AI

AI and the future of breast cancer

By Tim Simpson, General Manager at Hologic

Published

on

With both incidence and mortality of breast cancer set to increase over the next 15 – 20 years, there is a clear need to ensure the breast health field is primed to ensure the entire continuum of care, from detection to treatment, is working as efficiently as possible.

Figures releases from the Royal College of Radiologists (RCR) last week revealed a bleak picture for patients with breast cancer in the UK.

NHS staff shortages in 97 per cent of cancer centres impacting quality of care, longer waiting times, and delays in treatment of 6-8 weeks.

Described as a ‘crisis in the cancer workforce’ and one that impacts the entire continuum of care for breast cancer patients.

The true integration of the care pathway must be a priority to deliver the best possible outcomes for clinicians to patients, delivering tailored solutions and clinical effectiveness.

But this requires a workforce to deliver.

With a 29 per cent shortfall in radiologists, who play a critical role in breast cancer detection, one opportunity would be for greater deployment of technology and artificial intelligence (AI) to alleviate some of this burden and support efficiencies in breast health services.

Optimising detection

Breast cancer can be treated extremely effectively, particularly when detected early.

National screening programmes that were introduced in 1980s saw a rise in survival rates globally, demonstrating the need to maintain and invest in these initiatives as incidence is projected to increase.

With a projected increase in cases comes an inevitably greater demand on resources.

In order to manage increased demand, and a growing decline in the number of radiologists, the breast health community must consider various ways to drive efficiencies in the system.

Risk stratification to prioritise those at most risk of developing serious disease should move from theory to practice.

While some risk factors have been recognised for years, notably family history, predictive models have been difficult to establish given the factors are varied and numerous.

The validation of such models should be a priority, in order to better serve those most at risk and also those at least risk, who can be screened at a lesser frequency.

The frequency of screening is one aspect that can be tailored according to risk. Another important consideration is defining the screening pathway according to the risk factors identified.

Breast density is mentioned by the European Commission in its recent recommendations as an area where ‘specific diagnostic measures’ should be reviewed.

Recent publications, like the TOSYMA trial, have given reasons for optimism, given it showed digital breast tomosynthesis (DBT) plus 2D mammography demonstrated a clinically and statistically significant increase in detection rates for invasive breast cancer in women compared to digital mammography (DM) alone.

Innovative technology like DBT has the power to transform breast cancer screening through increased accuracy at the detection stage, but different imaging technologies have their role to play along the full continuum of breast cancer care from staging, to assessing the success of treatment and continuous monitoring post treatment.

Introducing artificial intelligence

The introduction of AI to help ease the burden on radiologists has been a topic of much discussion in recent years, as volumes of digital images increase without the corresponding levels of radiologists to support them.

There are varying degrees of support AI can offer from workflow prioritisation, which sees the most relevant or suspicious regions highlighted to the image reader, to being responsible for detecting breast cancer.

Workflow prioritisation is extremely relevant across the whole continuum of care given the heavy workload of breast health professionals across Europe.

For radiologists, the use of AI for workload prioritisation is particularly relevant for enhanced imaging technologies like DBT, which generates more images per patient and can increase reading time.

There remain many questions around AI, from ethical concerns to queries on its performance amongst diverse populations.

However, the hurdles should be confronted head on to help manage the realities of an increasing disease burden compounded by workforce shortages.

Click to comment

Trending stories

Exit mobile version