By Susan Harvey, MD
With an abundance of patients to see and data to analyze, health care professionals can become overwhelmed with their workload, especially as they face national and practice standards to simultaneously work efficiently while delivering the highest quality care to their patients. As a response to these pressures, triaging patient cases and information is a tactic that clinicians have used for years. While triaging is necessary to work efficiently and maintain high quality patient outcomes, manual triage practices can be challenging, creating an undeniable need for solutions that can help streamline the process.
Here, we consider the triaging needs of a breast imager and the breast cancer screening process.
As a woman enters the screening process, understanding her individual risk can inform the patient pathway. Broadly, risk includes genetic markers, family history, previous breast cancer diagnosis and breast tissue density. Each of these components influence the breast imager and the imaging recommendations. One clear example is breast tissue density, as women with very dense breasts are four to five times more likely to develop breast cancer than women with less dense breasts. This data is according to studies published in The New England Journal of Medicine and the Journal of the National Cancer Institute.
Traditionally, radiologists visually assess the images, using both FFDM and digital breast tomosynthesis (DBT), to determine their patient’s breast density category according to the Breast Imaging Reporting and Data System (BI-RADS). This categorization of breast tissue density has been proven to be variable as these assessments are subjective according to various studies, such as “Accuracy of Assigned BI-RADS Breast Density Category Definitions,” published in the National Library of Medicine. Unfortunately, this does lead to inconsistencies in patient recommendations and care.
There is, however, quantitative technology using AI available, which has begun to standardize density reporting, such as Quantra™ 2.2 Breast Density Assessment Software, which uses texture and pattern analysis and offers more consistent, more reliable scoring.
As new technologies are developed, such as Digital Breast Tomosynthesis (DBT), time consuming changes in workflow can be mitigated. For example, due to the large increase in the number of images provided by DBT, prioritizing which cases may have cancers and should be read first can be challenging. Certain AI technology, however, can help streamline and facilitate new workflows.
AI technology now exists that uses the large data sets produced by DBT systems, and constructs and processes the images into interleaved 1 mm images, while identifying clinically relevant regions of interest to focus on, and preserving important features. AI technology can also enhance accuracy by highlighting suspicious features on the images that may require additional imaging. This can potentially reduce unnecessary callbacks, benefiting patients and providers by improving outcomes.
In conclusion, working accurately and efficiently is a top priority in health care and this is why triage is widely used. The example of breast cancer screening is just one demonstration of how AI will work alongside and augment clinicians’ capacity and capabilities, creating a positive impact on patient outcomes and workflow.
Susan Harvey, MD, is Vice President of Global Medical Affairs in the breast and skeletal health division at Hologic, Inc.