve replacement unassociated with her left breast cancer. You can see the abscess formation and a bit of the wire deep in the abscess collection on these midline ultrasound images. A pseudoaneurysm that occurred after a stereotactic biopsy, manual compression for 30 minutes was able to control the bleeding. Seek surgical consultation for failed percutaneous drainages for large multiloculated collections for cleanup if abscess has ruptured through the skin, and rarely, if needed, for uncontrolled bleeding after biopsy procedures. This is a leave-me-alone. It's a post-surgical seroma six months after lumpectomy and radiation in a 79-year-old asymptomatic patient. And here, a reminder not to forget inflammatory carcinoma. You can see the photo orange pattern on the skin and the coronal image lower left. And skin and subcutaneous edema and masses infiltrating the fibroglandular tissue on both the transverse and sagittal images. So, in summary, first-line treatment of breast abscesses is no longer surgical incision and drainage. Symptomatic lactating patients should be seen emergently and evaluated with ultrasound. If mastitis is diagnosed, antibiotic therapy and ultrasound follow-up are recommended. And if abscess is found, aspirate it and attempt to drain it. Post-surgical fluid collections and asymptomatic lumpectomy and radiation therapy cancer patients are leave-me-alone and no when surgical intervention is necessary. Don't forget inflammatory carcinoma masquerading as breast abscess when the patient is refractory to treatment. Thank you very much for your attention. And our next presentation is by Dr. Tula Destounis, and it's on parenchymal enhancement in breast MRI, a clue to breast cancer risk. I'm Stamatia Destounis. We'll be talking on background parenchymal enhancement, a clue to increased breast cancer risk. When we look at background parenchymal enhancement, this is normal enhancement of patient's fibroglandular tissue during contrast-enhanced MRI. The normal breast parenchymal will enhance and fluctuate with hormonal cycles or hormonal therapy. An assessment of volume of the enhancement and the intensity is determined on the first post-contrast image at 90 seconds. BP is not necessarily related to the amount of fibroglandular parenchymal present, and a patient with extremely dense tissue may have little to no BP. The factors that affect BP are endogenous estrogen, as determined by the menstrual cycle, menopausal status, hormone therapy use, tamoxifen use, and we know that BP may vary between individuals and also vary over time in the same individual. The enhancement may be symmetric or asymmetric. So if it is symmetric, it may be mirror image enhancement. This typically speaks of normal type of BP, a benign process. There may be some preferential enhancement in the upper outer quadrant because of how blood supply localizes or the inferior aspect of each breast. If the enhancement is asymmetric and more prominent in one breast, this could be due to a benign process or malignant, but it has to be evaluated, such as the case on the lower right. Now, when we assess BP, we assess it visually. And here, there's inter-observer variability with what we classify as what the BP is. Now, we have a classification of minimal, mild, moderate, or marked background parenchymal enhancement that we should all be using in our reports. So when we look at breast density, similar to that, the presence of VPE can affect image interpretation. Several authors have presented this. It's important to get the background parenchymal classification correct. Mel Sather looked at training different readers to agree and to improve their agreement over time on what the VPE is, and this was sustained. Also within the same reader, improvement got better and agreement got better over time with training and learning. Now, the relationship of breast density and risk is well-known to us, given that increasing mammographic breast density and breast cancer risk, there is a relationship. It is less understood what the relationship may be between background parenchymal enhancement on MRI and breast cancer risk. The studies to date have shown some mixed results, most likely because of difference in study design and variability in the patient population, and also the approach to assessing the background parenchymal enhancement. Several studies have reported on the association of background parenchymal enhancement with breast cancer and evidence for VPE as having a role as an additional marker for higher risk for breast cancer. However, there's just as many studies that have found no definite association between VPE and breast cancer risk. Now, Dr. King looked at VPE and breast cancer risk, and they looked at the odds ratio and compared it to normal controls. So the odds ratio for breast cancer increased significantly with increasing background parenchymal enhancement. There was increased odds for breast cancer associated with moderate or marked VPE, and this was evident in pre- and postmenopausal women. The breast cancer odds also increased with increasing fibroglandular tissue, but the VPE findings remain significant and independent even