The Effect of Noise on Hemorrhage Detection using CT Scan Imaging​
Joao Cardoso Jasmine King Pushti Desai Preethi Krishnan
joao.cardoso@duke.edu jazzie.king@duke.edu pushti.desai@duke.edu preethi.krishnan@duke.edu
| Paper PDF |
Solution Pipeline:
Physical Transformations:
|
CT images are affected by multiple physical like sensor noise, temperature (thermal/Johnson-Nyquist noise), electrical charges, background radiation, and photon counting. In this study, we simulate CT noise by applying Gaussian and Poisson noise to the sinogram of CT Scans and utilize the reconstructed images to detect Intracranial Hemorrhage (ICH) using the publicly available RSNA dataset. Our results show that a physics-informed VGG pipeline can perform effectively with simulated noise.
Paper:
|
Code and Data:
|
Link to the original challenge:
| |