The mesoscale structure of chitosan in electrodeposited films is of considerable interest as it affects local molecular transport, cell guidance, and write/read/storage capabilities of the material. To understand how fabrication conditions affect the mesoscale structure of chitosan, time-lapse microscopy of birefringence signal, related to chitosan molecular concentration and organization, and co-registered brightfield signal, indicating the moving front of the film boundary, were collected from a film deposited on a cathode from a 1% (w/v) chitosan solution during chronopotentiometry. Birefringence time-series images were predicted using a deep learning neural network approach, then applied to different chitosan films formed in the presence of 1% agarose. Birefringence signal extends within the chitosan film to the moving front and grows in the film with continued operation of the cathode. A multi-scale residual neural network trained 197,761 parameters in about 8 seconds/epoch to learn birefringence from co-registered brightfield in 256x256x256 (pixel x pixel x time series frame) datasets. Different transformations of the image-series training data were evaluated: x-y, x-t, and y-t. The structural similarity index of birefringence signal predictions to ground truth was highest from y-t trained data, which captured space-time covariance. Predicted birefringence signal from the chitosan-agarose interpenetrating network had a lower structural similarity index compared to the ground truth chitosan-only network. The difference, visible in the structural similarity index map, was characterized as a loss of striated texture and sharp interface in birefringence from chitosan-agarose. This deep learning approach is proposed as a way to couple optical signals from chitosan across multiple fabrication conditions to determine relative effects on dynamic spatiotemporal patterning.

Deep learning prediction of optical signals in self-assembly experiments of electrofabricated chitosan films

Macromolecular Chemistry: The Second Century

[Author Benefits](https://scimeetings.acs.org/?utm_source=pubs_content_marketing&utm_medium=website&utm_campaign=0320_MCF_NPI_Launch_Spring_Homepage&ref=pubs_content_marketing)

[How to use SciMeetings](https://storage.googleapis.com/dl.morressier.com/acs/SciMeetings-ACSSpring2020-HowToGuide_M3020-Final.pdf)

[Frequently Asked Questions](https://storage.googleapis.com/dl.morressier.com/acs/SciMtng_ACSNMSp2020_FAQ_M3120_Final.pdf)

ACS Spring 2020 National Meeting & Expo

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.