AI for breast cancer: As of January 2021, Breast Cancer is the second most common cancer among women in the United States, there are more than 3.8 million women with a history of breast cancer in the United States.
To identify or diagnose breast cancer, doctors frequently utilise ultrasonography, mammography, MRIs, or biopsy.
Researchers from NYU and NYU Abu Dhabi (NYUAD) have built a revolutionary artificial intelligence system that reaches radiologist-level accuracy in recognising breast cancer in ultrasound pictures, according to a new study released this week.
Their findings are published in the journal Nature Communications, in a paper titled, “Artificial intelligence system reduces false-positive findings in the interpretation of breast ultrasound exams,” and was led by Farah Shamout, Ph.D., NYUAD assistant professor emerging scholar of computer engineering and colleagues.
“Though consistently shown to detect mammographically occult cancers, breast ultrasound has been noted to have high false-positive rates, the researchers wrote. “In this work, we present an AI system that achieves radiologist-level accuracy in identifying breast cancer in ultrasound images.”
The system was also tested for its performance on a database of over 30,000 images from the National Cancer Institute (NCI).
The CNN correctly identified malignancies in 96.67 percent of cases out-performing radiologists who achieved 93.21 percent accuracy and board-certified sonographers with 91.39 percent accuracy.
The results highlight that when it was combined with radiologists in a hybrid model, the highest accuracy rate of 98.52 percent was achieved for the detection of malignancies.
The researchers note that their system could be used to help patients in regions that do not have easy access to breast cancer screening or treatment.
The findings show that CNN demonstrated significantly higher accuracy, sensitivity, and specificity than the benchmark system. The CNN also improved the detection of invasive cancers by 26 percent.
“Our work suggests that AI can aid clinicians by decreasing the number of false-positive findings, and decreasing the time clinicians need to spend looking at these images.” The researchers note that a limitation of their study is the lack of follow-up on patients who were initially found with malignancies by means other than ultrasound.
“In addition, we did not have an opportunity to perform breast biopsy on these patients,” they said. Therefore, we are not certain whether the AI system can predict biopsy outcomes as accurately as it does imaging diagnoses.
Future work should investigate this question by leveraging large-scale datasets containing all clinical information on breast cancer cases including mammography and ultrasound findings with biopsy results.