Massively Parallel Image Processing Using Molecular Algorithms
IEEE Transactions on Molecular, Biological, and Multi-Scale Communications · 2026 · 12 · 1236-1252
DOI: 10.1109/tmbmc.2026.3737672 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Molecular, Biological, and Multi-Scale Communications, 2026. Verfasst von Christopher James Austin und 2 weiteren Autor:innen. Die Arbeit nennt 48 Literaturangaben. Gefördert von: DNA Microfactory for Autonomous Archiving, Next Generation Molecular Data Storage, Synthetic and Scalable Data Platform for Medical Empowered AI (AISym4MED) Projects under the European Union's Horizon Europe Research and Innovation Programme.
Abstract (Originalsprache)
Digital images are growing rapidly in size and number, while DNA-based storage and molecular computation have mostly been studied separately. We present a theoretical SIMD $\|$ DNA architecture for image storage and in-molecule image processing, consisting of a multi-cell-type encoding and a library of interoperable algorithms for inversion, thresholding, brightness adjustment, noising, and horizontal flipping. The main contribution of the paper is architectural: we show how image structure can be embedded into SIMD $\|$ DNA registers so that a common encoding supports multiple image-processing programs and exposes useful forms of parallelism across bits, channels, pixels, and images. We also report a limited wet-lab proof of concept for a simplified inversion pipeline on a 1-bit, $8\times 8$ image, together with a synthetic sequence-design workflow and a readout strategy based on Klenow-fragment elongation. The experiment recovers 25/64 pixels correctly (39.1%), revealing substantial gaps between theoretical design, kinetic simulation, and current laboratory performance. We therefore present the experimental component as preliminary validation of part of the architecture rather than as validation of the full multi-cell-type system, and we use it to analyse failure modes, sequence-design weaknesses, and directions for improving reproducibility and reliability.
Förderung
- DNA Microfactory for Autonomous Archiving
- Next Generation Molecular Data Storage
- Synthetic and Scalable Data Platform for Medical Empowered AI (AISym4MED) Projects under the European Union's Horizon Europe Research and Innovation Programme
Zitieren
APA
Austin, C. J., Egan, I. S., & Heinis, T. (2026). Massively Parallel Image Processing Using Molecular Algorithms. IEEE Transactions on Molecular, Biological, and Multi-Scale Communications, 12, 1236-1252. https://doi.org/10.1109/tmbmc.2026.3737672
Vancouver
Austin CJ, Egan IS, Heinis T. Massively Parallel Image Processing Using Molecular Algorithms. IEEE Transactions on Molecular, Biological, and Multi-Scale Communications. 2026;12:1236-1252. doi:10.1109/tmbmc.2026.3737672
BibTeX
@article{Austin2026,
author = {Austin, Christopher James and Egan, Isabella Sequeiros and Heinis, Thomas},
title = {Massively Parallel Image Processing Using Molecular Algorithms},
journal = {IEEE Transactions on Molecular, Biological, and Multi-Scale Communications},
year = {2026},
volume = {12},
pages = {1236--1252},
doi = {10.1109/tmbmc.2026.3737672}
}
EUVIMED