High-throughput serial two-photon tomography with BakingTray and BrainSaw

To understand the brain, scientists must see it in three dimensions. The brain is not a flat map but a dense, tangled forest of connections stretching through a volume of tissue. For decades, researchers have struggled to capture this full volume without destroying the very relationships they seek to study. Traditional methods often involve slicing the brain into thin sheets, mounting them on glass slides, and trying to reassemble the puzzle later. This process is fraught with problems: the delicate slices can tear, warp, or get lost, breaking the continuous story of how one part of the brain connects to another. While other techniques exist to make tissue transparent and image it whole, they often require complex chemical treatments that take weeks and demand specialized equipment. There is a need for a method that is fast, reliable, and capable of handling many samples at once, allowing scientists to map the brain's architecture with high precision while keeping the tissue intact during the imaging process.

A team of researchers has developed a solution that combines a powerful imaging technique with a new, open-source software platform to make this possible. They call their system a "BrainSaw," a name that reflects its function: it slices and images the brain simultaneously. The core technology relies on a two-photon microscope, which uses a focused beam of light to see deep inside tissue, paired with a vibratome, a device that cuts thin slices of tissue. Instead of removing the slices, the system images the fresh, exposed surface of the block after every cut. As the microscope captures the image of the surface, the machine slices off that layer, revealing a new surface underneath, and the process repeats thousands of times. This creates a seamless, three-dimensional movie of the entire organ. The innovation in this work is not just the hardware, but the software that controls it, called BakingTray. This software acts as the conductor, telling the microscope when to scan, when to cut, and how to stitch the millions of images together into a single, usable 3D dataset.

The researchers built twelve of these systems around the world, starting in 2016, and have put them to work in a busy facility at the Sainsbury Wellcome Centre in London. Over the years, these machines have imaged more than 7,000 samples, contributing to dozens of scientific papers. The system is designed to be robust enough to run unattended for hours, often overnight, allowing researchers to process multiple brains in a single session. The software is flexible enough to handle different types of microscopes and hardware setups, meaning labs can build their own systems or convert existing microscopes into these high-throughput imagers. The team demonstrated that the system is reliable, with the vast majority of imaging sessions completing without needing a restart. When interruptions do occur, they are usually due to simple user errors or predictable technical hiccups, and the system is designed to resume quickly, often within an hour.

The power of this approach is best seen in what it allows scientists to do. One major application is locating the precise path of an electrode that was inserted into a living animal's brain to record activity. By dipping the electrode in a fluorescent dye before insertion, the track it leaves behind glows under the microscope. The BrainSaw can image the entire brain, trace the dye, and map the exact location of the recording site in three dimensions, removing the guesswork from where the data came from. Another key use is counting cells. The system can image a whole brain at a resolution fine enough to see individual cell bodies, and automated software can then count every labeled neuron in the entire organ. This allows researchers to compare cell numbers across different animals or brain regions with a level of detail that was previously impossible to achieve at scale.

Beyond simple counting and tracking, the system reveals how different types of brain cells connect to one another. In one experiment, the researchers targeted specific inhibitory neurons in a small region of the brain called the ventral lateral geniculate nucleus. They used a virus to make these neurons glow and to mark their synaptic terminals, the tiny points where they communicate with other cells. By imaging the whole brain, they could see not just where the neurons sent their long axons, but exactly where those axons formed connections. They found that different subtypes of these neurons, even when located in the same small area, projected to different parts of the brain and formed connections in distinct patterns. This kind of detailed mapping helps scientists understand how specific circuits are built and how they might function.

The technique is also versatile enough to handle tissue that has been stained after the animal has died. The researchers showed that they could image brains from a mouse model of Alzheimer's disease that had been treated with a dye to highlight amyloid plaques, the clumps of protein associated with the disease. Even though the tissue had to be chemically treated to make it receptive to the stain, the system could still image the entire brain, revealing the widespread distribution of these plaques. This demonstrates that the method is not limited to genetically modified animals but can be applied to a wide range of biological questions, from studying the immune system in the spleen to mapping the spread of disease.

The success of the project relies heavily on the software, which is free and open for anyone to use. The authors emphasize that while the hardware requires skill to build and maintain, the software removes the barrier of needing custom code for every new setup. It allows a community of researchers to share designs and improvements, creating a standard way to perform these complex imaging tasks. The system is not perfect; it cannot image the brain of a much larger animal like a ferret because the tough outer covering of the brain is too difficult to cut, and it is not designed for the extreme magnification needed to trace a single axon down to its finest branch. However, for the vast majority of neuroscience questions that require a view of the whole brain, it offers a fast, reliable, and accessible path forward. By turning a complex, destructive process into a streamlined, automated workflow, the BrainSaw and BakingTray have made it possible for more scientists to explore the brain in three dimensions, turning the impossible task of mapping the entire organ into a routine part of modern research.