Towards Real-Time Hyperspectral Multi-Image Super-Resolution Reconstruction Applied to Histological Samples

Area of Science:

• Medical Imaging
• Spectroscopy
• Computational Imaging

Background:

• Hyperspectral Imaging (HSI) offers valuable spectral signature analysis for organic and non-organic elements in medical applications.
• Acquiring HSI data is challenging, with limited commercial sensors and often insufficient spatial resolution in critical spectral bands.
• Large data volumes in HSI pose significant image processing challenges.

Purpose of the Study:

• To enhance the spatial resolution of hyperspectral histology samples using super-resolution techniques.
• To develop a computationally efficient algorithm for HSI super-resolution.
• To maintain the spectral integrity of pixels during spatial resolution enhancement.

Main Methods:

• Utilizing multiple hyperspectral images of the same scene captured with sub-pixel shifts.
• Implementing a novel super-resolution algorithm designed for hyperspectral microscopic systems.
• Focusing on a low computational intensity approach for practical image processing.

Main Results:

• Successfully enhanced the spatial resolution of hyperspectral histology samples.
• The proposed algorithm maintains the spectral signatures of individual pixels.
• Achieved performance comparable to state-of-the-art super-resolution techniques.
• Demonstrated a computationally efficient solution for HSI data.

Conclusions:

• The developed super-resolution approach effectively improves spatial detail in hyperspectral histology.
• The method is computationally efficient, addressing a key challenge in HSI processing.
• This technique holds promise for real-time applications in medical hyperspectral imaging.