This notebook provides a minimal working example of a Deep Learning model for the detection of intravenous contrast enhancement in head and neck or lung CT scans. The model was trained using 1979 contrast and non-contrast head and neck and chest CT images from six different datasets acquired at multiple sites. The dataset is composed of images with variable volume sizes, field of view, and slice thickness.
We test the model by implementing an end-to-end (cloud-based) pipeline on publicly available chest CT scans hosted on the Imaging Data Commons (IDC), starting from raw DICOM CT data and ending with a DICOM SEG object storing the segmentation masks generated by the AI pipeline. The testing dataset we use is external and independent from the data used in the development phase of the model (training and validation) and is composed of a wide variety of image types (from image acquisition settings, to the presence of the contrast agent, to the presence, location and size of a tumor mass).
The way all the operations are executed - from pulling data to data postprocessing and the standardisation of the results - have the goal of promoting transparency and reproducibility.
Run this model in Google CoLab.
Please cite the following article if you use this code or pre-trained models:
Ye, Z., Qian, J.M., Hosny, A., Zeleznik, R., Plana, D., Likitlersuang, J., Zhang, Z., Mak, R.H., Aerts, H.J. and Kann, B.H., 2022. Deep Learning–based Detection of Intravenous Contrast Enhancement on CT Scans. Radiology: Artificial Intelligence, 4(3), p.e210285, https://doi.org/10.1148/ryai.210285.
The original code is published on GitHub. The original code is published using the Apache-2.0 license.