Laboratory for Knowledge Inference in
Medical Image Analysis

  Home About us Research The Team News Code and Data  

Pathology Images: KIMIA Path24

The KIMIA Path24 is a data set that was proposed in the following paper:

Classification and Retrieval of Digital Pathology Scans: A New Dataset
Morteza Babaie, Shivam Kalra, Aditya Sriram, Christopher Mitcheltree, Shujin Zhu, Amin Khatami, Shahryar Rahnamayan, H.R. Tizhoosh. CVMI Workshop @ CVPR 2017

We had 350 whole scan images (WSIs) from diverse body parts at our disposal. The images were captured by TissueScope LE 1.0. The scans were performed in the bright field using a 0.75 NA lens.

We manually selected 24 WSIs purely based on visual distinction for non-clinical experts which means, in our selection, we made conscious effort to select a subset of the WSIs such that they clearly represent different texture patterns.

There are 1325 patches of size 1000x1000 pixels. The training data can be extracted form the scans in any way appropriate for specific algorithms. With preset values, approximately 27,000 training patches of size 1000x1000 pixels can be extracted.

We calculate two accuracies: patch-t0-scan accuracy, and whole-scan accuracy. The total accuracy is the product of these two values.


If you are interested in using the KIMIA Path24 data set, please sign this agreement and send it to Hamid Tizhoosh (last_name AT uwaterloo DOT ca). Note for Students: The agreement has to be signed by your department/supervisor.

:: Download the paper

:: Also see: KIMIA Path24 Readme file

:: Our Sponsors

KIMIA Path24



© KIMIA Lab, University of Waterloo, Canada, 2013-2019 - Using Artificial Intelligence to Infer Knowledge from Medical Images