search

wentaozhu / deep-mil-for-whole-mammogram-classification

Zhu, Wentao, Qi Lou, Yeeleng Scott Vang, and Xiaohui Xie. "Deep Multi-instance Networks with Sparse Label Assignment for Whole Mammogram Classification." MICCAI 2017.
MIT License
113 stars 36 forks source link

Benign/Malignant labels in original InBreast dataset #17

Closed akbwaj closed 5 years ago

akbwaj commented 5 years ago

I have received the original InBreast dataset from the author, but can not find any description of Benign/Malignant labels in there. I do, however notice a label.txt file that you are using to generate labels. May I know where is this file generated from and what do the labels mean?

wentaozhu commented 5 years ago

https://github.com/wentaozhu/deep-mil-for-whole-mammogram-classification/blob/master/inbreast.py

readlabel() function is the one to turn the label to benign and malignant.

Generally, if the score is bigger than 3, it is malignant. Otherwise it is benign. You can find it from the paper.

Deep Multi-instance Networks with Sparse Label Assignment for Whole Mammogram Classification. MICCAI 2017.

akbwaj commented 5 years ago

Thank you for the information. I am actually looking for the Benign/Malignant findings after the biopsy. These are not provided anymore as part of the dataset. Thanks again!

yawwG commented 3 years ago

Actually, we can find the biopsy result from the medical report, then pick the corresponding label.