Real-time tumor tracking for proton therapy using deep learning and X-Ray fluoroscopy

Details

Supervisors

Macq, Benoît ; Loyen, Estelle

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en informatique, à finalité spécialisée

Abstract

Cancer is a leading cause of death worldwide, accounting for nearly 10 million deaths in 2020, or nearly one in six deaths. Among those, lung cancer is the biggest killer, causing more than 1.8 million deaths in the same year. Radiotherapy is one of the most common treatment methods for tumors, it uses ionizing radiation to destroy cancerous cells. Radiotherapy can take a multitude of forms, the most frequent one being external radiotherapy using photons as the radiation delivering mechanism. In recent years, proton therapy, another form of external radiotherapy has been steadily gaining traction. It uses protons as the radiation delivery medium, and as a result, can be more precise in its treatment thanks to a particularity of its radiation dose deposition curve named the Bragg peak. Because of this peak, the radiation can be deposited locally at the position of the tumor. However, for tumors located on moving organs such as the lungs, the position of the tumor can shift during the treatment, reducing the treatment's accuracy and requiring large security margins to cover the range of movement. In this work, we develop and analyze the performance of a real-time tumor segmentation model based on deep learning. This new approach consists of detecting the exact position of the tumor with a convolutional neural network to adjust the target of the treatment based on X-ray fluoroscopic imaging. If this approach proves successful, it would remove the need for large security margins and lead to a more accurate treatment with less compromise.

Citations

• APA

Chatzopoulos, E. (2022). Real-time tumor tracking for proton therapy using deep learning and X-Ray fluoroscopy. https://dial-mem.test.bib.ucl.ac.be/handle/2078.2/27629