CAD in concurrent reading mode boosts breast cancer detection, shortens read times

Computer-aided detection (CAD) can help inexperienced radiologists detect cancers on automated breast (AB) ultrasound (US), particularly in asymptomatic women.

Moreover, CAD used in concurrent-reading mode improves the diagnostic performance of both novice and experienced readers, while shortening overall reading times, according to a June 18 study published in Radiology.

“Currently, most CAD systems are used at the second-reading mode for breast cancer screening,” wrote co-lead authors Shanling Yang and Xican Gao, from Xijing Hospital of the Fourth Military Medical University in China, and colleagues. “However, this mode prolongs the reading time, especially for large volume imaging. Our study showed the use of concurrent-reading mode reduced the time required to read AB US images with similar sensitivity and specificity compared with second-reading mode.”

In second-reading mode, CAD is applied after the radiologist has completed their assessment. In concurrent-reading mode, it is used at the start. The latter technique may limit a reader’s vigilance, thereby reducing sensitivity, the researchers explained. It also goes against recommendations for CAD mammography reads which suggest the computer recommendations only appear after an initial assessment is done.

With that in mind, the team compared the performance and reading time of three radiologists with 1-3 years of US experience and three with 5-10 years of experience when using AB US CAD to detect breast cancer. They were asked to read the images without CAD, using CAD in second- reading mode and in concurrent reading mode.

The researchers retrospectively collected 1,485 AB US images from 1,452 women (36.4% asymptomatic) between 2016 and 2017. The images included 282 malignant lesions, 695 benign and 508 healthy lesions.

Overall, both groups improved when using the CAD. The mean area under the receiver operating characteristic curve rose from 0.88 without assistance to 0.91 at second-reading mode and 0.90 at current-reading mode. Sensitivity also rose in both modes.

According to the authors, CAD could be especially helpful for novice readers in women who are asymptomatic, as sensitivity jumped from 67% without assistance to 88% when using CAD at both second-reading and concurrent-reading modes.  

CAD used in concurrent-reading mode could also positively impact radiologist workflow. When CAD is used concurrently, readers could save 16 seconds compared to not using CAD at all, and they could perform reads 27 seconds faster compared to CAD used in second-read mode.

“Thus, our results suggest that the concurrent-reading mode with shorter reading time is preferable for workflow efficiency,” the researchers wrote.

In a related editorial, Priscilla J. Slanetz, MD, MPH, with Boston University Medical Center’s Department of Radiology in Boston, highlighted the time saved with CAD in concurrent-reading, noting such efficiency could have wide-reaching impact.

“The fact that CAD significantly shortened interpretation time is important, especially if either state or federal legislation ends up mandating, or even recommending, additional screening with US for women with dense tissue on mammograms,” Slanetz noted. “This savings in interpretation time should not be underestimated because it will allow higher throughput with no need to increase radiologist staffing, thereby providing greater access to this promising screening technology.”

""

Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.

Around the web

A total of 16 cardiology practices from 12 states settled with the DOJ to resolve allegations they overbilled Medicare for imaging agents used to diagnose cardiovascular disease. 

CCTA is being utilized more and more for the diagnosis and management of suspected coronary artery disease. An international group of specialists shared their perspective on this ongoing trend.

The new technology shows early potential to make a significant impact on imaging workflows and patient care.