Di Gong | Searcharxiv

arXiv2025

New Constraints on Axion Mediated Dipole-Dipole Interactions

The search for axions sits at the intersection of solving critical problems in fundamental physics, including the strong CP problem in QCD, uncovering the nature of dark matter, and understanding the origin of the universe's matter-antimatter asymmetry. The measurement of axion-mediated spin-dependent interactions offers a powerful approach for axion detection. However, it has long been restricted to regions outside the 'axion window' due to a significant trade-off: the need to effectively suppress the magnetic leakage from highly polarized spin sources while simultaneously detecting sub-femtotesla level exotic physics signals at sub-decimeter-scale distances. In this work, we report new experimental results on axion-mediated exotic spin-spin interactions using an iron-shielded SmCo$_5$ spin source in combination with a specially designed self-compensation comagnetometer. Employing a composite shielding structure, we achieved a suppression of the magnetic field by up to $10^{11}$. This enabled us to establish new constraints on the coupling between electrons and neutrons, surpassing previous experimental limits by more than 10000 times within the axion window. Furthermore, we also set strongest constraints on the coupling between electrons and protons. The proposed method holds substantial potential not only for advancing the search for new physics beyond the Standard Model but also for enabling transformative applications in biological and chemical research.

hep-ph↗

arXiv2020

Learn to Segment Retinal Lesions and Beyond

Towards automated retinal screening, this paper makes an endeavor to simultaneously achieve pixel-level retinal lesion segmentation and image-level disease classification. Such a multi-task approach is crucial for accurate and clinically interpretable disease diagnosis. Prior art is insufficient due to three challenges, i.e., lesions lacking objective boundaries, clinical importance of lesions irrelevant to their size, and the lack of one-to-one correspondence between lesion and disease classes. This paper attacks the three challenges in the context of diabetic retinopathy (DR) grading. We propose Lesion-Net, a new variant of fully convolutional networks, with its expansive path re-designed to tackle the first challenge. A dual Dice loss that leverages both semantic segmentation and image classification losses is introduced to resolve the second challenge. Lastly, we build a multi-task network that employs Lesion-Net as a side-attention branch for both DR grading and result interpretation. A set of 12K fundus images is manually segmented by 45 ophthalmologists for 8 DR-related lesions, resulting in 290K manual segments in total. Extensive experiments on this large-scale dataset show that our proposed approach surpasses the prior art for multiple tasks including lesion segmentation, lesion classification and DR grading

cs.CV↗