Summary
This collection consists of Computed Tomography (CT) images of the mediastinum and abdomen in which lymph node positions are marked by radiologists at the National Institutes of Health, Clinical Center. Radiologists at the Imaging Biomarkers and Computer-Aided Diagnosis Laboratory labeled a total of 388 mediastinal lymph nodes in CT images of 90 patients and a total of 595 abdominal lymph nodes in 86 patients.
The collection is aimed at the medical image computing community for developing and assessing computer-aided detection methods. Automated detection of lymph nodes can be an important clinical diagnostic tool but is very challenging due to the low contrast of surrounding structures in CT and to their varying sizes, poses, shapes and sparsely distributed locations. This data set is made available to make direct comparison to other detection methods in order to advance the state of the art.
Studies that have used this collection include the following (Please cite in any publications resulting from using this data):
References
- Roth, Holger R and Lu, Le and Seff, Ari and Cherry, Kevin M and Hoffman, Joanne and Wang, Shijun and Liu, Jiamin and Turkbey, Evrim and Summers, Ronald M. A new 2.5 D representation for lymph node detection using random sets of deep convolutional neural network observations. Medical Image Computing and Computer-Assisted Intervention--MICCAI 2014, p520-527, 2014. (link)
- Seff, Ari and Lu, Le and Cherry, Kevin M and Roth, Holger R and Liu, Jiamin and Wang, Shijun and Hoffman, Joanne and Turkbey, Evrim B and Summers, Ronald M. 2D view aggregation for lymph node detection using a shallow hierarchy of linear classifiers. Medical Image Computing and Computer-Assisted Intervention--MICCAI 2014, p544-552, 2014. (link)
- The Digital Object Identifier http://dx.doi.org/10.7937/K9/TCIA.2015.AQIIDCNM should also be used to refer to this data set.
Acknowledgements
We would like to acknowledge the individuals and institutions that have provided data for this collection:.
National Institutes of Health, Bethesda MD. Special thanks to Dr. Holger R. Roth and Dr. Ronald Summers, Imaging Biomarkers and Computer-Aided Diagnosis Laboratory, Grant Magnuson Clinical Center.
Data Access
Imaging Data
You can view and download these images on TCIA by clicking 
and selecting the CT Lymph Nodes collection. You can also access the data via it's Digital Object Identifier page: http://dx.doi.org/10.7937/K9/TCIA.2015.AQIIDCNM.
Collection Statistics | (updated 2015/03/16) |
|---|---|
Modalities | CT |
Number of Patients | 176 |
Number of Studies | 176 |
Number of Series | 176 |
Number of Images | 110,103 |
| Images Size (GB) | 57.8 |
If you are unsure how to download this collection, please view Searching by Collection or refer to TCIA's User's Guide for more detailed instructions on using the site.
Related Data
Annotation files: MED_ABD_LYMPH_ANNOTATIONS.zip (new 6/24/2015). The annotations include a folder for each case with text files of voxel indices, physical coordinates, size measurements and a MITK point set file (.mps), which can be visualized using the MITK workbench (Note: only release 2014.10.0 and later supports visualization of point set files using the "point set interaction plugin"). Abdominal size measurements include the longest and shortest axis in axial view of a lymph node. The shortest axis is used for the RECIST criteria. The mediastinal set only includes the shortest axis.
The DICOM files were created from volumetric images (Analyze and NifTI) using this from ITK:
Computer-generated candidate detections for mediastinal and abdominal lymph nodes (produced by methods in [K. Cherry et al., SPIE Med. Img. 2014] and [J. Liu et al., SPIE Med. Img. 2014]]). See attached: MED_ABD_LYMPH_CANDIDATES.zip (new 9/14/2015).