Reconstruction of undersampled scanning ion conductance microscopy images through zero-shot learning-based...

Area of Science:

• Biomedical Imaging
• Nanotechnology
• Computational Microscopy

Background:

• Scanning Ion Conductance Microscopy (SICM) offers non-contact, nanoscale imaging crucial for biomedical research.
• Slow imaging speeds in SICM hinder the observation of rapid biological processes.
• Current accelerated imaging methods like compressed sensing (CS) and deep learning face limitations in reconstruction quality and data requirements for SICM.

Purpose of the Study:

• To develop an accelerated imaging framework for SICM using computational reconstruction.
• To overcome the limitations of existing methods by proposing a data-efficient approach.
• To enable high-speed SICM for capturing dynamic biological events.

Main Methods:

• A zero-shot super-resolution (SR) framework utilizing artificial neural networks is proposed.
• The method leverages internal image statistics to train image-specific SR networks, eliminating the need for external datasets.
• Reconstruction from undersampled SICM measurements is performed.

Main Results:

• The proposed zero-shot SR method demonstrates superior reconstruction accuracy compared to bicubic interpolation, CS, and baseline zero-shot SR (ZSSR) algorithms.
• It achieves comparable image quality with significantly fewer sampling points than CS methods, enabling faster imaging.
• The framework shows robust performance under noisy conditions, proving its practical applicability.

Conclusions:

• The developed zero-shot SR framework provides a practical strategy for high-speed SICM imaging.
• This approach effectively reconstructs high-fidelity images from undersampled data, overcoming data scarcity issues.
• The study paves the way for enhancing imaging speed in other data-scarce scanning probe microscopy techniques.