after adjusting the fibroglandular tissue. So King thought that VPE has the potential to serve as an additional tool for risk stratification in high-risk women undergoing breast MRI imaging examinations. Now, the authors in this study by Telegraphel investigated whether background parenchymal enhancement and breast cancer would correlate. The study researched any significant difference of VPE pattern distribution in the case of benign or malignant lesions. There were 386 patients, 180 were premenopausal, group one, and 206 postmenopausal, group two. The MRI images were classified as normal VPE, minimal, mild, moderate, or marked, and the two groups were further subdivided into three categories based on the MRI findings, negative, benign, or malignant lesions. The distribution of these patterns of VPE within the two groups and within the three MRI categories was calculated. What was found was there was a statistically significant difference in the VPE types in negative patients and benign lesions versus malignant ones. There was a higher prevalence of moderate and marked VPE found among malignant lesions, and a predominance of minimal and mild VPE among negative patients, whether premenopausal or postmenopausal, and benign lesions, premenopausal or postmenopausal. This meta-analysis by Thompson, looking at VPE as a risk factor for breast cancer, they reviewed 18 studies in the literature. The results showed that greater than minimal VPE was associated with higher odds of breast cancer in women with high risk. The association was not found for women with average risk. This suggests higher levels of background parenchymal enhancement may have the potential to predict the development of breast cancer, particularly invasive breast cancer, especially among women with high risk. The findings were consistent regardless of the study design, regardless of the VPE assessment, or the timing of the VPE measurement and VPE assessment by different radiologists. And then when they validated this analysis of case control studies, they removed unmatched studies, and that further produced similar effect results of at least mild VPE, or at least moderate VPE and breast cancer risk. Arasu looked at the association between MRI, VPE, and breast cancer risk, and found that more women with cancer had mild, moderate, or marked VPE than women without cancer, 80% versus 66%. When compared with minimal VPE, increasing VPE levels were associated with significantly increased cancer risk, and VPE should be considered for risk prediction models for women undergoing breast MRI. Here is the association between MRI, VPE, and future primary breast cancer risk in the different types of breast parenchymal enhancement and women with and without cancer. They also found that when you look at MRI background parenchymal enhancement and mammographic breast density, the women with cancer compared without cancer had the higher proportion of mild, moderate, or marked VPE, respectively, When they looked at heterogeneity denser, extremely dense breast, they also had more of the dense breast tissue pattern. When you combine VPE and density, the women with cancer had a higher proportion of mild, moderate, or marked VPE, and also a higher proportion of heterogeneity or extremely dense breast compared with women without cancer. Sippel looked at a study cohort of over 4,686 screening MRIs in 2,446 patients. 85% had minimal or mild VPE and 15% moderate or marked. And what they found was if you had moderate or marked VPE at screening, this was associated with greater risk of developing breast cancer within a year compared to minimal or mild VPE. This confirmed the relationship that higher VPE has some kind of correlation with breast cancer risk. They also adjusted for multiple confounding risk factors such as age, breast density, screening indication, and background parenchymal enhancement remain independent predictor of higher risk patients having screening MRI. So Sippel felt that identifying these patients with hormonally responsive breast tissue is important in order to allow more tailored screening along with other clinical factors, different types of treatment and prevention strategies for the patients at high risk. This is a patient that had a screening MRI. She has moderate to marked VPE and a 22.8% calculated lifetime breast cancer risk, and the small enhancing mass near the nipples in invasive carcinoma. Another patient here with moderate VPE and a 26% calculated lifetime breast cancer risk and an irregular enhancing mass in the left breast that was a small invasive ductal carcinoma. Dr. Grimm looked at the relationship between VPE on high risk screening MRI and future breast cancer risk. They subdivided the screening MRI patients into cancer cases and also control patients that were matched by age and high risk indication at a two to one ratio. There were non-significant differences in age, race, ethnicity, breast density, and chemoprevention therapy between the groups. What the study showed was that in patients with greater than minimal VPE, these patients were 2.5 times more likely to develop future cancer. So with VPE assessment, given the important