Using Progressive Context Encoders for Anomaly Detection | AI in Healthcare | Generative Modeling

Keywords: Digital Pathology, Anomaly Detection, Melanoma
About this event
Abstract
Whole slide imaging (WSI) is transforming the practice of pathology, converting a qualitative discipline into a quantitative one. However, one must exercise caution in interpreting algorithm assertions, particularly in pathology where an incorrect classification could have profound impacts on a patient, and rare classes exist that may not have been seen by the algorithm during training. A more robust approach would be to identify areas of an image for which the pathologist should concentrate their effort to make a final diagnosis. This anomaly detection strategy would be ideal for WSI but given the extremely high resolution and large file sizes, such an approach is difficult. Here, we combine progressive generative adversarial networks with a flexible adversarial autoencoder architecture capable of learning the “normal distribution” of WSIs of normal skin tissue at extremely high resolution and demonstrate its anomaly detection performance. Our approach yielded pixel-level accuracy of 89% for identifying melanoma, suggesting that our label-free anomaly detection pipeline is a viable strategy for generating high quality annotations - without tedious manual segmentation by pathologists. The code is publicly available at github.com/Steven-N-Hart/P-CEAD.
Paper : www.biorxiv.org/content/10.11...
About the Author :
Quincy Gu is a 2nd -year predoctoral student at the Mayo Clinic College of Medicine and Sciences in biomathematics, bioinformatics, and computational biology. He also holds his bachelor’s degree from the University of Minnesota in mathematics. With primary interests in medicine and engineering, he served as an undergraduate research assistant at the University of Minnesota Medical School, where he did research on colorectal cancer patients’ survival analysis and breast cancer MRI image processing. His previous research experience convinced him that he could be a dedicated medical researcher in the future by pursuing a doctoral degree.
After joining Mayo Clinic, he has focused his research on correlating genomic signatures with histologic features from H&E slides to reduce the number of unnecessary testing procedures and identify patients who would benefit from genetic predisposition testing. With special interest in melanoma, he is developing artificial intelligence algorithms in the new subfield of digital pathology, which aims to provide automatic melanoma diagnosis and individualized therapeutic options.
Quincy is looking forward to continuing his education in the medical doctoral training program after graduating from the PhD program. He is particularly interested in being a physician-scientist to provide patient care and contribute to skin cancer research as his long-term career goal.
The code is open-source and we would be doing a quick walkthrough on how to get results as well.
The code is publicly available at github.com/Steven-N-Hart/P-CEAD
Paper : www.biorxiv.org/content/10.11...