TCGA-SBU-TIL-Maps: AI-derived Tumor Infiltrating Lymphocyte maps for the TCGA collections

This collection contains the spatial distributions and patterns of tumor-infiltrating lymphocytes (TILs) computed at the patch level in whole slide tissue images from a subset of cancer types in TCGA. The spatial distributions and patterns  were computed automatically using two distinct AI algorithms developed by the Joel Saltz lab at Stony Brook University. The collection includes both the binary classification (TIL-positive or not) and the prediction probability generated by the AI model for each included whole slide image.  Each image in the collection was partitioned into a uniform grid of 50x50 square micron image patches; each patch was assigned a classification probability by the AI models. 

The computational results  included in this dataset were produced using the algorithms described in [1,2]. The data generated using the algorithm described in [1] were shared earlier on The Cancer Imaging Archive [3].

This dataset contains the annotations harmonized into DICOM Segmentation representation.

This dataset is available from the NCI Imaging Data Commons (IDC), and can be explored interactively in the IDC Portal using this link: https://portal.imaging.datacommons.cancer.gov/explore/filters/?analysis_results_id=TCGA-SBU-TIL-Maps

Specific files included in the record are organized in the manifests by the individual TCGA collections that were processed. The suffix of the manifest indicates its content, which is the list of pointers to the public Google Cloud Storage (GCS) or Amazon Web Services (AWS) buckets containing the files included in the collection:

1. -gcs.s5cmd: GCS-based manifest (to download the files described in the manifest, execute this command: pip install --upgrade idc-index && idc download manifest).

2. -aws.s5cmd: AWS-based manifest (to download the files described in the manifest, execute this command: pip install --upgrade idc-index && idc download manifest).

3. -dcf.dcf: Gen3-based manifest (see details in https://learn.canceridc.dev/data/organization-of-data/guids-and-uuids). 

[1] Saltz, Joel, Rajarsi Gupta, Le Hou, Tahsin Kurc, Pankaj Singh, Vu Nguyen, Dimitris Samaras et al. "Spatial organization and molecular correlation of tumor-infiltrating lymphocytes using deep learning on pathology images." Cell reports 23, no. 1 (2018): 181-193.

[2] Abousamra, Shahira, Rajarsi Gupta, Le Hou, Rebecca Batiste, Tianhao Zhao, Anand Shankar, Arvind Rao et al. "Deep learning-based mapping of tumor infiltrating lymphocytes in whole slide images of 23 types of cancer." Frontiers in oncology 11 (2022): 806603.

[3] Saltz, J., Gupta, R., Hou, L., Kurc, T., Singh, P., Nguyen, V., Samaras, D., Shroyer, K. R., Zhao, T., Batiste, R., Van Arnam, J., The Cancer Genome Atlas Research Network, Shmulevich, I., Rao, A. U. K., Lazar, A. J., Sharma, A., & Thorsson, V. (2018). Tumor-Infiltrating Lymphocytes Maps from TCGA H&E Whole Slide Pathology Images [Data set]. The Cancer Imaging Archive. https://doi.org/10.7937/K9/TCIA.2018.Y75F9W1