BioEMTalks: ConnectomX and the Katana Microtome — Precision Engineering for Serial Block-Face Imaging
Abstract
Founded in 2018 and based in Oxford, ConnectomX specialises in advancing serial block-face imaging (SBFI). In this BioEMTalks presentation, we provide an overview of the company, the Katana microtome, and the key technological innovations that enable nanoscale precision in a uniquely compact form factor.
The Katana Microtome
Unlike traditional, bulky alternatives, the Katana is an ultra-compact and portable in-situ microtome. Standing at just 56 millimetres tall, it sits directly on the microscope stage. Through advanced engineering and simulation, ConnectomX achieved single-nanometre precision and high structural stiffness within this tiny form factor, making it one of the leading commercial solutions for serial block-face imaging.
Key Technological Innovations
Advanced Oscillating Knife
The Katana uses piezoelectric actuators to actively control the diamond knife's lateral movement. This significantly reduces cutting force, allowing the microtome to cleanly slice through incredibly delicate structures without tearing them — a critical capability when working with sensitive biological samples at the nanoscale.
Automated Cutting Diagnostics
One of the persistent challenges in serial block-face imaging is visually undetectable cutting errors. ConnectomX has developed an automated diagnostics system that analyses image correlation throughout a stack, automatically alerting users to irregular cuts — such as alternating thick and thin sections — before they compromise the dataset.
Knife-Edge Tracking Camera
A specialised mirror and camera system stays continuously focused on the moving knife edge. This eliminates blur, ensures safe sample approaches, and allows researchers to monitor the most subtle interactions between the knife and the sample surface in real time.
Physical Voxel Printing
As an outreach and data visualisation project, ConnectomX developed a unique method to 3D print datasets directly from raw biological volume data (voxels), bypassing the standard requirement to convert data into surface meshes. This approach allows the physical recreation of massive biological structures — some containing up to 4.9 billion printed voxels — bringing volume electron microscopy data into the real world.