Individual lung cancer survival estimation based on radiomics analysis
When a patient dies in the hospital through a long treatment with painful side effects and very high financial cost, both the patients and their families suffer a lot during this period. We might ask: could it have been avoided, what if the clinicians acted differently? This is impossible to know because we cannot go back in time. Nonetheless, if we know the result in advance, we may act differently so that the decision based on estimated survival time can better improve the patients’ health care. The goal of this project is trying to leverage the recent deep learning techniques to facilitate the development in the lung cancer survival estimation. The hypothesis is that deep learning algorithm can train a model that can extract additional information beyond our experts and this information can serve for the survival estimation. The national lung cancer screening trial (NLST) dataset provides an unparalleled resource about patients with diagnosis images, clinic data, treatment, nodule findings, demographics, and other patient data. ‘These data can be used to generate a knowledgeable data mining system so that the sophisticated personalized diagnostic decision-making process is possible.
An individual survival estimation system is developed by using the NLST dataset.
Xiaoshui Huang, Xiaoshui.firstname.lastname@example.org
Tianqi Chen, email@example.com