Conventional CAD(computer-aided detection / diagnosis)has been commercialized mainly in mammography and has achieved some results, but there are not a few problems that hinder its widespread use such as performance. Deep learning is a new technology that overcomes this hurdle, and CAD is now undergoing further evolution / diversification as AI-CAD. Recent prominent journals include those that show that AI-CAD performs as well as or better than physicians, those that compare the performance of multiple AI-CADs and verify their effectiveness, those that show that AI-CAD can reduce the number of false positives and the number of recalls compared to conventional CAD, and those that show the effectiveness of AI-CAD in automatically classifying normal and possibly abnormal images in screening(triage type) to significantly reduce the workload of physicians. The CEO of ScreenPoint Medical in the Netherlands predict that the performance of mammography AI-CAD currently on the market will reach the average radiologist's interpretation level, and will reach the level of high-level radiologists or higher around 2025. However, even so, the current situation is that the performance will be further improved compared to “AI-CAD” alone or “doctor” alone, not by automatic diagnosis, but by the collaboration of “AI-CAD + doctor”. This paper describes the progress of the latest AI-CAD, the problems to be solved, and the future utilization of the new AI-CAD in breast cancer screening, based on the contents of the author's keynote lecture, with some of the contents updated.

View full abstract

[ background ] Breast cancer screening has been performed with mammography images. Mammography reading skills are known to be varied among readers. Artificial intelligence (AI) could help to increase overall accuracy rate and to reduce work overload to make the system more efficient. Previous study revealed the promising result on automatic interpretation of mammography images. Our aim is to clarify whether the accuracy rate interpreting various types of mammography is not inferior to the previous study. [materials and methods] KBOG, the non-profit organization in Kanagawa prefecture has initiated the clinical trial, called KBOG1701 trial. Nine institutes had prepared more than a thousand mammography images with annotation information. Cropped images were created from mammography images and were used to train AI. [result] Based on cropped images, the accuracy, sensitivity, specificity and AUC rates were 81%, 88%, 74%, 0.880, respectively. Trained model was implemented on a web server to automatically scan mammography images on the internet. [conclusion] Using CAD system based on AI could be helpful to reduce false negative risk and increase overall accuracy rate. AI could be the game changer on the breast screening system.

View full abstract

[in Japanese]

View full abstract

Automated Breast Ultrasound System, ABUS) has been developed as a supplemental breast imaging tool for dense breast women. A huge amount of whole breast scanned images are sent to the reading station where a breast imager interprets and reports a huge volume of images, which brings about each reader's unexpected stresses from longer reading time as well as uneasiness about accuracy. Intense needs for appropriate solution has been claimed by breast imagers in vain. This paper was edited prioritizing the readers' familiarity with QviewCAD™ System. (abbreviated as QVCAD) for correct and better comprehension as well as the top level of QVCAD's performances. 1. Comprehensible close investigation of histories of developments of ABUS and QVCAD, to learn when and how both were interlocked and/or synchronized. 2. Learning from valuable experiences presented by Dr. Ian Grady (Director of North Valley Breast Clinic, Redding, CA. USA), profound understanding by graphic visualization of effects of ABUS as well as QVCAD so that more readers feel stronger affinity especially with QVCAD. 3. Introduction of high level of integrity of QVCAD and stunning user interface named as C-thru? Navigator. 4. Performance of QVCAD in terms of shortening of interpretation time and achievements of high levels of diagnostic accuracy are all referred from world's top-level journals as well as presentations by key opinion leaders. Timely new movements taking place in 2021, such as US FDA's decision of downgrading QVCAD's level, from class 3 (PMA) to class 2 (510k), which may influence Japan's PMDA. In Japan, new governmental actions such as foundation of the Digitalization Agency, AI Hospital Concept by Cabinet Office, and so on. “ABUS with QVCAD New Era” is now getting closer.

