Talk / Overview

I will present Genentech efforts to automate the FDA-accepted radiological standard metric quantifying treatment response of solid tumors: RECIST. Two Deep Learning approaches will be discussed: one relying on human-segmented lesions for training and the other, a weakly-supervised approach, not requiring such individual lesion segmentations. The automation will speed out drugs-to-approval time as well as help physicians to better match patients to therapies.

Talk / Speakers

Yury Petrov

Principle Data Scientist, Genentech

Talk / Highlights


Automated Tumor Burden Assessment from CT-scans

With Yury PetrovPublished March 12, 2020

AMLD / Global partners