In this episode of Data Skeptic, our host Kyle Polich is joined by Gabriel Maicas, a PhD candidate at the University of Adelaide, to discuss machine learning systems that can be used by radiologists to improve their accuracy and speed of diagnosis.
Thanks to The Great Courses Plus for sponsoring this episode.
Medical imaging is a highly effective tool used by clinicians to diagnose a wide array of diseases and injuries. However, it often requires exceptionally trained specialists such as radiologists to interpret accurately. AI techniques such as deep learning have revolutionized image analysis, and are expanding the reach and improving the and efficiency and quality of medical imaging.
Radiologists are typically trained to solve tasks of increasing difficulty, examining small sets of practice images for each task. Meanwhile, most machine learning-based medical image analysis systems require much larger training image sets and are modeled to solve specific, yet complex classification problems. Given this discrepancy in training approaches between radiologists and ML-based systems, Gabriel and his colleagues recently proposed a novel training approach based on the training program that radiologists undergo.
Gabriel and his colleagues leveraged meta-training that models a classifier based on a series of tasks. His team suggests that a meta-training approach can be used to pre-train medical image analysis models.