ML For Mitosis Detection from Big Medical Images

A Large Scale Deep Learning Approach for the Mitosis Detection from Big Medical Images

The strongest indicator of a cancer patient's prognosis is the number of mitotic bodies that a pathologist manually counts from the high-resolution whole-slide histopathology images. It is challenging to automate the process of mitosis detection due to the limited training datasets and the intensive computing involved in the model training and inference. This presentation introduces a large-scale deep learning approach to train a two-stage CNN-based model with high accuracy to detect the mitosis locations directly from the high-resolution whole-slide images. The whole pipeline, including data preprocessing, model training, hyperparameter tuning, and inference, is parallelized by utilizing the distributed TensorFlow, Apache Spark, and HDFS. The experiences and techniques in this project can be applied to other large scale deep learning problems as well.

Speaker : Fei Hu (IBM)

Fei Hu is a staff software engineer at IBM Center for Open-Source Data and AI technologies (CODAIT). His work focuses on deep learning and big data frameworks. He is passionate about open source technologies with active contributions and published more than 15 papers about big array-based data management and mining.