H.R. Tizhoosh
University of Waterloo

Vision : Images : Brain



Cancer Imaging Archive provides a variety of medical image data.



» News

» Free Inquiries

» Publications

» Image Data

» Source Code

» Students

» Links

» Supporters

» Partners

» Resume

» Contact


:: University of Waterloo

:: KIMIA Lab

:: Centre for Bioengineering and Biotechnology (CBB)


  Home Research Teaching KIMIA Lab  

Benchmark Image Datasets

Breast Ultrasound

These are 35 breast ultrasound scans collected from Web. We do not own them. All images have been segmented by an image-processing expert with extensive experience in breast lesion segmentation. Hence, the segments may not
perfectly comply with anatomical knowledge in a clinical context. We have not classified the images but most of them “appear” to be cysts (benign, e.g. images numbers 3,5 and 19), and some of them “seem” to be tumors (malignant, e.g. image numbers 6, 8 and 28).

All images and their manual segmentations have been saved in TIFF format.

Note: Make sure that segments are binary when you read them (e.g. in Matlab via BIN = logical(imread(‘1G.tif’));)

Breast Ultrasound Segmentation Benchmark

Please cite the following paper when using these breast ultrasound images:

Download Breast Ultrasound Images with Manual Segments

Also see:


© H.R. Tizhoosh, Systems Design Engineering, University of Waterloo, Canada, 2001-2015