Reviewed by Mila PereraSep 26 2022
Biomedical engineers from Duke University have developed a new method of cipher creation using bacterial growth patterns.
Simulated bacterial growth patterns such as this can be used to encode secret messages. Image Credit: Duke Pratt School of Engineering
Bacteria grow in particular ways depending on circumstances such as nutrient levels and space restrictions.
The researchers created a virtual bacterial colony and then adjusted growth parameters and the quantities and sizes of simulated bacterial dots to create an alphabet based on how colonies appeared after filling a petri dish. This encoding technique is known as emorfi.
The final simulated pattern assigned to each letter does not always match perfectly, so the encoding is not one-to-one. The researchers did find, however, that a machine learning software application could learn to differentiate between the patterns to identify the intended letter.
A friend may see many images of me over the course of time, but none of them will be exactly the same. But if the images are all consistently reinforcing what I generally look like, the friend will be able to recognize me even if they are shown a picture of me, they have never seen before.
Lingchong You, Professor, Biomedical Engineering, Duke Pratt School of Engineering
The encoder produces a film out of a sequence of patterns to encrypt messages, each corresponding to a distinct letter. Although these patterns appear similar to the untrained eye, the computer program can tell them apart.
Provided that the receiver knows the original conditions of their development, they should be able to read the coded messages, while an outsider should not be able to decipher the code without their own AI.
The David and Lucile Packard Foundation, the Office of Naval Research (N00014-20-1-2121), the National Science Foundation (MCB-1937259), and the Google Cloud Research Credits program provided funding for this study.
Journal Reference
Liu, J., et al. (2022) Distributed information encoding and decoding using self-organized spatial patterns. Patterns. doi:10.1016/j.patter.2022.100590