Standard 25k Selection

A standardized selection of 25,000 participants, optimized to meet the requirements of most analyses, is available to streamline the selection process. This Standard 25k Selection is available via download with an approved CDAS project with a fully executed Data Transfer Agreement (DTA) and includes:

• All participants with a T0-T5 lung cancer (n=585) or T0-T5 non-target cancer (n=12)
• All participants with any abnormal/suspicious screens (n=10,152)
• Randomly selected participants with any abnormal/not suspicious screens (n=7,781)
• Randomly selected participants with all negative screens (n=6,470)

Due to the size of the images for 25,000 participants (~89,000 images totaling 850GB ), x-ray images are organized into 12 batches (ranging in size from 30 - 83.5 GB) to facilitate download. By default, the batches are arranged in equally-sized training and testing sets comprised of 6 batches each. The training and testing sets have been created to have equal numbers of cases as well as controls.

For investigator convenience, the first batch of images from the training and test batches have been set aside as a screen-based case/control set (~6,000 images totaling ~60GB). The Case-Control Subset provides investigators with a powerful analytic set of images and data totaling ~60GB. This subset is designed to provide the most relevant lung cancer cases and matched controls at a more manageable size. The Case-Control Subset provides investigators with the x-ray for the year of diagnosis for all cancers diagnosed within one year of an x-ray image. The selection also includes 15 control images for every case image, including: a 5:1 ratio of screen images of abnormal/suspicious, abnormal/not suspicious, and negative screens, respectively. All control images are matched to cases based on study year and smoking status at the time of randomization. In detail, the Case-Control Subset includes:

• Study image from the year of diagnosis of a screen-detected cancer or interval cancer within one year of screening (n=374)
• Study image from five randomly selected screening years with an abnormal/suspicious screen result for every case (n=1870)
• Study image from five randomly selected screening years with an abnormal/not suspicious screen result for every case (n=1870)
• Study image from five randomly selected screening years with a negative screen result for every case (n=1870)
Note: For ~6% of study years with an image, participants have multiple chest x-ray images that will be included.

Sampling weights were calculated to show how the randomly selected individuals and images reflect on the greater PLCO screened population. The Standard 25k Selection includes person-based weights while each image in the Case-Control Subset is assigned a screen-based weight. For more information on the sampling weights, please see the user's guide.

Custom Selections

Investigators may opt to select their own custom populations, limited to at most 25,000 participants. However, custom selections are only available via hard drive and are not able to be downloaded from the CDAS website. The requestor must bear the expense of the hard drive and all related shipping expenses, including prepaid return shipping. Signature requirement and acknowledgement of physical possession are required in a timely manner. Contact Us if you have any questions.