Personalized Screening is Key to Breast Imaging AI Market Reaching $205M by 2025

10 Factors Radiologists Should Consider When Selecting AI Vendors
Dr. Sanjay M. Parekh Ph.D. , Sr Market Analyst at Signify Research Ltd.

As October draws to a close, so does Breast Cancer Awareness Month (BCAM), or National Breast Cancer Awareness Month (NBCAM) as it is known in the US. Breast imaging AI vendors are heavily promoting this initiative to raise awareness of the disease and highlight the importance of women attending breast cancer screening appointments. Breast cancer screening programs are an important public health intervention, and initiatives like the BCAM may prompt eligible women to attend their screening appointments. However, uptake of screening only adds to the growing backlog of unread breast imaging scans, which was compounded further by the COVID-19 pandemic.

Breast imaging AI: a growth opportunity

AI solutions are increasingly playing a role in supporting breast cancer screening programs. These solutions are evolving from detection-only AI algorithms of comparatively limited clinical value to well-rounded clinically useful solutions that incorporate decision-support and risk-based screening, which aim to deliver improved patient outcomes.

Breast imaging is one of the more mature clinical applications of the medical imaging AI market. It was worth just over $60 million in 2020, and despite a blip in growth due to the COVID-19 pandemic, it is projected to grow to just over $205 million (CAGR 22%) by 2025. The breast imaging AI market is heavily dominated by the US, which accounted for over 80% of the total market in 2020, and this is projected to continue in the coming years. Most AI solutions target FFDM (2D mammography) scans, specifically for breast lesion detection and breast density assessment (see Figure 1). 

Figure 1: Breast imaging AI market forecast by application

Factors driving the uptake of breast AI solutions

The backlog of unread scans is a growing concern. One way healthcare providers are looking to address this is by adopting AI solutions in the workflow to improve radiologists’ productivity; for example, breast imaging AI solutions that offer triage capabilities enable cases to be sorted and prioritized in the worklist based on the presence of suspicious regions of interest. 

A combination of a demonstrable ROI with breast imaging AI solutions through a reduction in reading times, a reduction in unnecessary biopsies, and improvement in clinical performance and reduced recall rates, are all factors facilitating the uptake of breast imaging AI. These solutions support radiologists when reading scans as well as being a quality-assurance tool in single-reader workflows. However, in some instances (e.g., in the US), AI solutions are being deployed in double-reader workflows as the second read. 

Another market driver for breast imaging AI is the increasing adoption of digital breast tomosynthesis (DBT) systems. In the US, more than 70% of screening centers now have DBT systems, and in Western Europe, shipments of DBT systems are forecast to be comparable to FFDM systems in 2021. DBT scans require a longer time for images to be read by radiologists, a factor restricting growth for this technology. However, breast AI solutions developed for DBT could help to address this and drive growth for both DBT technology as well as AI solutions.

There is also a growing interest in the use of AI to interpret breast ultrasound scans. AI has the potential to improve cancer detection rates, improve diagnostic accuracy and improve radiologist workflow productivity, especially for time-consuming Automated Breast Ultrasound (ABUS) exams. Perhaps most importantly, AI can support less experienced clinicians with interpreting breast ultrasound scans, which will expand the reach of breast cancer screening in parts of the world where mammographic screening is not available. 

A shift to personalized breast cancer screening

Healthcare providers are also looking to widen the net for screening to identify at-risk women earlier, for example, by lowering the age of screening. This is being driven by recommendations from leading medical bodies, such as the US CDC, which is increasingly promoting risk or genetic testing for women, and the American College of Radiologists (ACR), which recommends risk assessment at the age of 30, and a tailored approach to screening thereafter. Early detection is key, and survival rates improve dramatically when the disease is caught early enough. However, a blanket approach to screening is not cost-effective, and personalizing breast cancer screening is necessary.

AI vendors are looking to address this market opportunity by developing risk-based screening tools that will also support healthcare providers to deliver a more tailored approach to breast cancer screening. In some instances, risk-based screening may result in an increased frequency of screening for high-risk women, but in other cases, it will reduce the frequency of screening for lower-risk women.

Identifying high-risk women earlier means that they can be offered more intensive screening, such as follow-up MRI scans after mammography. In contrast, for low-risk women, the intervals between screening may be lengthened, reducing the rate of false positives, an unintended consequence of over-screening. These personalized and risk-based screening approaches will inevitably generate a greater volume of scans, but AI solutions that support radiologists to read this increased volume of scans will increasingly be sought-after by healthcare providers.

Risk-based breast cancer screening AI tools incorporate multiple risk factors found in a screening mammogram, including a patient’s age, breast density, and subtle mammographic features, for example, to determine their risk of developing breast cancer. AI tools that quantify breast density are also invaluable in regions (e.g., Asia) where there is a high prevalence of women with denser breast tissue, a risk factor for developing cancer. In the mid-term, these risk-based AI solutions will look to leverage patient data (risk factors) from the EHR to improve risk assessment for screened patients, and vendors may look to develop partnerships with EHR vendors to support this.

In the long term, as risk-based solutions mature and are increasingly deployed in screening programs, they will incorporate other findings, such as genetic testing results (e.g., identifying mutations in the BRAC1 or BRAC2 genes that are likely to increase a patient’s risk of developing breast cancer). This trend gathered momentum earlier this year when Volpara Health acquired a breast cancer risk assessment company, CRA Health, expanding its reach across the breast screening workflow. This acquisition continued Volpara Health’s journey from a company that made a useful AI-based breast density assessment tool, to comprehensive breast health IT firm aggressively targeting personalized screening.

Barriers holding back the breast imaging AI market

The slow uptake of breast imaging AI solutions as part of screening programs outside the US remains a barrier to market growth. A double-reader workflow for breast cancer screening remains necessary, and unlike the US, many European countries are still hesitant to adopt the use of AI as a second read. 

The uptake of breast cancer screening also remains low in some Asian countries. More significantly, there remains a lack of national breast cancer screening programs in many growth markets, for example, in two of the most populous nations in the world, China and India.

Other barriers to market growth for breast AI include limited reimbursement for such solutions, especially outside the US, and a lack of recognition and recommendation for AI in clinical practice guidelines from authoritative bodies. For example, there is a lack of legislation promoting breast density assessment in Asian markets, where this is more prevalent.

The future outlook

Breast cancer screening programs that adopt AI to: (i) improve radiologist workflow productivity, (ii) enhance clinical decisions and reduce unnecessary biopsies, and (iii) personalize screening by identifying higher-risk individuals, are forecast to rise in upcoming years. 

Those vendors that offer clinically valuable, comprehensive AI solutions that address multiple aspects of the screening pathway will have a strong advantage over vendors that only offer comparatively narrow tools. Ultimately, identifying at-risk women at an earlier stage with better informed clinical decisions will enable them to benefit from earlier intervention and treatment and receive better follow-up management, significantly improving care outcomes.


About Dr. Sanjay M. Parekh

Dr. Sanjay M. Parekh Ph.D. is a senior market analyst at Signify Research Ltd. The company is headquartered in Cranfield, U.K. Signify Research is an independent supplier of market intelligence and consultancy services to the global healthcare technology industry. Our major coverage areas include Healthcare IT, Medical Imaging and Digital Health. Clients include technology vendors, healthcare providers, and payers, management consultants and investors.