clinical implications of the level of VPE on breast MRI, it is becoming increasingly important to accurately determine VPE level. In summary, VPE has been found to be a biomarker for breast cancer risk and variations in methods of VPE assessment need to become standardized. My references, this is my team, my medical outcomes and research team, and I wanna thank you for your time. I want to introduce Dr. Marsha Javid that will be speaking on personalized breast imaging art and practice of digital decision-making. Hello, this is Marsha Javid coming to you from Haifa, Israel at Rambam Healthcare Campus. In this Essentials of Breast Imaging course, I'll be speaking about personalized breast imaging, the art and practice of digital decision-making. And that's a mouthful. What we're really talking about in this next few minutes is radiomics. Radiomics is using the power of computers to data mine, do characterization and analysis, and combine multiple datasets to provide personalized care, risk stratification, and treatment planning based on objective information. If you think back to the beginning of imaging, when we first started doing ultrasound, MRI, and CT on a large scale in the 1980s, we began with descriptions, morphology. How did things look? What size are they? Where are they located? And we've come a long way since then. Radiomics and the use of artificial intelligence will enable us to extract data in unprecedented quantities analyze that data and come up with information that is unreachable otherwise, because we have extended our analysis and our capabilities beyond human limits. It's just too much to deal with in the human brain. So radiomics integrates information from every possible source relevant to the patient's care, clinical information, histology, serology, laboratory values, genomic information, and imaging, extracts features, analyzes data, and creates predictive models that will help us to do a better job of achieving optimal outcomes. Quantitative features from digital images are achievable now. Along with tissue signatures, we can look at metastasis to lymph nodes, the appearance of the nodes themselves and of the primary tumor. The expression of certain biomarkers like CHI67, status of receptors like ER, PR, HER2, TN, luminal A and B and more, angiogenesis, or micro-vessel density per unit volume of tissue, and even contrast kinetics. All of this combined with the clinical and imaging data allows us to achieve tailored management of patients and specific care that they deserve. If you search PubMed with the word biomarker, it returns 923,691 results. This is a hot topic. However, there is no oversight body for this, and there is no consensus on exactly how or what needs to be done. What are the benchmarks for performance? What are the clinical endpoints? Is it gonna be progression-free survival, disease-free survival, or conventional resist parameters, which basically all come down to Kaplan-Meier statistics. But we can do much more than we've already done. In this new world of radiomics, which is basically imaging data analysis for decision support, data is king. It rules. Whomever owns the data is in charge. So we're talking about using and harnessing artificial intelligence, more specifically machine learning, and ideally deep learning to help us to achieve what we're trying to do here. Let's talk a little about deep learning. It's a type of machine learning where many data resources are processed. There's little importance to pre-processing by humans in general, and it's probably more accurate than traditional machine learning. This is reliant on many layers of software-based calculators called neurons in a neural network. The information from many of these are spatially close together, and complex patterns in large datasets can be sorted using an approach like this. We use what is learned to process new data. So these interconnected neural networks allow us to find new meanings. Kernels will filter these elements and get rid of noise and save only the most important information. So the output of one neuron is gonna be the input of the next, and these hidden layers allow us to learn features of the data in a feature hierarchy. And here's a schematic of a neural network. The input layer, the hidden layers, and the output layer are not visible, but they're important, and we're gonna see much more of this as time goes by. It's already happening. Machine learning can identify, flag, and add relevance to the findings that we're seeing. It can offer advice to a radiologist, speed up our workflow by performing repetitive tasks much more quickly than people can. It can improve image quality, accuracy, can help us to separate abnormal from normal tissue by using segmentation. It can show us hidden information we might not see in patterns, and generate synthetic images from conventional imaging, and it can provide us with continuous variables that are relevant to patient management. So it's a powerful tool. So what's the big deal with AI? We get rid of repetitive high-volume tasks. We improve performance and accuracy. We use computer learning independently and adaptive intelligence. We