Can Robots Read Your CT Scan?

Can Robots Read Your CT Scan? Posted By:
...

At a plenary session of the World Conference on Lung Cancer in September, 2019, in Barcelona, Spain, Hugo Aerts, PhD presented fascinating data on the use of artificial intelligence (AI) in clinical practice. Professor Aerts has utilized "deep-learning," which consists of very intense programming of computers or robots with numerous real-life examples. In the past, these strategies have not been sophisticated enough to have been utilized in medical situations.

Professor Aerts discussed his work on developing the ability of AI to read and draw conclusions from CT scans evaluating tumors in the lung. In his lab in Boston, in conjunction with Harvard and the Dana Farber Cancer Institute, he develops the AI by entering numerous CT scans with varieties of lung tumors to "teach" the machine to recognize lung tumors in the images. When putting the AI computers to the test, he found that the AI algorithm was more accurate than final-year radiology residents in 90% to 95% of cases. He emphasized that there is no engineering, he simply gives the AI many tumor examples and the computer learns to identify them automatically.

He went on to demonstrate that while evaluating for tumors, the technology can also describe and diagnose cardiac complicationsresulting in a prediction model that could accurately predict the risk of heart attack. Currently, standard radiology interpretation of CTs of the chest does not go into such detail unless the cardiac disease is blatantly obvious, for time and cost reasons. But in this case, if the radiology is done using a computer or robot, it performs the risk-assessment task automatically.

Lastly, he revealed his plans for radiation oncology dose-planning. Again, by implementing a number of real-life radiation plans, over time, the AI computer can begin to develop a safe and effective radiation plan. All of these AI examples could potentially save costs and reduce human error; however, controversy remains over the human aspect as well. A trained clinician will interpret scans or radiation doses while taking account of other conditions that may not be considered by the AI device. Plus, there is always the idea that this is taking away jobs. Much more data and studies will need to be completed before this type of technology could become mainstream.

References
  • Xu Y, Hosny A, Zeleznik R, et al. Deep learning predicts lung cancer treatment response from serial medical imaging. Clin Cancer Res. 2019;25:3266-3275.

Share

Filed under: Miscellaneous , Oncology/Hematology

Related
Artificial Intelligence in Healthcare: Our Next Great Innovation, Part II—AI’s Use in Technology and the APP Impact

Artificial Intelligence in Healthcare: Our Next Gr ...

Most of us have heard someone invariably shout out, “Hey, Siri!” or “Alexa . . . !” These ex ...

Filed under: Health Policy and Trends, Miscellaneous


Continue Reading
Autonomy and Motivation for the Healthcare Workforce

Autonomy and Motivation for the Healthcare Workfor ...

I recently discovered the New England Journal of Medicine podcast Not Otherwise Specified. In full t ...

Filed under: Health Policy and Trends, Miscellaneous, Practice Management/Career, NPs & PAs


Continue Reading
Implicit vs Explicit Bias in Healthcare: A Crash Course

Implicit vs Explicit Bias in Healthcare: A Crash C ...

We all have it—bias, that is. It affects how we interact with others and our relationships, includ ...

Filed under: Miscellaneous, Practice Management/Career, NPs & PAs


Continue Reading
When a Window Is More Than Just Something We Look Through

When a Window Is More Than Just Something We Look ...

I am sure you have heard the phrase “window of time” applied to many different aspects of life. ...

Filed under: Miscellaneous, NPs & PAs


Continue Reading
Artificial Intelligence in Healthcare: Our Next Great Innovation

Artificial Intelligence in Healthcare: Our Next Gr ...

In 1935, the great British mathematician Alan Turing described a math machine that had limitless cap ...

Filed under: Health Policy and Trends, Miscellaneous, NPs & PAs


Continue Reading
Primary Hyperoxaluria: Pathophysiology, Diagnosis, and New RNAi Treatment Approaches

Primary Hyperoxaluria: Pathophysiology, Diagnosis, ...

PH PathophysiologyPrimary hyperoxalurias (PHs) are rare and caused by a genetic defect of glycolate ...

Filed under: Miscellaneous


Continue Reading