Breast Cancer Screening Enhanced With Radiologist, Machine Learning Combo


  • A study published in JAMA Network Open states that combination of radiologists and machine learning can increase accuracy of breast cancer screening.
  • Many similar studies have also drawn the same conclusion.

According to a study published in JAMA Network Open, machine learning (ML) tools can enhance the accuracy of breast cancer screenings when combined with assessments from human radiologists. A commonly used tool called mammography screening can detect breast cancer at early stage. Researchers noted that multiple clinical trials showed that early detection via mammography screening could reduce mortality.

However, the group of researchers that conducted the study, also stated that mammography screening could not always be perfectly accurate. Roughly 9% to 10% of the 40 million American women undergoing breast cancer screening each year are recalled for additional diagnostic imaging. Usually, 4% to 5% of these women are ultimately diagnosed with breast cancer. This can lead to patient anxiety and unnecessary treatment or interventions.

Aiming to improve breast cancer screening accuracy by combining evaluations from radiologists with machine learning algorithms, the team based its study on results from Dialogue on Reverse Engineering Assessment and Methods (DREAM) initiative. DREAM is a crowd-sourced competition that sought to assess whether artificial intelligence (AI) algorithms could beat or meet radiologist interpretive accuracy.

DREAM initiative was conducted by investigators at IBM Research, Kaiser Permanente Washington Health Research Institute, Sage Bionetworks, and the University Of Washington School Of Medicine. Bringing together a broad, international scientific community, this initiative tested multiple algorithms for improved breast cancer detection.

Using hundreds of thousands of screening mammograms from women in the US, the team trained and validated the algorithms. Researchers trained the algorithms, using images alone or combined images, previous examinations, and clinical & demographic risk factor data that would confirm or rule out cancer within 12 months.

The top-performing algorithm achieved an area under the curve of 0.858 and a specificity of 66.2% which was lower than community-practice radiologists’ specificity of 90.5%. Researchers combined the top-performing algorithm with US radiologist assessments and this experiment yielded a higher area under the curve of 0.942 and resulted in a specificity of 92.0%.

Dr. Christoph Lee, professor of radiology at the University of Washington School of Medicine and co-first author of the paper, said:

“Based on our findings, adding AI to radiologists’ interpretation could potentially prevent hundreds of thousands of unnecessary diagnostic workups each year in the United States. Robust clinical validation is necessary, however, before any AI algorithm can be adopted broadly.”

Other studies also give thumbs-up to Radiologist-ML Combo

With AI being used more in clinical settings, more healthcare professionals depending on the technology, viewing it as important support tool. A survey conducted by GE Healthcare and MIT Technology Review revealed that 79% of providers believed AI tools to be helpful in reducing clinician burnout. This allowed professionals to deliver more patient-centered, engaging care. The survey said:

“Healthcare institutions have been anticipating the impact that artificial intelligence (AI) will have on the performance and efficiency of their operations and their workforces—and the quality of patient care. Contrary to common, yet unproven, fears that machines will replace human workers, AI technologies in healthcare may actually be‘re-humanizing’ healthcare, just as the system itself shifts to value-based care models that may favor the outcome patients receive instead of the number of patients seen.”

An unrelated study from New York University (NYU) showed that combining AI with analysis from human radiologists could significantly improve breast cancer detection. This would help radiologists reduce the number of biopsies needed. Senior study author Krzysztof J. Geras, PhD, assistant professor in the Department of Radiology at NYU Langone, said:

“Our study found that AI identified cancer-related patterns in the data that radiologists could not, and vice versa. AI detected pixel-level changes in tissue invisible to the human eye, while humans used forms of reasoning not available to AI. The ultimate goal of our work is to augment, not replace, human radiologists.”