Funnel ReLU outperforms standard activations in medical image deblurring benchmark
A systematic benchmark of fifteen activation functions across CT, mammography and MRI deblurring finds that the spatially adaptive Funnel ReLU ...
A systematic benchmark of fifteen activation functions across CT, mammography and MRI deblurring finds that the spatially adaptive Funnel ReLU ...
A new MLP-based segmentation method combining mixer layers with channel attention achieved a Dice coefficient of 81.45 percent on the ...
A new deep learning network called MTF-Net combines multi-scale transformer attention with cross-branch feature fusion to segment brain tumors from ...
The CARD framework reached a Macro-AUROC of 75.9% in semi-supervised chest X-ray disease classification, rising to 77.0% when pre-diagnostic clinical ...
A narrative review finds that deep-learning synthetic CT can improve cone-beam CT images and dose calculations, but clinical validation remains ...
Diabetic foot ulcers are among the most feared complications of diabetes, and now a team of researchers in India has ...
A topology-aware deep learning model keeps coronary vessels connected in noisy X-ray angiograms, reaching 88.3 percent Dice on the DCA1 ...
A new dual-branch U-Net called PDSA-DBUNet combines dynamic sparse attention with a polynomial mixer to achieve record abdominal CT segmentation ...
A multicenter study found that a deep learning model analyzing full ultrasound videos significantly improved diagnostic accuracy for suspicious breast ...
A two-stage deep learning framework can automatically segment subpleural lung lesions on grayscale ultrasound and distinguish benign from malignant cases, ...
© 2026 Scienmag - Science Magazine
© 2026 Scienmag - Science Magazine