Understanding complex biological systems requires mapping them in three dimensions at high resolution. Whether tracing neural connections across a whole brain or analyzing cellular architecture within an organ, researchers face a fundamental trade-off between spatial resolution and volumetric throughput. Acquiring nanometer- or micrometer-scale data across millimeter- to centimeter-sized tissue volumes traditionally required months or years of painstaking manual labor and slow microscope operations.
To overcome this bottleneck, a new paradigm of high-throughput volumetric imaging has emerged. By automating physical tissue handling, eliminating mechanical settling delays in microscopes, and leveraging non-destructive penetrating radiation, these techniques compress data acquisition timelines from months to hours.
Eliminating Mechanical Latency in Scanning Systems
Traditional high-resolution imaging relies on a "stop-and-start" approach: a mechanical stage moves a specimen to a specific coordinate, waits for vibrations to settle, captures an image tile, and moves to the next position. When imaging a large volume requiring millions of tiles, these mechanical settling times dominate the acquisition timeline.
The first major insight of high-throughput imaging is to replace slow mechanical movements with inertialess electronic or optical scanning. For instance, in electron microscopy, Zheng et al. developed beam deflection Transmission Electron Microscopy (bdTEM). Instead of moving the physical stage for every single image tile, their system uses electromagnetic deflectors to rapidly steer the electron beam across a $3 \times 3$ grid of tiles for each individual stage movement. This hybrid approach achieves a net imaging rate of 0.3 gigapixels per second, making millimeter-scale reconstructions feasible in weeks rather than months.
In optical microscopy, a parallel strategy involves continuous, synchronized scanning. Rather than stopping the stage at all, the Volumetric Imaging with Synchronized on-the-fly-scan and Readout (VISOR) technique developed by Wu et al. moves the specimen continuously while synchronizing the camera's frame readout with the stage's velocity. This completely eliminates mechanical overhead, allowing an entire mouse brain to be mapped at synaptic resolution in under 1.5 hours. A similar synchronization principle is utilized by Chen et al. in Weakly supervised learning analysis of Aβ plaque distribution in the whole rat brain to rapidly generate the terabyte-scale datasets required for whole-organ disease profiling.
Automating and Parallelizing Physical Sectioning
Because light and electrons cannot deeply penetrate large, opaque tissues, specimens must often be physically cut into ultra-thin sections to achieve high-resolution imaging. Historically, handling these delicate slices was a manual, error-prone process. High-throughput pipelines solve this by automating section collection and parallelizing the imaging workload.
One robust approach is tape-based automated sectioning. As reviewed by Peddie et al. in Volume electron microscopy, volume electron microscopy (vEM) frequently employs automated tape-collecting microtomes to catch ribbons of ultra-thin tissue slices directly onto a continuous conveyor belt. This concept is translated to optical imaging by Larsen et al. with AutoCUTS-LM, which utilizes a transparent plastic tape system to collect sections from archived human brain tissue for rapid 3D digital reconstruction.
When scaling up to petascale volumes, such as the human cerebral cortex fragment mapped by Shapson-Coe et al. in A connectomic study of a petascale fragment of human cerebral cortex, automated ultra-thin sectioning is paired with multibeam scanning electron microscopy (mSEM). By shooting multiple simultaneous electron beams at the tape-mounted sections, data acquisition accelerates by orders of magnitude.
To bypass the throughput limit of a single microscope, researchers can also parallelize the tissue itself. Takemura et al. demonstrated this in A Connectome of the Male Drosophila Ventral Nerve Cord by partitioning large biological specimens into independent macroscopic slabs using a heated diamond knife. These slabs are then distributed across multiple electron microscopes to be imaged simultaneously, drastically compressing the total project timeline.
In the optical domain, automated sectioning is frequently integrated directly into the imaging loop to create serial-sectioning tomographies:
- Min et al. developed Serial optical coherence microscopy for label-free volumetric histopathology, combining automated mechanical sectioning with deep-tissue optical imaging to reconstruct whole organs in under an hour without chemical labels.
- Jiang et al. designed an "image-then-collect" platform in A platform for efficient identification of molecular phenotypes of brain-wide neural circuits that pairs structured illumination microscopy with a robotic water-flow slice collection system, digitizing a whole mouse brain in 10 hours while preserving the physical slices for downstream staining.
- Gong et al. developed the Brain-wide Positioning System (BPS), which integrates wide-field structured illumination with real-time tissue staining to map an entire brain at single-cell resolution in just three days.
Section-Free Volumetric Penetration
The most direct way to eliminate sectioning bottlenecks is to avoid cutting the tissue altogether. For small, transparent, or live organisms, advanced light-sheet microscopy provides a solution. Lemon et al. demonstrated this in Whole-central nervous system functional imaging in larval Drosophila, using a high-speed light-sheet architecture to capture real-time neural activity across an entire central nervous system simultaneously.
For larger, opaque specimens where light scatters severely, high-energy synchrotron X-ray imaging offers a non-destructive alternative. As explained by Westneat et al. in Advances in Biological Structure, Function, and Physiology Using Synchrotron X-Ray Imaging, synchrotron-based phase-contrast techniques exploit variations in X-ray phase shifts rather than simple absorption. This allows for micrometer-scale resolution of internal soft-tissue structures within intact millimeter- to centimeter-sized organisms.
To map large neuroanatomical structures efficiently, Dyer et al. utilized propagation-based phase-contrast microtomography (Quantifying Mesoscale Neuroanatomy Using X-Ray Microtomography). This approach leverages high-energy synchrotron beams to rapidly capture mesoscale tissue volumes without physical sectioning, drastically reducing the time required to map structural organization compared to electron microscopy. Similarly, Katzke et al. scaled this methodology to high-throughput phenomics in High-throughput phenomics of global ant biodiversity, using the high flux density of synchrotron X-rays to rapidly generate 3D models of intact biological specimens without staining or physical alteration.
Finally, mathematical acceleration can also optimize volumetric acquisition. In magnetic resonance imaging, Lesch et al. demonstrated in Ultrafast 3D Bloch–Siegert B‐mapping using variational modeling that tailored variational regularization strategies can exploit the distinct spatial characteristics of morphology to achieve up to 100x acceleration factors, enabling rapid 3D mapping within a single breath-hold.
Key Foundational Work
Whole-central nervous system functional imaging in larval Drosophila (2015) — contributed 1 concept elements to this direction.
Advances in Biological Structure, Function, and Physiology Using Synchrotron X-Ray Imaging (2008) — contributed 1 concept elements to this direction.