MammoScreen AI Tool Improves Diagnostic Performance of Radiologists in Detecting Breast Cancer

A clinical investigation published in Radiology: Artificial Intelligence demonstrated that the concurrent use of a new artificial intelligence (AI) tool improved the diagnostic performance of radiologists in the detection of breast cancer by mammography without prolonging their workflow.¹

Researchers used MammoScreen, an AI tool designed to identify regions suspicious for breast cancer on 2D digital mammograms and determine their likelihood of malignancy. The system produces a set of image positions with scores for suspicion of malignancy that are extracted from the 4 views of a standard mammogram.

“The results show that MammoScreen may help to improve radiologists’ performance in breast cancer detection,” Serena Pacilè, PhD, clinical research manager at Therapixel, where the software was developed, said in a press release.²

In this multireader, multicase retrospective study, a dataset including 240 digital mammography images were analyzed by 14 radiologists by a counterbalance design, where each half of the dataset was read either with or without AI in the first session and vice versa for a second session, with the 2 sessions separated by a washout period. End points assessed by the investigators included area under the receiver operating characteristic curve (area under the curve [AUC]), sensitivity, specificity, and reading time.

Overall, the average AUC across readers was 0.769 (95% CI, 0.724-0.814) without the use of AI and 0.797 (95% CI, 0.754-0.840) with AI. The average difference in AUC was 0.028 (95% CI, 0.002-0.055; P = .035). The investigators said these data indicate greater interreader reliability with the aid of AI, resulting in more standardized results.

Further, average sensitivity was increased by 0.033 when AI support was utilized (P = .021). Reading time changed dependently with the AI-tool score.

For those with a low likelihood of malignancy (< 2.5%), the time was about the same in the first reading session and slightly decreased in the second reading session. For those with a higher likelihood of malignancy, the reading time was generally increased with the use of AI.

“It should be noted that in real conditions, additional factors may have an impact on reading time (ie, stress, tiredness, etc), and that those factors were obviously not considered in the present analysis,” explained the authors.

Importantly, the main limitation of this study was that the used dataset was not representative of normal screening practices. Specifically, a high rate of false-positive readings may have resulted due to readers awareness of the dataset being enriched with cancer cases, causing a laboratory effect. Moreover, because readers had no access to prior mammograms of the examined patients, other images, or additional patient information, the assessment was more challenging than a typical screening mammography reading workflow.

“…the overall conclusion of this clinical investigation was that the concurrent use of this AI tool improved the diagnostic performance of radiologists in the mammographic detection of breast cancer,” wrote the authors. “In addition, the use of AI was shown to reduce false negatives without affecting the specificity.”

In March, the FDA cleared MammoScreen for use in the clinic, where it could aid in reducing the workload of radiologists. Moving forward, the investigators plan to continue to explore the behavior of the AI tool on a large screening-based population and its ability to detect breast cancer earlier.

References:
1. Pacilè S, Lopez J, Chone P, Bertinotti T, Grouin JM, Fillard P. Improving breast cancer detection accuracy of mammography with the concurrent use of an artificial intelligence tool. Published November 4, 2020. Radiology: Artificial Intelligence. doi:10.1148/ryai.2020190208

2. AI tool improves breast cancer detection on mammography. News release. Radiological Society of North America. Published November 4, 2020. Accessed December 3, 2020. https://www.eurekalert.org/pub_releases/2020-11/rson-ati110220.php