arXiv subjects

Osamah Sufyan

Publications and source records attributed to Osamah Sufyan.

3 recordsLinked to original sources

arXiv2026

PU-UNet: Stable Multiplicative Interactions for Medical Image Segmentation

Many dense prediction networks rely on additive feature transformations and model higher-order feature interactions only implicitly. Product units provide an explicit mechanism for multiplicative feature modeling, but their logarithmic--exponential formulation can cause numerical instability, which has limited their use in deep dense prediction networks. In this work, we propose Product-Unit U-Net (PU-UNet), a residual U-Net that integrates stable product-unit residual blocks into rich low-resolution stages for medical image segmentation. The proposed formulation combines smooth positivity mapping with log-domain clipping, enabling stable multiplicative feature learning with negligible computational overhead. On ISIC 2018, Kvasir-SEG, and BUSI, PU-UNet achieves Dice scores of 0.942, 0.959, and up to 0.925, respectively. Compared with a matched Residual U-Net baseline, PU-UNet consistently improves Dice and IoU while keeping parameters, FLOPs, and inference latency nearly unchanged, and reduces the image-level false-positive rate on normal BUSI cases from 0.077 to zero. Ablation studies suggest that the gains are associated with product-unit interactions, are strongest under low-resolution placement, and benefit from the proposed stabilization design. These results suggest that stable product-unit residual learning can be an effective way to enhance U-Net-style segmentation networks with explicit multiplicative interactions.

cs.CV↗

arXiv2026

Double circular dichroism high harmonic spectroscopy: An ultrafast probe for topological photocurrents

Understanding optical responses of topological matter is a central problem for enabling optoelectronic applications based on topological physics, which is of fundamental concern for photocurrents control and spectroscopy. Currently, schemes for sensing ultrafast photocurrents and separating their bulk/surface contributions are lacking. We introduce here double circular dichroism (DCD) harmonic spectroscopy as an all-optical probe of ultrafast dynamics in topological materials. In this scheme, pump and probe pulses are circular with helicities that are independently controlled, yielding the circular dichroism of the circular dichroism -- a time-resolved response evaluating how probe-induced dichroism depends on pump helicity. While DCD vanishes in symmetric systems, it survives in broken time-reversal symmetry materials including Chern insulators. We theoretically demonstrate this concept through simulations in a Haldane nanoflake, where a pump laser manipulates chiral current-carrying states, and intense probe pulses drive harmonic emission. We show that DCD originates from both bulk and edge-localized states, but these have opposite signs, similar magnitudes, and a different amplitude scaling. Hence, DCD could allow efficient separation of bulk/edge contributions to photocurrents. Variation of the electronic structure and laser parameters further reveals anomalies that might be useful for probing topological attributes of photocurrents in select harmonics. Overall, our work introduces DCD as a potentially powerful approach for disentangling bulk/boundary photo-responses in broken-symmetry quantum matter, and could also be implemented in other pump-probe spectroscopies based on photoelectrons and absorption, as well as other chiral systems.

cond-mat.other↗

arXiv2026

Anatomy-Informed Deep Learning for Abdominal Aortic Aneurysm Segmentation

In CT angiography, the accurate segmentation of abdominal aortic aneurysms (AAAs) is difficult due to large anatomical variability, low-contrast vessel boundaries, and the close proximity of organs whose intensities resemble vascular structures, often leading to false positives. To address these challenges, we propose an anatomy-aware segmentation framework that integrates organ exclusion masks derived from TotalSegmentator into the training process. These masks encode explicit anatomical priors by identifying non-vascular organsand penalizing aneurysm predictions within these regions, thereby guiding the U-Net to focus on the aorta and its pathological dilation while suppressing anatomically implausible predictions. Despite being trained on a relatively small dataset, the anatomy-aware model achieves high accuracy, substantially reduces false positives, and improves boundary consistency compared to a standard U-Net baseline. The results demonstrate that incorporating anatomical knowledge through exclusion masks provides an efficient mechanism to enhance robustness and generalization, enabling reliable AAA segmentation even with limited training data.

cs.CV↗