View full abstract

More than 60,000 Japanese women are diagnosed with breast cancer annually, and 13,000 die from the disease.Therefore, breast cancer screening is very important to achieve mortality reduction. We are conducting research and development on deep learning of DWIBS, an image with excellent contrast that emphasizes microscopic water diffusion, and mammary MRI images with various contrasts such as T1WI and T2WI. Using the deep learning model Xception, we developed a diagnostic aid model for fat-suppressed T2-weighted images and diffusion-weighted images. The combination of AI to assist diagnosis by taking advantage of the characteristics of MRI images may further contribute to the goal of reducing the mortality rate of breast cancer screening.

View full abstract

[in Japanese]

View full abstract

In response to the results of J-START for women in their 40s, it is possible to add auxiliary diagnostic imaging such as US examination in MG examination of high-density breast (Dense Breast: DB) whose detection sensitivity decreases. It is being considered. However, since DB exists at a certain rate even in the 50s and over, there are many unclear issues such as the effects in other age groups and the effects by breast composition. In recent years, compressed breast thickness (CBT) has been attracting attention as an aid in examining the actual state of DB. Based on the report that the detection rate of breast cancer is not low for thin breasts with a CBT of less than 3cm even in DB, CBT is incorporated into the breast composition determination recommended by the sperm mechanism, and it is expected to be a screening stratification factor. In this study, it was found that there is a certain tendency in the relationship between breast composition and CBT in the screening population (general population). Since the relationship with the detected cancer has not been verified, it was difficult to refer to the screening stratification. However, there is room for consideration of its usefulness by comparing it with clinical information such as discovered cancer and intermediate stage cancer. it was thought.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

In this research, we investigated whether compressed breast thickness (CBT) can be helped by diagnostic stratification factor and we verified the correlation between CBT and other factors. Method: Using 3D imaging MLO mammography we verified 8 items. We categorized 4 groups from 460 cases. The groups range from 20mms to over 50mms, in 10mm increments. Results Discussion: The patient's mean age was 53.8 years, mean compression was 103.5N and mean CBT was 39.7mm. Dense breast rate has a tendency to decline when CBT is large. There is only a low positive correlation between AGD and CBT, likewise BMI and CBT results are positively correlated. As CBT got larger, tumor rates became higher. Rate of both benign and malignant tumors increased. Furthermore, sensitivity of malignant lesions increased after adding a 3D mammography. After a 3D mammography the larger the CBT is, the higher the rate of category change. Compared to CBT 20mm level, CBT 50mm or higher levels showed higher rates of tumors and malignancy. For breasts with smaller CBT it's possible to manually check for tumors at an early stage, however for breasts with larger CBT it becomes difficult to manually check for tumor, which seems likely to delay their finding. Previous research concluded that the risk of developing breast cancer has a tendency to increase in proportion with BMI. In this research CBT increased proportionally with BMI, and the malignancy rates were higher when CBT was large. Among breasts with larger CBT there tends to be a low proportion of dense breast. In addition, the detection sensitivity of malignant tumors was high. Conclusion: According to these results, we believe the risk of breast cancer increases with larger CBT. In summary, there is a higher possibility of early detection of breast cancer by 3D mammography for breasts with larger CBT. It is important to increase breast awareness, and awareness of all available examination, particularly among patients with larger CBT breasts.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

Purpose: The aim of our study was to compare the clinical performance between digital breast tomosynthesis-guided vacuum-assisted biopsy(DBT-group) and stereotactic vacuum-assisted biopsy(ST-group). Patients and Methods: We retrospectively reviewed the data of 82 women with 83 lesions who underwent ST-VAB between April 2019 and March 2020 and 35 women who underwent DBT-VAB between April 2020 and November 2020. In all cases, calcification had been detected on mammographic screening. The technical feasibility of the procedure and the procedure time were compared between the two groups. Results: The calcifications were retrieved in all cases(100%). No differences were observed in the calcification morphology or malignancy rate between the two groups. The time taken for positioning and the total examination time were significantly shorter in the DBT-VAB group. Conclusions: DBT-VAB seems to be a safe procedure that can be performed rapidly, and the observation period after DBT-VAB is also short; however, the sample sizes were smaller. Further studies are needed to evaluate the diagnostic ability of DBT-VAB.

