IMAIOS, a company experienced with developing new tools to help healthcare professionals in their daily practice, is very proud to announce its partnership with Duke University School of Medicine.
Together, IMAIOS and DUSM developed an online training application that allows any physician to practice and progress with the annotation and diagnosis of breast scans, using a database of 200 very high-resolution digital breast tomosynthesis (DBT) images. The users of this free web-based tool can use a simple medical image viewer to select where they believe breast lesions lie within such scans, and then compare their results with the actual location of the lesions.
To supplement IMAIOS’s application, Duke University School of Medicine developed an algorithm which identifies and remembers the annotation behavior and potential errors of any physician who uses this tool. As more users practice with this tool, the algorithm can learn to predict future erroneous annotations by users. Using this interface regularly, doctors can self-assess the growth of their skills over time, and better understand their own diagnostic methodology.
This tool will not only allow physicians to self-assess, but will also be a great way to understand the process that leads to diagnosis as more data is collected from the physicians using the application. In turn, elucidating this diagnostic process could lay the groundwork for further research in understanding how to automate it, with artificial intelligence and machine learning.