Radiomics to 1. Identify Patients at Risk for Developing Pneumonitis, 2. Differentiate Immune Checkpoint Inhibitor-induced Pneumonitis From Other Lung Inflammation and 3. Distinguish Tumour Pseudo-progression From Real Tumour Growth, in Patients With Non-small Cell Lung Cancer Treated With Anti-PD1 or Anti-PD-L1
Radiomics to 1. Identify Patients at Risk for Developing Pneumonitis, 2. Differentiate Immune Checkpoint Inhibitor-induced Pneumonitis From Other Lung Inflammation and 3. Distinguish Tumour Pseudo-progression From Real Tumour Growth, in Patients With Non-small Cell Lung Cancer Treated With Anti-PD1 or Anti-PD-L1
The investigators will develop a radiomics signature for immune checkpoint-induced pneumonitis in 40 patients with a pulmonary event under anti-PD1 or anti-PD-L1 (cases) and 40 patients without a pulmonary event under anti-PD1 or anti-PD-L1 (controls).
On the basis of the case-control study of patients treated with anti-PD1 or anti-PD-L1, they will further optimise the model using reinforcement machine learning. The model will then be validated in 300 prospective patients.
Preliminary analyses on a dataset showed a clear distinction in radiomics features for patients with and without pneumonitis from anti-PD1 or anti-PD-L1. Prior experience of the investigators of training and validating radiomics signatures combined with their preliminary exploratory results presented here, will be used to develop a radiomics signature for immune checkpoint-induced pneumonitis in 40 patients with a pulmonary event under anti-PD1 or anti-PD-L1 (cases) and 40 patients without a pulmonary event under anti-PD1 or anti-PD-L1 (controls).
On the basis of the case-control study of patients treated with anti-PD1 or anti-PD-L1, the investigators will be able to further optimise the model using reinforcement machine learning. The model will then be validated in 300 prospective patients.
Inclusion Criteria:
Exclusion Criteria: