Jayashree Kalpathy-Cramer, Sandy Napel, Dmitry Goldgof, Binsheng Zhao. (2015). Multi-site collection of Lung CT data with Nodule Segmentations. The Cancer Imaging Archive. http://doi.org/10.7937/K9/TCIA.2015.1BUVFJR7
This dataset (also known as the “moist run” among QIN sites) contains CT images (41 total scans) of non-small cell lung cancer from: the Reference Image Database to Evaluate Therapy Response (RIDER), the Lung Image Database Consortium (LIDC), patients from Stanford University Medical Center and the Moffitt Cancer Center, and the Columbia University/FDA Phantom. In addition, 3 academic institutions (Columbia, Stanford, Moffitt-USF) each ran their own segmentation algorithm on a total of 52 tumor volumes. Segmentations were performed 3 different times with different initial conditions, resulting in 9 segmentations formatted as DICOM Segmentation Objects (DSOs) for each tumor volume, for a total of 468 segmentations. This collection may be useful for designing and comparing competing segmentation algorithms, for establishing acceptable ranges of variability in volume and segmentation borders, and for developing algorithms for creating cancer biomarkers from features computed from the segmented tumors and their environments.
Kalpathy-Cramer, J., Zhao, B., Goldgof, D., Gu, Y., Wang, X., Yang, H., … Napel, S. (2016, February 3). A Comparison of Lung Nodule Segmentation Algorithms: Methods and Results from a Multi-institutional Study. Journal of Digital Imaging. Springer Nature. http://doi.org/10.1007/s10278-016-9859-z
Note: In December 2018 it was discovered that an update to NSCLC Radiogenomics mistakenly resulted in the deletion of the segmentation data for this analysis set. We are currently investigating whether it is possible to restore the data. In the meantime this dataset can be downloaded using the links below which exclude the Stanford NSCLC Radiogenomics subset of the analyses.
NOTE: On 9/14/2015 this DOI was updated to resolve problems with 9 of the segmentations being incorrectly labeled. The Series Instance UIDs in the original data set which have since been deleted from TCIA are:
These have been replaced with the following new segmentation series: