Application of deep learning for efficacy of radiation treatment for lung cancer

Lung cancer is one of the most fatal cancers in the United States. The type of treatment heavily depends on the stage and histology of lung cancer. Often, we need to use radiotherapy in conjunction with chemotherapy and surgery to treat patients with lung cancer. The aim is to kill the cancer cells while minimizing radiation to healthy tissue and important organs in the vicinity of treatment area. Although modern radiation oncology has a proven and predicable role for lung cancer treatment with gains in quality, efficacy, toxicity, and outcomes, still we have local treatment failure. Also, the damage to the normal tissue around the tumor (fibrous) can lead to severe post treatment issues/toxicities for the patients. Therefore, the goal of this project is to quantify the effect of radiation in eliminating the tumor and predicting post-treatment toxicities/complications. For this purpose, we develop a convolutional neural network algorithm for tracking the efficacy of treatment. We will use pre- and post-treatment CT scan and PET images, with blood test results and radiation dose data of the patients. In such way we will predict outcomes of the treatment in terms of dose coverage to the tumor target and OARs per Quantitative Analysis of Normal Tissue Effects in the Clinic.