Shape-Encoded Hydrogel Sensor Particles Enable Multiplex Odorant Detection Through Deep-learning Classification.

Area of Science:

• Biotechnology
• Sensor Technology
• Machine Learning

Background:

• Developing portable, cell-based biohybrid sensors for simultaneous odorant detection faces challenges in distinguishing sensor cell types.
• Current methods struggle with reliable identification of diverse sensor cells within a single sensing platform.

Purpose of the Study:

• To develop a shape-encoding strategy for hydrogel particles to enable shape-based identification of distinct sensor cell types.
• To apply deep learning for accurate classification of particle shapes and enable multiplexed odorant detection.

Main Methods:

• Hydrogel particles were engineered into unique shapes, each corresponding to a specific sensor cell type expressing a distinct odorant receptor (OR).
• A convolutional neural network (CNN) was trained to classify particle shapes from fluorescence images.
• The shape identification scheme was applied to time-lapse images of mixed particles exposed to single odorants.

Main Results:

• The CNN achieved high accuracy in classifying particle shapes, enabling reliable assignment of particle identity.
• Shape-specific fluorescence signals were extracted, revealing distinct odorant-dependent responses.
• Observed responses correlated with the known ligand specificities of the expressed odorant receptors.

Conclusions:

• Shape-encoded hydrogel particles combined with deep learning offer a scalable, position-independent method for multiplexed odorant detection.
• This framework facilitates the development of compact, high-throughput biohybrid sensors for safety, environmental monitoring, and diagnostics.