Melanie Wegener | Searcharxiv

arXiv2026

Tracking the Brownian motion of DNA-functionalized magnetic nanoparticles for conformation analysis beyond the optical resolution limit

Brownian motion provides access to hydrodynamic properties of nanoscale objects independent of their optical resolvability. Here, we present a diffusion-based approach to infer effective particle size distributions of DNA-functionalized magnetic nanoparticles (MNPs), consisting of a magnetic core and a polystyrene shell, in a regime where direct geometric sizing is limited by optical diffraction. Using multi-particle tracking microscopy, we analyze the Brownian dynamics of MNPs grafted with double-stranded DNA (dsDNA) of varying contour length under low-salt conditions. A physically motivated model is introduced that relates dsDNA contour length to an effective hydrodynamic diameter via an attenuated corona description. The measured diffusion coefficient distributions exhibit a systematic and monotonic dependence on dsDNA length in quantitative agreement with the model. While the tracked objects are predominantly dsDNA-mediated agglomerates rather than isolated nanoparticles, clustering does not obscure the length-dependent signal. Instead, the dsDNA corona determines the hydrodynamic scaling, whereas agglomeration mainly introduces an offset and distribution broadening. These results demonstrate that Brownian dynamics enables robust readout of biomolecular length scales even far below the optical resolution limit. The distribution-based approach is inherently tolerant to polydispersity and aggregation, making diffusion-based tracking a simple and promising strategy for future biotechnological and biomedical assays.

cond-mat.soft↗

arXiv2022

An efficient deep neural network to find small objects in large 3D images

3D imaging enables accurate diagnosis by providing spatial information about organ anatomy. However, using 3D images to train AI models is computationally challenging because they consist of 10x or 100x more pixels than their 2D counterparts. To be trained with high-resolution 3D images, convolutional neural networks resort to downsampling them or projecting them to 2D. We propose an effective alternative, a neural network that enables efficient classification of full-resolution 3D medical images. Compared to off-the-shelf convolutional neural networks, our network, 3D Globally-Aware Multiple Instance Classifier (3D-GMIC), uses 77.98%-90.05% less GPU memory and 91.23%-96.02% less computation. While it is trained only with image-level labels, without segmentation labels, it explains its predictions by providing pixel-level saliency maps. On a dataset collected at NYU Langone Health, including 85,526 patients with full-field 2D mammography (FFDM), synthetic 2D mammography, and 3D mammography, 3D-GMIC achieves an AUC of 0.831 (95% CI: 0.769-0.887) in classifying breasts with malignant findings using 3D mammography. This is comparable to the performance of GMIC on FFDM (0.816, 95% CI: 0.737-0.878) and synthetic 2D (0.826, 95% CI: 0.754-0.884), which demonstrates that 3D-GMIC successfully classified large 3D images despite focusing computation on a smaller percentage of its input compared to GMIC. Therefore, 3D-GMIC identifies and utilizes extremely small regions of interest from 3D images consisting of hundreds of millions of pixels, dramatically reducing associated computational challenges. 3D-GMIC generalizes well to BCS-DBT, an external dataset from Duke University Hospital, achieving an AUC of 0.848 (95% CI: 0.798-0.896).

cs.CV↗