York researchers from the Lassonde School of Engineering and Sunnybrook Hospital have teamed up to reveal a potential solution to a clinical issue regarding breast cancer treatment.
The study was published February 10 in Nature’s journal Scientific Reports by associate professor Ali Sadeghi-Naini, who also serves as York research chair in the electrical engineering department. The study was also done in collaboration with Sadeghi-Naini’s graduate students Hamidreza Taleghamar and Seyed Ali Jalalifar.
The study highlighted a methodology that could assist in allowing physicians to predict the effectiveness of chemotherapy for individual breast cancer patients, prior to or immediately following treatment.
As reported by YFile in February, “The study’s methodology involves an AI program designed to learn how to make minute distinctions between Quantitative Ultrasound (QUS) parametric images of cancerous tumours, at a volume and level of detail that would be impossible for human beings,” continuing on to say that the study builds upon previous research on QUS parametric imaging.
Dr. Christo El Morr, PhD, associate professor of health informatics at York’s school of health policy and management, shares their insights on the use of AI to improve healthcare and how great of an impact it can have on the future of cancer screening.
“There is potential for AI to improve health, including healthcare delivery, but it’s good to be cautious about the impact of any technology in any domain. Machine-learning algorithms learn from past data, and the latter reflects our social biases (e.g., gender, racial). We need in-depth multidisciplinary studies to identify AI and mitigate unintended undesirable consequences of AI impact.
“Most importantly, as a society, we need to define what is desirable, and that is a vital ethical question for the development of a humanist future.”
Sadeghi-Naini recognized that breast cancer screenings, while being a key determinant in early detection and treatment, require many clinical resources— especially time.
“AI platforms can assist clinicians in rapid and accurate annotation of images to streamline the breast cancer screening workflow. The role of an AI platform in this application is currently defined as helping the expert radiologists in rapid review and interpretation of images. However, one can envision that such platforms may eventually be able to generate radiology reports for most of the routine images and only triage the complicated cases for radiologist review.”
Sadeghi-Naini further adds that current AI programs save valuable time by identifying and quantifying patterns humans can’t see, as well as monitor treatment effectiveness.
“Some of these patterns, identified from large-scale imaging datasets, may potentially be correlated to specific characteristics of breast cancer tumours in terms of microstructure, metabolisms and physiology that are linked to the level of aggressiveness and responsiveness of the tumour to chemotherapy.
Sadeghi-Naini continues on to say that data-driven AI models “discover such correlations through complicated computational methods and apply them to predict the therapy response in new patients.
“AI and quantitative imaging techniques can also be applied early after the start of a prescribed treatment to assess whether the tumour is responding and to predict the therapy outcome. Such predictive models can help clinicians in adjusting the standard treatment for individual patients, or even switch to alternative treatments before it is potentially too late.”
Hoping that the use of AI in healthcare can assist in creating a more equitable system and help clinicians with their tasks as well, Sadeghi-Naini says, “AI can facilitate equity in access to healthcare services in the society at large by streamlining the current clinical workflows, and permitting more efficient allocations of clinical resources.”