uncover hidden information we might not otherwise see. And here we go. The data has the information. So the question is, what is data? Is it intellectual property? Does it belong to the patient? Does it belong to the institution? Who owns it? This is a big question. There are many other, both ethical and practical questions that will arise. So how do we get real? We improve feature extraction and accuracy, and use it for the benefit of the patients, not for the benefit of profiting for the institution or an entity privately. We improve diagnostic performance for the good of the patient. We enable better data mining for comprehensive evaluation and combining, synthesizing multiple sources of data, and we can move faster with workflow. AI can work. It's already starting to show up in the peer-reviewed published literature. It may not be a standalone process right now because somebody has to be responsible. What happens if an AI algorithm needs a mammogram and they make a mistake? It makes a mistake. Who's responsible? What about integrating hardware and software? We have many different vendors. The platforms may or may not be interchangeable. Some are, some are not. Can it be used for second readings? Probably. It can reduce costs, but we have to do a cost-benefit analysis ultimately and see if it really is an enduring change, an endurable change. The data extraction not found by humans is a big plus. Personalized screening allows us to give specific counseling and risk assessment to patients on an individual basis. High-risk mutations like BRCA1 and 2, TP53, P10, CDH1, and STIK11 are revealed and can guide what we do next if we know about them. Patients can make informed decisions when they're given this information. Moderate-risk mutations are ATM, CHEK2, and PALB2. And the goals are to find it fast, find it early, improve the outcome by decreasing mortality. The current tools for personalized screening involve use of genetic testing panels. Typically, these consider the patient's history of breast cancer before the age of 50 or six years, history of ovarian cancer, and positive family history of close relatives with breast, ovarian, colon, endometrium, prostate, and pancreatic cancer. NCCN uses the Gale model, which is without a positive family history of breast or ovarian cancer, or the Tyrocusic, which is a lengthy and detailed system that predicts the risk of breast cancer. These risk assessment models are in common use today. Gale model has eight questions, usually age at menopause, age at first birth, and prior breast biopsies. It considers family history of up to only the first degree family members. The class model from the CDC is an extensive list of parameters used, and it considered first and second degree relatives and age of breast cancer in each. The extended class also includes positive family history of ovarian cancer and risk of bilateral disease. Tyrocusic includes elevated estrogen exposure and prior benign breast biopsies. It takes an extensive family history, including BRCA1 and 2 risk estimates, and includes breast density. However, it's very time-consuming. The BRCA mutation models, BRCA Pro and Bodacia, are breast and ovarian cancer analysis of disease incidents and carrier estimation algorithms. This includes the probability of being a carrier of BRCA1 and 2. It is the only model with family history taken beyond the second-order relatives, and it has been validated in European and Ashkenazi Jews with a sensitivity of 85%. However, its use in other populations is uncertain. So the holy grail here is risk stratification with a synthesis of all data from radiomics, AI and deep learning, and genetics. And we will be using individual recommendations that will improve survival by performing early detection, which is the ultimate goal that we all share, saving women's lives. Thank you for your attention. The next speaker will be Georgia Giacomis-Spear, who will speak about supplemental breast screening with automated ultrasound. Hello, everyone. I look forward to speaking to all of you today. I would like to thank Dr. Mendelsohn and Dr. Strohler for the invitation to speak in this session, Supplemental Breast Cancer Screening with Automated Breast Ultrasound, and specifically the why and how. Here are my disclosures. We're gonna introduce breast density here today and talk about why automated breast ultrasound is a possible solution for adjunctive screening. We're gonna go over a few cases and talk about the how and success factors upon implementation. The average American woman has a one in eight chance of developing breast cancer during her lifetime. We know that regular screening with mammography has decreased mortality by 45%. We also know that mammography is the gold standard for breast cancer screening, and it is the only screening imaging test shown to reduce deaths due to breast cancer. But we also know that sensitivity of mammography is decreased, and specifically in women who are at high risk for developing breast cancer and women with dense breasts. So why is breast density so important? Well, it limits sensitivity of mammography due to the masking effect and increases our