Artificial Intelligence Assists in Monitoring Breast Cancer Diagnosis for Radiologists
Artificial intelligence (AI) is proving to be a valuable tool in the field of radiology, particularly in the detection of breast cancer through mammograms. A recent study, originally published by Cosmos, has revealed that the use of AI significantly improves the visual search patterns and efficiency of radiologists in identifying breast cancer.
The study involved 12 radiologists who read mammography examinations for 150 women, 75 with breast cancer and 75 without. The AI tool provided a "level of suspicion" score that influenced how radiologists approached each case. When AI scores were low, indicating low suspicion, radiologists moved more quickly through normal cases. Conversely, high AI suspicion scores prompted them to conduct more careful and detailed examinations of challenging or subtle findings.
One of the key benefits of AI integration is the enhanced focus on suspicious areas. Eye-tracking studies reveal that radiologists equipped with AI decision support spend more time examining regions of mammograms that contain lesions. This targeted attention leads to better detection accuracy, reducing the likelihood of missed cancers without increasing the overall reading time per case.
Radiologists assisted by AI also demonstrated higher sensitivity in cancer detection compared to unaided reading. The AI tool acts as a guide that highlights areas of potential concern, allowing radiologists to recognize lesions more effectively.
The study aimed to investigate the impact of AI on the search patterns of radiologists, which was previously unknown. The eye tracking data showed that radiologists with AI support spent more time examining the regions that contained lesions, demonstrating a more targeted and efficient approach.
However, researchers emphasize that radiologists must be properly educated to critically interpret AI outputs to avoid overreliance, which could lead to unnecessary additional imaging or missed diagnoses. Radiologists remain ultimately responsible for final decisions.
The use of AI in mammography interpretation carries a relatively low risk of incorrect information, but radiologists should feel accountable for their own decisions. A mammogram, an X-ray of the breast used to detect abnormalities, is one of the most effective ways to detect early signs of breast cancer.
In conclusion, AI acts as an effective decision support tool that guides radiologists’ visual search, increasing efficiency and accuracy in breast cancer detection on mammograms through enhanced lesion detection and adaptive reading strategies without adding to the time burden. Further studies are being conducted to develop methods to predict if AI is uncertain about a decision, aiming to improve both the performance and efficiency in breast cancer screening.
AI's integration in radiology, particularly for breast cancer detection, significantly improves the efficiency of radiologists. The study revealed that radiologists focusing on AI-suggested suspicious areas led to better detection accuracy, reducing missed cancers without increasing reading time. However, for effective use, radiologists need to critically interpret AI outputs to avoid overreliance and maintain their accountability in final decisions. AI is proving to be a valuable tool in breast cancer screening through efficient and effective mammogram interpretations, but further studies are needed to predict AI's uncertainty and improve performance in breast cancer screening.
