Dataset¶

The training dataset (n=150 cases) is now available on Zenodo! Please follow this link to download the data: https://zenodo.org/records/11199559. 

Training Dataset Folder/File Structure¶

The training dataset is uploaded as a ZIP archive to Zenodo. The data is provided with the following folder hierarchy:

• Top-level folder (named "HNTSMRG24_train")
  • Patient-level folder (anonymized patient ID, example: "2")
    • Pre-radiotherapy data folder ("preRT")
      • Original pre-RT T2w MRI volume (example: "2_preRT_T2.nii.gz")
      • Original pre-RT tumor segmentation mask (example: "2_preRT_mask.nii.gz")
    • Mid-radiotherapy data folder ("midRT")
      • Original mid-RT T2w MRI volume (example: "2_midRT_T2.nii.gz")
      • Original mid-RT tumor segmentation mask (example: "2_midRT_mask.nii.gz")
      • Registered pre-RT T2w MRI volume (example: "2_preRT_T2_registered.nii.gz")
      • Registered pre-RT tumor segmentation mask (example: "2_preRT_mask_registered.nii.gz")

For more details on the dataset download please visit the Zenodo repository. 

Patient Cohorts¶

Patients with histologically proven HNC who underwent radiotherapy (RT) at The University of Texas MD Anderson Cancer Center (MDACC).  Predominantly oropharyngeal cancer (OPC) or cancer of unknown primary. 

Imaging Data¶

T2-weighted (T2w) anatomical sequences of the head and neck region taken at MDACC. Data will be a mix of fat-suppressed and non-fat-suppressed images. All patients are immobilized using a thermoplastic mask. Raw images were automatically extracted from a centralized institutional imaging repository (Evercore). Images include pre-RT (1-3 weeks before start of RT) and mid-RT (2-4 weeks intra-RT) scans; example shown below. Pre-RT and mid-RT image pairs for a given patient will be consistently either fat-suppressed or non-fat-suppressed.

Figure 1. Example of pre-RT and mid-RT T2w (non-fat-suppressed) scans for a patient. 

Segmentation Information¶

Primary gross tumor volumes (abbreviated GTVp) - at most 1 per patient (can be 0), and metastatic lymph nodes (abbreviated GTVn) - variable number per patient (can be 0). 

Multiple physician expert observers (n = 3 to 4) have independently segmented GTVp and GTVn structures for all cases (pre-RT and mid-RT) based on MRI images provided. Based on recent literature from our group (PMID: 36761036), a minimum of 3 annotators is suggested to yield acceptable segmentations when combined via the simultaneous truth and performance level estimation algorithm (STAPLE) in these structures. Therefore, we have collected independent segmentations from >=3 annotators for each structure. All annotators were medical doctors with at least 2 years of experience in head and neck cancer segmentation. All annotators had access to patient medical histories and any previous relevant imaging (e.g., PET/CT) via the patients chart. Final verification of segmentation quality was performed by experienced radiation oncology faculty members with greater than 10 years of experience.  Segmentations were combined via the STAPLE algorithm to yield the final ground truth segmentation for each case; an illustrative example is shown below. For a small subset of cases where there was extreme disagreement between observers, only a single contour from an experienced radiation oncology faculty member was used. Note: For this challenge, we will only provide consensus segmentations to participants. Individual observer segmentations will not be provided but will be made publicly available upon the challenge's completion (i.e., via TCIA).

Figure 2. Example of the STAPLE consensus process combining multiple segmentations into a single final consensus segmentation. 

The final label mask has one of three possible values: background = 0, GTVp = 1, GTVn = 2 (in the case of multiple lymph nodes they are concatenated into one single label). A visual example is shown below. 

Figure 3. A visual example of the mask labeling scheme for this challenge. Background = 0, primary gross tumor volume (GTVp) = 1 , metastatic node gross tumor volume (GTVn) = 2. Visualization performed in 3D Slicer. 

Data Pre-Processing¶

Anonymized DICOM files (images and structure files) are converted to NIfTI format (.nii.gz) for ease of use by participants. All images are cropped from the top of the clavicles to the bottom of the nasal septum (oropharynx region to shoulders), allowing for more consistent image field of views and removal of identifiable facial structures. 

Training and Test Data¶

The same patient cases will be used for the training and test sets of both tasks of this challenge. Therefore, we plan to release a single training dataset that can be used to construct solutions for either segmentation task. As of June 14th, we have released the training dataset on Zenodo here: https://zenodo.org/records/11199559. 

For a given patient case, the following training data will be provided in .nii.gz format:

• Original pre-RT T2w MRI volume with original pre-RT segmentation mask.
• Original mid-RT T2w MRI volume with original mid-RT segmentation. 
• Registered pre-RT T2w MRI volume with registered pre-RT segmentation mask - More details on why these files are provided are mentioned on the Tasks and Evaluation page. We also provide a example of how registrations were performed on our GitHub.

The test data provided (via Docker containers), however, will be different for the two tasks. Participants must be cognizant that only certain specific files will be provided for their Docker containers depending on which task they are submitting for. More details on what is expected of participants algorithms and what will be provided during testing is discussed in the Tasks and Evaluation page and Submission Instructions page.

Misc.¶

• Data from the training and test sets are representative of real-world cases from a large cancer institute treating HNC. Training and test sets will be partitioned such as to contain similar distributions based on dataset characteristics such as image fat-suppression status, tumor response, TNM staging,  etc. 
• Only the challenge organizers (i.e., MDA Fuller Lab) will have access to the ground-truth segmentations (labels) for the test cases until final publication of data. 
• Ethics approval was obtained from the University of Texas MD Anderson Cancer Center Institutional Review Board with protocol number RCR03-0800. This is a retrospective data collection protocol with a waiver of informed consent.