Complete the Google Form to access the Github where you can download the hidden Rad dataset. (Google Form)
Then we will send the access right.
For the sample study, the first release data sample is open in the public github directory:
for task 1, go to https://github.com/hidden-rad/Task1. (MIMIC license is required.)
for task 2, go to https://github.com/hidden-rad/Task2.
The data for Task 2 is provided in CSV format, containing relevant information for each case in a structured tabular form. Each row in the CSV file represents a single case, with the following columns:
A1 (First Impression):
Initial observations by radiologists. This data serves as the starting point for diagnostic reasoning and may include preliminary hypotheses based on observed findings.
Example: 'emphysema', 'pleural effusion', 'pneumothorax'
A2 (Anatomical Location):
Specific anatomical locations relevant to the case, helping narrow down the area of focus and refining the diagnostic flow.
Example: 'Subcutaneous tissue', 'Parenchyme', 'RUL(Right Upper Lobe)', 'Parenchyme', 'RML(Right Middle Lobe)', 'Parenchyme', 'RLL(Right Lower Lobe)', 'Parenchyme', 'LUL(Left Upper Lobe)', 'Parenchyme', 'LLL(Left Lower Lobe)', 'Pleural', 'Right Pleural'
A3 (Thoracic Spine Levels):
Information on thoracic spine levels involved in the case, providing additional context for spine-related findings.
Example: {'begin': 1, 'end': 12}
A4 (Final Impression):
The concluding impressions by radiologists after evaluating all available information. This represents the final diagnostic insight derived from the data.
Example: 'emphysema', 'pleural effusion', 'pneumothorax'
Causal section:
A unique identifier for each case, used to locate the associated ground-truth causality report for validation.
Example: 083663b6-04df-418e-a959-3d0a331d3d81
Each of these components contributes to the overall diagnostic flow, replicating a radiologist’s structured thought process.