AI and Oncology: An Unexpected but Useful Pairing

Reading time: 4 minutes

Susan Egbert

Did you know that researchers are trying to integrate artificial intelligence (AI) with cancer treatments? There are plenty of studies in oncology available that range from diagnostic to treatment that are peer-reviewed (experts looking at the articles before publishing). This is useful for AI as the more data that is available (examples being large public datasets), the more that the AI can be developed into something useful for researchers. Traditionally, clinical research has been mostly using statistics as a way to explain the data. However, AI takes the data, tries to find patterns, and makes prediction patterns based on it. Researchers today are working with AI in order to develop more accurate treatment scenarios to help guide decisions on how to best proceed with a patient’s therapy options.  

So far the role of AI has been seen in cancer detection, screening, diagnosis, classifications, cancer genomics, tumor microenvironment, and biomarkers1. Primarily, most of the AI technology is seen in screening for cancers like lung and breast cancer that are highly common across the population2. The first FDA-approved AI device was for cardiovascular purposes (ECG to detect abnormal heart rhythms) in 20142, with more oncology AI devices likely being approved especially for diagnostic purposes with AI making headway in image analysis. AI treatment devices for oncology will probably gain more popularity with oncologists once there is enough data for AI to make the appropriate and accurate algorithms to treat patients.

There are two concepts that are necessary to know in order to understand AI a bit more. These are machine learning and deep learning. Machine learning is where the machine is able to learn and improve patterns and models of analysis. Deep learning is the concept where the machine-learning method utilizes complex and deep networks to finalize a highly predictive model. Both machine learning and deep learning are central in the AI revolution of the management of cancer patients.

Figure 1: From Figure 1 of C. Luchini et al, 2021.  This figure talks about the current use of AI in devices in a while the types of cancer AI is used in is in b

Traditional mammograms have a high false-positive/negative reading and involve intense work by healthcare professionals to rule these readings out from the true reading.   AI-assisted mammograms have been able to cut these false positives/negatives by about 69%3,4,5. Reducing false readings will allow practitioners to evaluate the presence of cancer, especially distinguishing some of the more challenging cases that might be presented. In colon cancer polyps,  the use of AI classifications of lesions based on CT colonography found that flat colonic lesions could be detected more accurately7. With AI assistance, the possibility of relying on less invasive methods would probably be seen in the near future for such things as detecting colon cancer. This would allow patients to not need a colonoscopy, which is generally not a pleasant experience for the patients.

Overall, Hamamoto et al, 2020 reviewed all the FDA-approved AI devices (61) and summarized what each of them did and the trends of these devices. Luchini et al, 2021 investigated all the FDA-approved AI-associated or associable devices in cancer. In total, they looked through 71 and made an overview of what devices did which function. In the span of a year Hamamoto’s review to Luchini’s review shows that this year span has had fewer FDA-approved devices (10) in comparison to some of the previous years (24 in 2019). This trend might account for the need to test new areas. Data will still need to be collected over a span of many years before there will be an extremely precise and accurate way for AI to be able to detect cancer less invasively. However, AI is making a great start into the world of oncology and we can expect more of it to come into the near future.

Edited by Kate Secombe

Works mentioned:

  1. Luchini, Claudio, Antonio Pea, and Aldo Scarpa. “Artificial intelligence in oncology: current applications and future perspectives.” British journal of cancer 126.1 (2021): 4-9.
  2. Hamamoto, Ryuji, Kruthi Suvarna, Masayoshi Yamada, Kazuma Kobayashi, Norio Shinkai, Mototaka Miyake, Masamichi Takahashi et al. “Application of artificial intelligence technology in oncology: Towards the establishment of precision medicine.” Cancers 12, no. 12 (2020): 3532.
  3. Schaffter, T.; Buist, D.S.M.; Lee, C.I.; Nikulin, Y.; Ribli, D.; Guan, Y.; Lotter, W.; Jie, Z.; Beng, H.D.; Wang, S.; et al. Evaluation of Combined Artificial Intelligence and Radiologist Assessment to Interpret Screening Mammograms. JAMA Netw. Open 2020, 3, e200265.
  4. Batchu, S.; Liu, F.; Amireh, A.; Waller, J.; Umair, M. A Review of Applications of Machine Learning in Mammography and Future Challenges. Oncology 2021, 99, 483–490. 
  5. Rodriguez-Ruiz, A.; Lång, K.; Gubern-Merida, A.; Teuwen, J.; Broeders, M.; Gennaro, G.; Clauser, P.; Helbich, T.H.; Chevalier, M.; Mertelmeier, T.; et al. Can we reduce the workload of mammographic screening by automatic identification of normal exams with artificial intelligence? A feasibility study. Eur. Radiol. 2019, 29, 4825–4832.
  6. Vobugari, Nikitha, Vikranth Raja, Udhav Sethi, Kejal Gandhi, Kishore Raja, and Salim R. Surani. “Advancements in Oncology with Artificial Intelligence—A Review Article.” Cancers 14, no. 5 (2022): 1349.
  7. Taylor, S.A.; Iinuma, G.; Saito, Y.; Zhang, J.; Halligan, S. CT colonography: Computer-aided detection of morphologically flat T1 colonic carcinoma. Eur. Radiol. 2008, 18, 1666–1673.

Image Credits: FDA.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Create a website or blog at WordPress.com

Up ↑

%d bloggers like this: