A small correction to a recently published study on the intricate dynamics of neuronal filopodia has drawn attention to a deceptively important detail in the way three-dimensional microscopy data are described. The correction, published in BMC Bioinformatics on 25 August 2026, amends the caption of Figure 2 in the article “Semi-automatic geometrical reconstruction and analysis of filopodia dynamics in 4D two photon microscopy images.” The revised caption clarifies that the axial resolution of the two-photon microscopy images was 0.1 × 0.1 × 0.5 micrometers, rather than the previously stated value. Although the change involves a single number, it directly affects how readers interpret the shape, scale, and measurable motion of the tiny cellular protrusions shown in the study.
Filopodia are slender, dynamic extensions that project from the surfaces of many cells, including neurons. In developing nerve cells, they can act as exploratory structures, probing the surrounding environment and helping axons establish or modify connections. Their behavior is highly dynamic: a filopodium may extend, retract, bend, branch, or disappear over short periods. Understanding these changes requires more than a static image. Researchers need to follow the structures through space and time, reconstructing their geometry in three dimensions while also tracking how that geometry evolves. This is the central challenge addressed by the original study, which presents a semi-automatic approach for reconstructing and analyzing filopodia dynamics in four-dimensional two-photon microscopy datasets.
The phrase “four-dimensional” refers to three spatial dimensions plus time. In the study, two-photon microscopy was used to image axon terminals, the specialized ends of neuronal projections where communication with other cells can occur. The technique produces stacks of optical sections at successive time points, creating a volumetric movie of living or preserved biological structures. Instead of analyzing only the brightness of individual pixels, the researchers sought to recover the underlying geometrical organization of filopodia. This can involve identifying the central axis of each protrusion, estimating its length and orientation, measuring curvature, and determining how those properties change as the axon terminal remodels. Such measurements can transform a visually complex image sequence into quantitative data suitable for biological comparison.
The corrected resolution value is essential because microscopy images do not have equal precision in every direction. In the revised Figure 2 caption, the lateral dimensions are given as 0.1 × 0.1 micrometers, while the axial dimension along the z direction is 0.5 micrometers. This means that the microscope can distinguish considerably finer detail across the image plane than along the optical axis. In practical terms, a feature may appear sharply defined in the x-y plane but become elongated, blurred, or less precisely localized in z. This anisotropy is common in three-dimensional optical microscopy and must be incorporated into image processing and geometric interpretation. Treating the voxels as isotropic when they are not could distort lengths, angles, volumes, and the apparent trajectories of filopodia.
A voxel is the three-dimensional equivalent of a pixel: a small volume element assigned an intensity value. In this dataset, each voxel represents a region measuring 0.1 micrometers by 0.1 micrometers by 0.5 micrometers. Because the z dimension is five times larger than either lateral dimension, a visually spherical object can appear stretched along the axial direction, while a thin filopodium may occupy only a small number of z slices. The corrected caption therefore provides readers with the physical scale needed to judge the examples accurately. It also helps researchers reproduce the analysis, calibrate software, and compare the reported results with data acquired using other microscopes or imaging settings.
The original research was developed by scientists affiliated with the Zuse Institute Berlin, Freie Universität Berlin, and the Stowers Institute for Medical Research. The author team included Blaž Brence, Josephine Brummer, Vincent J. Dercksen, Mehmet Neset Özel, Abhishek Kulkarni, Neele Wolterhoff, Steffen Prohaska, Peter Robin Hiesinger, and Daniel Baum. Brence and Brummer contributed equally to the work. Their collaboration brought together expertise in computational geometry, image analysis, microscopy, and neurobiology. That combination is increasingly important as modern microscopes generate datasets too large and complex to analyze reliably through manual inspection alone.
Semi-automatic analysis occupies a useful middle ground between entirely manual tracing and fully automated segmentation. Manual reconstruction can be accurate for a small number of structures, but it is slow and subject to differences between observers. Fully automated methods, meanwhile, may struggle when filopodia overlap, fade in intensity, merge with the parent axon terminal, or change shape between frames. A semi-automatic workflow can allow algorithms to propose candidate structures or trajectories while giving researchers the ability to correct ambiguous decisions. This approach preserves biological oversight while reducing repetitive work. In a time-resolved volume, even modest improvements in efficiency can make it possible to examine many more terminals and filopodia than would be feasible by hand.
The technical problem is particularly demanding because filopodia are close to the limits of optical resolution and are embedded in irregular cellular surfaces. Their apparent brightness can vary because of labeling, illumination, local background, and movement through the imaging volume. A protrusion may also be oriented obliquely relative to the imaging plane, making its true length difficult to estimate from a two-dimensional projection. Geometrical reconstruction attempts to solve this by representing the structure in three dimensions, correcting for the physical spacing of voxels, and extracting measurements from the reconstructed object rather than from a flattened image. Accurate voxel calibration is therefore not a cosmetic detail: it is part of the mathematical foundation of the analysis.
The correction does not change the study’s central methods, conclusions, authorship, or research question. It corrects the caption accompanying Figure 2, which displays volume-rendered examples of axon terminals with filopodia. Panel a shows two axon terminals in an x-y orientation, while panel b illustrates the poorer axial resolution in the z direction. The corrected caption identifies the resolution as 0.1 × 0.1 × 0.5 micrometers cubed, with the amended value highlighted in the correction notice. The original article has been updated accordingly. By formally documenting the change, the journal ensures that readers who cite, reproduce, or extend the work have access to the correct imaging specifications.
The episode also highlights a broader issue in the era of data-intensive biology: a single parameter can influence an entire chain of scientific interpretation. Resolution affects segmentation, segmentation affects reconstruction, reconstruction affects geometry, and geometry affects biological conclusions about growth, retraction, orientation, and structural stability. Transparent corrections allow that chain to remain traceable. As researchers increasingly use machine learning, computational modeling, and automated image analysis to study living cells, precise descriptions of acquisition settings will become even more important. The corrected filopodia study offers not only a computational framework for examining neuronal structure in four dimensions, but also a reminder that trustworthy science depends on accurately reporting the physical limits of the images on which every measurement rests.
Subject of Research: Semi-automatic 4D geometrical reconstruction and analysis of filopodia dynamics in neuronal axon terminals using two-photon microscopy.
Article Title: Correction: Semi-automatic geometrical reconstruction and analysis of filopodia dynamics in 4D two photon microscopy images
Article References: Brence, B., Brummer, J., Dercksen, V. J. et al. “Correction: Semi-automatic geometrical reconstruction and analysis of filopodia dynamics in 4D two photon microscopy images.” BMC Bioinformatics 27, 197 (2026). The correction refers to: Brence et al. “Semi-automatic geometrical reconstruction and analysis of filopodia dynamics in 4D two photon microscopy images.” BMC Bioinformatics 27, 48 (2026).
Image Credits: AI Generated
DOI: https://doi.org/10.1186/s12859-026-06557-2
Keywords: Filopodia, axon terminals, neuronal development, two-photon microscopy, 4D imaging, three-dimensional reconstruction, image analysis, voxel resolution, computational geometry, biological image processing