risk for developing breast cancer. And also the prevalence of breast density is high. 40% of women that we screen have what we call dense breast tissue. They fall in the higher of the two breast density categories based on VIRADs. Breast density makes an impact on mammography. And as a woman's tissue is more dense, the sensitivity and specificity for detecting cancer decreases significantly. In fact, we can miss 30 to 50% of breast cancers with screening with mammography alone. Breast density also imposes an increased risk for developing breast cancer up to four to six fold. And it is thought that this is due to growth factors present in the breast stroma. There's also a higher rate of interval cancers that develop in women with higher breast density. The majority of these cancers are ER negative, higher tumor grade, and larger in size, pretending a worse prognosis and a worse outcome. And just to define interval cancers, interval cancers are those detected in a patient between routine screenings. When we couple this information with the legislative movement and patient-informed laws, we see that as of 2020, 38 states have mandatory breast density-informed legislation. In fact, in our state, in the state of Illinois, I had the honor and privilege to serve as medical expert during the initiation and enactment of our breast density notification law that went into effect on January 1st of 2019. In addition, eight states and the District of Columbia have reimbursement legislation for supplemental screening. Our state, Illinois, is included. What that means is that if a woman has dense breast, she is allowed to have an ultrasound or an MRI covered by insurance to circumvent limitations of mammography. So what we have here is informed patients, and that increases demand for supplemental screening. This is a cartoon that's slightly outdated in its verbiage, but it really gets the point across. Patients know more about their diseases than we do, so we must utilize our advanced technologies so we can circumvent these issues. We have a plethora of imaging modalities in our armamentarium, including digital breast homosynthesis, which is really the next evolution of mammography. We have screening whole breast ultrasound and breast MRI, which is typically reserved for high-risk patients in the screening setting, and those with newly diagnosed breast cancer. There's also an emerging role of abbreviated MRI in the setting of dense breasts. We have molecular imaging, PEM, contrast-enhanced mammography. So we need to scrutinize each of those to see what could be utilized in our practices. But we need to keep in mind how we measure the impact of supplemental screening. And we look at the most important prognostic factors, which include reduction in interval cancer rate, reduction in the number of node-positive cancers, and a reduction in stage two to four disease. So therein lies the advantage of screening breast ultrasound because it has excellent contrast resolution. It overcomes superimposition of tissue. It overcomes breast density, and the equipment is widely available. There's no ionizing radiation, and it's very well tolerated by patients. No IV contrast, no claustrophobia. And specifically, let's talk about why automated breast ultrasound could be a solution. While the examination is performed by a technologist in less than 15 minutes using standardized screening protocols, image quality is less operator-dependent, and images get uploaded to a workstation, which then has post-processing capabilities and 3D interpretation. Large fields of view is also an advantage. The coronal plane really mimics the anatomic appearance of the breast and allows us to have additional information about how we can possibly detect a breast cancer. It is reproducible, and there's no ionizing radiation. And specifically, when you look at automated over handheld ultrasound, reproducibility and consistency are key. It's easy to perform. It doesn't require physician time. There's consistent image acquisition time that reduces unexpected delays and allows for a streamlined workflow. Physician time for interpretation decreases with experience and there are advantages, again, to post-processing at the workstation with the advantage of having reconstructed coronal views that are helpful in detecting architectural distortion. So let's talk about our background, our story, and the how in our practice. Well, I'm chief of breast imaging at North Shore University Health System, which is a teaching affiliate of the University of Chicago. Our practice is comprised of 14 dedicated breast imaging radiologists. We do not have ultrasound technologists. We have four comprehensive hospital-based breast centers and five satellite imaging centers, and our practice is very high volume. We have a high-risk breast clinic and the Kellogg Comprehensive Cancer Center. And we recognized a gap in clinical care when it comes to women with dense breasts and the limitations with mammography alone. So we implemented a supplemental screening program with automated breast ultrasound in November of 2015. The eligibility criteria for implementation of this particular modality