MyProject: Trainable CT Scan Scheduling for Optimised Radiation Dose

Wan Wang      Zhengxuan Cao      Hrishikesh Shetty

ww214@duke.edu     zc218@duke.edu     hns22@duke.edu

Paper PDF

Computed Tomography (CT) scans are crucial for disease diagnosis but increase the risk of radiation exposure due to high-intensity X-ray beams. A new machine learning model aims to optimize radiation dose and image quality in chest CT scans through a sophisticated multi-layer architecture. This includes a physical layer for data acquisition adjustments, a reconstruction network for image processing, and a validation network to ensure image quality. The model adjusts scanning parameters to reduce radiation while maintaining diagnostic effectiveness. Tests on a dataset processed through the model and standard methods showed comparable accuracy, demonstrating the model's success in safely reducing radiation doses.

Paper: Paper PDF

Code and Data: Code: Onedrive Datasets: Mask Training Dataset Covid Dataset