Chest X-ray and Chest CT for TB Detection; Contrast-Enhanced Mammography for Women with Breast Implants; Risk-Based Approach to Optimizing Mammography During Crisis; Plus, Natural Language Processing for COVID-19 Case Volume Prediction
Welcome to Diagnostic Imaging’s Weekly Scan. I’m senior editor Whitney Palmer.
Before we get to our featured interview with Dr. Ricardo Cury of Radiology Associates of South Florida, a Radiology Partners practice, about the use of natural language processing to predict the number of COVID-19 cases, here are the top stories of the week.
Tuberculosis is, thankfully, rare in the United States, but recent reports show the decline is cases is starting to slow. That makes quicker, earlier detection a must. In a new study published in The Lancet Infectious Diseases, a team of researchers from Harvard Medical School determined that using chest X-rays and chest CTs to image patients with suspicious symptoms can expediate diagnosis and, potentially, limit transmission. A delay in diagnosis can also lead to greater disease progression. By examining 18.9 million medical insurance claims, they found that patients who did have a delayed diagnosis were more likely to have complications, such as irreversible lung damage and fungal lung infections. They were also much more likely to transmit the infection to other members of their household. Of the 1,026 household members who lived with the 456 patients identified with active TB, 25 percent became infected. But, conducting a chest X-ray or chest CT did lead to faster diagnosis when patients presented with symptoms, and even if there is some over-testing, they said, it is an inexpensive way to detect an infectious disease that is still present in the United States.
A new study has shown that contrast-enhanced mammography performs equivalently to breast MRI in women who have breast implants. The results published in the American Journal of Roentgenology shows that this emerging breast imaging technique could provide breast cancer screening for women who don’t have access to MRI or who cannot undergo the scan for a variety of reasons, including having a pacemaker. In a study of 17 women with breast implants, 11 of whom had dense breasts, the team determined that results from breast MRI and contrast-enhanced mammography were consistent for all 17 women – they identified invasive ductal carcinoma in 15 women, as well as invasive lobular carcinoma and ductal carcinoma in situ in one woman each. Contrast-enhanced mammography also picked up six additional lesions that were confirmed by MRI, two of which were malignant. MRI did not pick up any cancers not seen on contrast-enhanced mammography, the team said.
It’s well known that mammography screening mostly came to a screeching halt at the beginning of the COVID-19 pandemic, and providers struggled to determine who to bring in for screening and how to do it. The American College of Radiology and the Society of Breast Imaging did offer guidance, but now a new approach might be available – a risk-based algorithm. According to investigators led by the University of California at Davis, using clinical indicators and a patient’s personal history with breast cancer can help providers pinpoint which patients to bring in for screening, ultimately catching the most cancers with the fewer studies. Based on their results, published in JAMA Network Open, they found that 12 percent of mammograms – those from women at high-to-very high risk for breast cancer detection – caught 55 percent of cancers identified. They used these results to create two algorithms for how mammography screening services can be prioritized during times of crisis or reduced capacity. The first model enables scheduling high-to-very high cancer detection scans when operating at 12-percent capacity, and the second schedules moderate-to-high cancer detection scans when at 20-percent capacity.
And, finally, this week, Diagnostic Imaging spoke with Dr. Ricardo Cury, chairman of radiology at Radiology Associates of South Florida and director of cardiac imaging at Miami Cardiac and Vascular Institute, about the use of natural language processing to evaluate COVID-19 chest CT reports. Based on an investigation with more than 400,000 chest CTs, Cury and his colleagues built a prediction model that can accurately estimate official COVID-19 case counts on a daily, weekly, and state-by-state basis. Here’s what he shared.
For more coverage based on industry expert insights and research, subscribe to the Diagnostic Imaging e-Newsletter here.