View full abstract

It is necessary to consider how to deal with the wish and literacy of the women in regard to breast cancer screening, in order to resolve the dense breast problem, including that of notification of women about their breast density on screening. Therefore, we created an original leaflet presenting the information needed to understand about breast cancer screening. After distributing the leaflets, we conducted a questionnaire survey of the women to determine their understanding of the contents of the leaflet, their level of knowledge and willingness to know about their breast density, and their willingness to undergo the next screening. The participants were 2253 women who underwent population-based breast cancer screening between March 2016 and March 2017. Over 90% of the women understood the contents of the leaflet that we had created and distributed. About a half of the women had prior knowledge about breast density, while only a minority(30.0%) were aware of the benefits and harms of breast cancer screening prior to reading the leaflets. Over 95% of the women expressed a willingness to undergo screening the next time, even after understanding the potential harms after reading the leaflet. We found that even though the women's current level of knowledge about breast cancer screening was insufficient, the women expressed a high level of desire to know about their breast density. Therefore, it is most important to establish a system in order to correctly convey the pros and cons, that is, the benefits and harms, of breast cancer screening.

View full abstract

Since 2008, the Iwate Cancer Society has been implementing a third-reader system in which a third reader makes the final judgment when the first and second readers offer different interpretations of screening mammography images. In this study, 351,134 women aged ≥40 years who underwent screening mammography (alone or with palpation) at the Iwate Cancer Society between 2008 and 2018 were included. Of the 7,282 third readings performed, 6,154(84.5%) were judged as not requiring further detailed examination. If a third reading was not performed, the recall rate was 4.1%, but finally, the results showed a recall rate of 2.4%, breast cancer detection rate of 0.27%, and a positive predictive value of 11.5%. In addition, third readings had not been performed for interval breast cancer until 2020. Of the 3,433 women for whom a third reading was performed during the period between 2014 and 2018,2,669 and 764 were judged by the first and second readers, respectively, as requiring further detailed examination. Of the cases judged by the second and first readers, 369(48.3%) and 227(8.5%), respectively, were judged as requiring further detailed examination by the third reader. Of the 84 cases of breast cancer diagnosed, 70(83%) were diagnosed by the second reader, while only 14(17%) were diagnosed by the first reader, with PPVs of 9.2% and 0.5%, respectively. Some kind of intervention is needed for cases with discrepant judgement between the first and second readers, and one of the potentially promising is the use of a third-reader system.

View full abstract

Breast awareness refers to awareness of a woman about the normal look and feel of her own breasts, which would enable her to immediately notice any changes in these attributes, potentially facilitating early diagnosis of breast cancer. Breast awareness is the first and important breast screening step for reducing breast cancer mortality. In many developed countries, breast cancer mortality has already decreased, whereas in Japan, the mortality is still increasing. The extremely low participation rate in screening mammography is the most important reason for the continued increase of breast cancer mortality in Japan. Poor breast awareness, poor awareness about breast cancer screening, and familial/hereditary breast cancer are crucial factors that contribute to the extremely low participation rate in screening mammography in Japan. In a cross-sectional online survey carried out to investigate the knowledge about the importance of breast awareness, breast cancer screening, and familial/hereditary breast cancer in 1000 Japanese women aged 20 and 69 years old, only 5% of the participants knew about breast awareness. In regard to awareness about breast cancer screening, 31% of the participants knew that mammography is the appropriate modality for breast cancer screening, and 22% were aware that screening mammography is recommended for women aged 40 years old and over in Japan. To our surprise, 71% of the participants were aware about familial/hereditary breast cancer. Thus, the percentages of Japanese women who knew about the importance of breast awareness and breast cancer screening were low, and we suggest the need to devise educational programs and campaigns to spread knowledge about breast awareness and improve participation in routine breast cancer screening.

