Cancer Screening using Deep Learning
summary
Arterys automates cardiac MRI segmentation on ssfp scans by running convolutional nets in the cloud and handling partially labeled clinical data. The pipeline uses TensorFlow on AWS GPUs and trains a single network with separate outputs per contour, ignoring missing masks in the loss. Using ~100-case validation, the right-ventricle median volume error drops from ~30% to ~10% when all data are used. Models currently train on 2D frames; 3D/time extensions are under exploration.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 0:49How Arturis works
- 1:42Cardiac Anatomy
- 3:05Cardiac MRI
- 7:15Neural Network Architecture
- 10:20Architecture
- 11:49Results
- 14:00Conclusion
- 14:34Questions
related talks
34mWhy AI in the Medical Field Both Sucks and Rocks...
11mInterview
29mCancer diagnostics enabled by machines: finding and classifying single tumor....
43mBuilding AI That Learns and Adapts: A Case Study in MRI Diagnostics
49mBuilding AI That Learns and Adapts: A Case Study in MRI Diagnostics
55mDeep Learning in Medical Image Diagnostics
24mMachine Learning & Medical Imaging: The Future of Early Cancer Detection
15mBrain Tumor Detection and Classification Using Deep Learning Models