Tour of HTT Data
- Presenter: Brandon D Gallas, Ph.D.
- When: December 17, 2021
- Format: Webinar hosted by the HTT project.
- Slides: 20211217-tourHTTdata.pdf (807 KB, uploaded by Brandon D. Gallas 1 year 3 months ago)
The goal of the High-Throughput Truthing (HTT) project is to produce a validation dataset established by pathologist annotations for artificial intelligence algorithms analyzing digital scans of pathology slides: data (images + annotations). This video contains a tour of the project’s github repository and how to analyze the pilot study data.
- The first half of the tour will be basics about explaining, accessing, and using the data, including basics about R, R Studio, git, and github.
- The second half of the tour will discuss the agreement analysis. The data collected from pathologists include a binary classification decision, and if positive, a density estimate. Because the density estimates are from humans, the data is challenging to model. To overcome this challenge, we will consider methods that treat the data as ordinal instead of ratio. The agreement analysis is still a work in progress, and feedback will be encouraged.
For more information on the HTT project and how to get involved, visit our eeDAP wiki homepage: https://ncihub.org/groups/eedapstudies
To see the pilot study data, visit our github repository: https://github.com/DIDSR/HTT