included patients with a bi-red density category C or D who are at average or intermediate risk, patients who had difficult clinical exams. And one thing that was very important was that each patient had to have a current mammogram, and we defined that as having a mammogram within nine months. And the reason for that is that we wanted to ensure that automated breast ultrasound was truly an adjunctive supplemental modality and did not replace mammography as the gold standard for screening. Our referrals came from our very own radiologists and technologists, the high-risk clinic, breast surgery department, and referring physicians. And training was also sort of impeccably designed so that we can batch train small groups of radiologists and technologists to ensure that they learned not only how optimal imaging looks, but also interpretation strategies to ensure success from the get-go. And then these trained radiologists ended up being super users. So when we continued to batch train our radiologists, they served as the experienced group to help ensure that we facilitated smooth transition into this modality. We also developed a breast imaging reference sheet, and this helped our referring physicians significantly because we made sure that we made them aware where the equipment was located, where they can send their patients, when a certain patient would be eligible for which particular exam and what criteria we utilized. Education becomes very important when implementing a new modality into the practice. And especially for referring physicians and our breast surgeons and high-risk clinic who refer our patients specifically to us so that they can be on board with this mission as well. So in the beginnings of our implementation, I initiated a retrospective IRB-approved study to study the first 500 patients in our practice. And we found six small node-negative invasive breast cancers in our first 500 patients. There were no DCIS cases. We had a high cancer detection rate of 12 per 1,000, a very low callback rate at 7%, which is significant when you're looking at the advantages and disadvantages of a new modality. We had a positive biopsy rate of 55%, and we've also initiated an ongoing prospective three-year clinical trial to accumulate more data so that we can publish more information on performance measures and outcomes of this particular modality. And as of late 2015, we have performed more than 50,000 automated breast ultrasound exams. And here's a compilation in a table of numerous automated breast ultrasound studies around the world. And it really shows that there's an incremental cancer detection rate above mammography alone that can be initiated with automated breast ultrasound. Some of the studies do show a high callback rate. So I think it's important to note that a lot of these studies range not only in timeline. So we've had advancements of technology improvements and image quality that really raise the bar of the performance measures and outcomes, allowing us to reduce our recall rates. Here, we have a case of a 53-year-old with prior-array breast lymphectomy, BI-RADS breast density C, and a negative mammogram. And as you can see, there's no mammographic evidence of malignancy. So she had automated breast ultrasound, and we detected a sub-centimeter hypocoic mass. And this particular patient was recalled for additional targeted exam, confirming the presence of a indeterminate lesion, and the patient returned for biopsy with pathology of invasive ductal carcinoma. Here, we have another 54-year-old woman, no family history, high breast density, mammogram was negative. Here are the CC and the MLO views. And again, we detected a sub-centimeter mass, a small invasive tumor on the automated breast ultrasound. Here's confirmation on the targeted ultrasound. And here's the post-procedure mammogram showing the clip in an area of dense tissue where no mammographic abnormality existed. Success factors in implementation of an automated breast ultrasound program include a team approach and communication, training that is tailored to ensure success, knowing patient's anatomy and knowledge of image acquisition and automated breast ultrasound displays is gonna be important, and strategies for decreasing false positive interpretation are essential. Also, concurrent mammographic evaluation becomes imperative. So in conclusion, mammography remains the gold standard for breast cancer screening, and new technologies show great promise as adjunctive screening tools in the arena where mammography underserves, and specifically in dense breasts in women who are at average to intermediate risk. Automated breast ultrasound shows great promise, and our initial patient experiences have been very positive. And in summary, technology's mission in breast imaging is to seek specificity without loss of sensitivity, and it's safe to say that breast imaging is not a one-size-fits-all solution, and a risk-based screening approach is the future. Thank you for your attention. Thank you.

Breast Imaging

Ultrasound Interventions

Breast MRI

Radiomics

Artificial Intelligence

Breast Cancer Risk

Automated Ultrasound

Dense Breasts