Amir Hossein Saleknia

State-of-the-art 3D pancreatic tumor segmentation no longer demands massive architectures or pre-training: a tailored 5.86M-parameter model boosts lesion sensitivity by 6.6 points simply by replacing standard pooling with lossless downsampling and linear-deformable attention.

Supervised classification wildly overestimates dataset bias in web-scale image collections, mistaking resolution artifacts for meaningful semantic differences.