Multi-Temporal Frames Projection for Dynamic Processes Fusion in Fluorescence Microscopy

TL;DR

This study tackles noise, temporal variability, and inconsistent visualization in fluorescence microscopy by fusing multi-temporal frames into a single high-quality 2D image using a modular preprocessing-plus-z-projection pipeline. The method emphasizes interpretability over opaque deep learning, enabling downstream segmentation, and is validated across 666 pipelines on 91 HL-1 Ca^{2+} signaling videos, achieving a 44.2% increase in detected cells compared with conventional frame stacking. Key contributions include a comprehensive, configurable framework combining CLAHE, Gamma Correction, and noise filtering with six projection strategies (MIP, AP, SP, PDP, STDP, QP), plus NR-IQA–driven guidance for image quality. The approach improves boundary delineation and annotation efficiency, with potential applicability to other multi-temporal imaging domains beyond fluorescence microscopy.

Abstract

Fluorescence microscopy is widely employed for the analysis of living biological samples; however, the utility of the resulting recordings is frequently constrained by noise, temporal variability, and inconsistent visualisation of signals that oscillate over time. We present a unique computational framework that integrates information from multiple time-resolved frames into a single high-quality image, while preserving the underlying biological content of the original video. We evaluate the proposed method through an extensive number of configurations (n = 111) and on a challenging dataset comprising dynamic, heterogeneous, and morphologically complex 2D monolayers of cardiac cells. Results show that our framework, which consists of a combination of explainable techniques from different computer vision application fields, is capable of generating composite images that preserve and enhance the quality and information of individual microscopy frames, yielding 44% average increase in cell count compared to previous methods. The proposed pipeline is applicable to other imaging domains that require the fusion of multi-temporal image stacks into high-quality 2D images, thereby facilitating annotation and downstream segmentation.

Figures (11)

• Figure 1: Systematic pipeline for multi-temporal frames fusion in (a) live-sample FM datasets via combined (b) image processing and (c) z-projection methods.
• Figure 2: Box plot of NR-IQR scores by projection method. Yellow-filled diamond-shaped dots within the boxes indicate mean values for each projection group.
• Figure 3: Qualitative analysis of NR-IQA metrics image quality evaluation performance on the HL-1 dataset. Individual rows show examples of images scoring a) the lowest, b) median and c) the highest PIQE (P), NIQE (N) and BRISQUE (B) scores across the full dataset, corresponding to a) high, b) medium and c) low scores assigned by each metric.
• Figure 5: Box-plot highlighting the 10 best preprocessing-projection algorithms combinations identified by PIQE, NIQE and BRISQUE scores. Data were normalised to allow a fair comparison between the three metrics. As before, yellow diamond-shaped dots within the boxes indicate mean values across the 77 images within the dataset, averaged across the specified preprocessing-projection method applied.
• Figure 6: Selected preprocessing and projection methods, reviewed with an expert experimentalist for cell annotation, including illustrative examples.