BioLite U-Net: Edge-Deployable Semantic Segmentation for In Situ Bioprinting Monitoring

TL;DR

BioLite U-Net targets real-time, edge-capable semantic segmentation of extrusion in bioprinting to enable closed-loop control under hardware constraints. It introduces a lightweight encoder–decoder network using depthwise separable convolutions, trained on a novel 787-image 3-class dataset and deployed on a Raspberry Pi 4B, achieving high segmentation accuracy with a tiny parameter count. The study benchmarks BioLite against MobileNetV2- and MobileNetV3-based baselines, showing a strong accuracy-efficiency tradeoff and real-time on-device inference, thus enabling smart, autonomous bioprinters. The work advances practical vision-based monitoring for bioprinting and establishes a public dataset and design principles for edge-focused segmentation in constrained environments.

Abstract

Bioprinting is a rapidly advancing field that offers a transformative approach to fabricating tissue and organ models through the precise deposition of cell-laden bioinks. Ensuring the fidelity and consistency of printed structures in real-time remains a core challenge, particularly under constraints imposed by limited imaging data and resource-constrained embedded hardware. Semantic segmentation of the extrusion process, differentiating between nozzle, extruded bioink, and surrounding background, enables in situ monitoring critical to maintaining print quality and biological viability. In this work, we introduce a lightweight semantic segmentation framework tailored for real-time bioprinting applications. We present a novel, manually annotated dataset comprising 787 RGB images captured during the bioprinting process, labeled across three classes: nozzle, bioink, and background. To achieve fast and efficient inference suitable for integration with bioprinting systems, we propose a BioLite U-Net architecture that leverages depthwise separable convolutions to drastically reduce computational load without compromising accuracy. Our model is benchmarked against MobileNetV2 and MobileNetV3-based segmentation baselines using mean Intersection over Union (mIoU), Dice score, and pixel accuracy. All models were evaluated on a Raspberry Pi 4B to assess real-world feasibility. The proposed BioLite U-Net achieves an mIoU of 92.85% and a Dice score of 96.17%, while being over 1300x smaller than MobileNetV2-DeepLabV3+. On-device inference takes 335 ms per frame, demonstrating near real-time capability. Compared to MobileNet baselines, BioLite U-Net offers a superior tradeoff between segmentation accuracy, efficiency, and deployability, making it highly suitable for intelligent, closed-loop bioprinting systems.

Figures (5)

• Figure 1: Overview of the proposed segmentation framework for real-time bioprinting monitoring.
• Figure 2: Proposed BioLite U-Net architecture with depthwise separable convolutions.
• Figure 3: Example images from the bioprinting dataset.
• Figure 4: Training curves of BioLite U-Net: loss, Dice, and IoU scores over 200 epochs.
• Figure 5: Qualitative segmentation results of BioLite U-Net on test samples.