“Application of Clinical and Radiomic Features for the Differential Diagnosis of Pneumonitis in Stage IV Non-Small-Cell Lung Cancer (NSCLC) Patients Treated with Immune Checkpoint Inhibitors”
Immunotherapy-Induced Pneumonitis (IIP) is a sporadic and unpredictable lethal side-effect which requires timely diagnosis for management with steroids. Non-small-cell lung cancer (NSCLC) patients treated with immunotherapy can also manifest with pneumonitis induced by causes other than immunotherapy, such as bacterial and fungal infections, sarcoid-like pulmonary reactions, chronic obstructive pulmonary disease exacerbations, and radiation-induced pneumonitis. The overlapping clinical manifestations of IIP and Other types of Pneumonitis (OP) has created a diagnostic challenge. Making a differential diagnosis in a single clinical center is error-prone due to indistinct radiological patterns that can be appreciated by visual inspection and also lack of large scale studies because of low incidence rates. In this talk, I will present the results of our study that we sought to devise clinical, radiomic, and combined models for estimating the risk of IIP in stage IV NSCLC patients who are receiving immunotherapy.
“Adaptive radiotherapy in the era of learning algorithms and big data”
The past decade has shown tremendous progress in dozens of perception and modelling tasks with impressive applications, from autonomous vehicles to playing a game of Go, or modelling the folding of proteins. It is natural, therefore, to ask whether neural networks could have similar success in domains with data that are less in quantity, stronger in structure due to the underlying physics of acquisition, and possibly dynamic over time by the nature of the task. A prime example of such a domain is adaptive radiotherapy, where the mechanics of the CT acquisition are well known, and used to reconstruct the anatomy of the patient and the tumor prior to radiation. In this talk, I will present the vision of POP-AART for online adaptive radiotherapy guided by neural network learning algorithms.
