Image-Domain Poisson-Perturbation Robustness of NCCT Slice Classification
The study tested how classifying NCCT slices changes when image quality is intentionally degraded by Poisson noise. They compared direct classification versus using a pre-denoiser before classification, finding no statistical benefit for joint denoising methods. While some fidelity improved, the results show that task-aware training does not statistically outperform simple direct classification for slice ranking.