View full abstract

The purpose of this study was to evaluate the diagnostic performance of automated breast ultrasound(ABUS)for breast diseases as compared to hand-held breast ultrasound(HHUS). Materials and Methods: A multicenter study was conducted on 496 subjects recruited from 6 hospitals between January 2018 and August 2019. All women underwent both ABUS and HHUS and the images were interpreted by 9 doctors using the guideline published by the Japan Association of Breast and Thyroid Sonology for judging the diagnostic category. Each case was interpreted by 3doctors and none of the doctors interpreted both ABUS and HHUS images from the same cases. Lesions judged as category 4 or over were counted as suspicious lesions, and lesions judged as category 3a or over were counted as detected lesions. Results: There cases comprised 235 malignant and 261 non-malignant cases. The overall diagnostic sensitivities of ABUS and HHUS were 54.3% and 78.5%, and the diagnostic specificities were 87.1% and 81.2%, respectively. Thus, while HHUS had a higher sensitivity（p<0.01), ABUS had higher specificity (p=0.002). On ROC analysis the AUCs for ABUS and HHUS were 0.755 and 0.862，respectively, being larger for HHUS (p<0.01). As for the detectability of malignancies by any of the 3doctors, the sensitivities of ABUS and HHUS were 91.1% and 99.1%, and the specificities were 26.4% and 23.4%, respectively; thus, the sensitivity of HHUS was higher(p<0.01). Conclusion: ABUS showed inferior sensitivity to HHUS in which targeted images were obtained under the clinical information.

View full abstract

A woman in her 30s working at a hospital was admitted to our hospital for a routine follow-up checkup and medication, as she had undergone right skin-sparing mastectomy and breast reconstruction at another hospital. The sonography at the time of the checkup showed two 10-mm infraclavicular lymphadenopathies with uniformly diffuse thickened cortices and fatty hila. As the patient had received her second dose of the COVID-19 vaccine in her left deltoid muscle 4 weeks prior to the test, the infraclavicular lymphadenopathies were considered as representing benign reactive lymph node enlargement in response to the COVID-19 vaccination. The lymphadenopathy was shrinking in a repeat sonography performed 12 weeks after the second dose of the vaccine. With the mass rollout of COVID-19 vaccinations, COVID-19 vaccination-related unilateral lymphadenopathy has become an important clinical condition to be recognized by clinicians, patients, and the general population. The lymphadenopathy is visible on imaging as early as 1 day after the vaccination, and in some cases, persists for more than 10 weeks after the vaccination. Therefore, imaging tests for non-urgent indications, such as screening mammography, should be scheduled prior to vaccination or be postponed to at least 6 weeks after the final vaccine dose; however, imaging tests for urgent clinical indications, such as treatment planning, active treatment monitoring, and assessment of new symptoms should not be delayed, irrespective of the vaccination status. An additional strategy to mitigate confounding findings would be to administer the COVID-19 vaccine on the side contralateral to the primary cancer (in patients with/with a history of cancer) or in the thigh. Ipsilateral lymphadenopathy following COVID-19 vaccination in the previous 6 weeks may be considered as being benign, and further imaging is not indicated. However, clinical management is recommended, with ultrasonographic examination, if the concern persists later than 6 weeks after the final vaccine dose. Recording of the vaccination status (date[s] of vaccination[s] and injection site [left or right, arm or thigh]) for all patients presenting for imaging and having that information readily available to the radiologist and physician at the time of interpretation can reduce unnecessary return visits by the patient and/or delays in the final assessment. Widespread patient education regarding COVID-19 vaccination-related unilateral lymphadenopathy is also needed, especially as this condition may be mistaken as a sign of malignancy. Imaging societies, clinicians, and news media outlets should spread awareness and educate the public about this clinical condition to minimize patient anxiety. At the time of vaccination, the possible development of axillary swelling should be highlighted, and the subjects should be informed that it is part of the normal immune response elicited by the vaccine. Patients also should be aware of the best times to schedule routine screening mammography after COVID-19 vaccination.

View full abstract