Loc, I. | Search bioRxiv

bioRxiv2025

3D-Printed Magnetic Levitation Device with Deep Learning-Assisted Particle Tracking and Analysis for High-Throughput Sorting

Intrinsic density-based particle separation is fundamental to biomedical research and materials science. Magnetic levitation offers an accessible and label-free approach; however, current platforms are limited by throughput, complex fabrication requirements, and manual analysis methods. Here, we demonstrate a high-throughput magnetic levitation-based microfluidic device fabricated using commercial 3D printing, integrated with dual automated analysis systems. The device features optimized magnet configurations and wide channel design (1 mm x 1 mm) that enables gentle separation (<1 PSI) at throughputs of 66 mL/hr-- a ten-fold improvement over existing levitation platforms. We developed two complementary analysis tools: "Phase" for static levitation height measurements, and a deep learning pipeline combining CNN-based particle classification (>95% accuracy) with SORT (Simple Online and Realtime Tracking) algorithm for dynamic analysis. The automated system showed excellent correlation with manual counting (Pearson coefficients: 0.91-0.99, p<0.001). Through systematic optimization of magnet spacing and paramagnetic medium concentration (150 mM Gd), the platform achieved robust continuous-flow sorting while maintaining exceptional purity (>90%) and resolving density differences as small as 0.03 g/mL. This work establishes a versatile platform for particle sorting, enabling sophisticated analysis without specialized facilities or extensive operator training, with broad applications in biomedical research and diagnostics.

bioengineering↗

bioRxiv2022

Raman Signal Denoising Using Fully Convolutional Encoder Decoder Network

Raman spectroscopy is a vibrational method that gives molecular information rapidly and non-invasively. Despite its advantages, the weak intensity of Raman spectroscopy leads to low-quality signals, particularly with tissue samples. The requirement of high exposure times makes Raman a time-consuming process and diminishes its non-invasive property while studying living tissues. Novel denoising techniques using convolutional neural networks (CNN) have achieved remarkable results in image processing. Here, we propose a similar approach for noise reduction for the Raman spectra acquired with 10x lower exposure times. In this work, we developed fully convolutional encoder-decoder architecture (FCED) and trained them with noisy Raman signals. The results demonstrate that our model is superior (p-value < 0.0001) to the conventional denoising techniques such as the Savitzky-Golay filter and wavelet denoising. Improvement in the signal-to-noise ratio values ranges from 20% to 80%, depending on the initial signal-to-noise ratio. Thus, we proved that tissue analysis could be done in a shorter time without any need for instrumental enhancement.

biochemistry↗