NVIDIA AI-Assisted Annotation SDK
With the NVIDIA AI-Assisted Annotation SDK, radiologists only need to give approximate points of interest, instead of hand drawing an entire region to detect an organ or an abnormality. Using deep learning techniques, the AI-Assisted Annotation SDK takes these approximate points as input along with the 3D volume data and returns an auto-annotated set of slices. Auto-annotation step is achieved using NVIDIA’s pre-trained deep learning models for different organs. In this way, neither the application developers nor the radiologists need any prior deep learning knowledge, but benefit from NVIDIA’s deep learning expertise out of the box.
The AI-Assisted Annotation SDK will include a Python-based flask server, REST API, documentation and integration examples for all the features.
- Assisted Annotation - Takes approximate points as input and returns a list of slices already annotated for different organs.
- Performance Speedup - Achieve 4x-10x speedup compared to manual editing depending on the organ being segmented.
- Polygon Editing - Reduces time in correcting annotation results by suggesting organ boundaries automatically.
- Transfer Learning Workflow - Workflows to enable labelling and training pipeline for improving annotation accuracy of pre-trained models using NVIDIA’s Transfer Learning Toolkit.
AI-Accelerated Annotation SDK includes NVIDIA’s Transfer Learning Toolkit for seamless labelling and training workflow integration consistently improving accuracy over time with additional data.
We were able to get our hands on NVIDIA’s AI Assisted Annotation technology and integrate it into our viewer in a couple of days’ time. We currently annotate a lot of images - sometimes on the order of 1000 or more a day, so any technology that can help automate this process could potentially have a significant impact in reducing the time and cost of annotation. We are excited to leverage the AI assisted workflows and work with NVIDIA to solve these critical medical imaging problems.Mark Michalski, Executive Director at MGH & BWH Center for Clinical Data Science