[Paper Review] Imaging at depth in tissue with a single-pixel camera

Vicente Durán, Fernando Soldevila+5 more|arXiv (Cornell University)|Nov 11, 2014

Random lasers and scattering media24 references3 citations

TL;DR

This paper presents a noninvasive, coherent-light-free method for deep-tissue imaging using a single-pixel camera and structured incoherent illumination. By applying compressive sensing to light fluctuations measured by a single detector, the technique successfully reconstructs high-fidelity images of objects embedded in 6 mm thick chicken breast tissue at visible wavelengths, demonstrating practical potential for biomedical diagnostics without time-gated or raster-scanning systems.

ABSTRACT

One challenge that has long held the attention of scientists is that of clearly seeing objects hidden by turbid media, as smoke, fog or biological tissue, which has major implications in fields such as remote sensing or early diagnosis of diseases. Here, we combine structured incoherent illumination and bucket detection for imaging an object completely embedded in a turbid medium. A sequence of low-intensity microstructured light patterns is launched onto the object, whose image is accurately reconstructed through the light fluctuations measured by a single-pixel detector. Our technique is noninvasive, does not require coherent sources, raster scanning nor time-gated detection and benefits from the compressive sensing strategy. We experimentally retrieve the image at visible wavelengths of a transilluminated target embedded in a 6mm-thick sample of chicken breast.

Motivation & Objective

• To overcome the challenge of imaging deeply embedded objects in turbid biological media such as tissue.
• To develop a noninvasive imaging method that avoids the need for coherent light sources, which are often impractical or damaging in biological settings.
• To enable high-resolution image reconstruction in scattering media using a single-pixel detector and structured illumination.
• To demonstrate the feasibility of compressive sensing in conjunction with bucket detection for deep-tissue imaging.
• To achieve clear image recovery in thick, highly scattering samples like chicken breast without time-gated or raster-scanning techniques.

Proposed method

• The method employs structured incoherent illumination, where a sequence of microstructured light patterns is projected onto a target embedded in a turbid medium.
• A single-pixel detector collects the total transmitted light intensity (bucket signal) for each illumination pattern, capturing the integrated response of the sample.
• Image reconstruction is performed using compressive sensing algorithms, leveraging the sparsity of the target image in a suitable basis to recover the object from fewer measurements than traditional methods.
• The technique avoids coherent illumination, time-gated detection, and mechanical scanning, relying instead on statistical correlation between illumination patterns and measured intensities.
• The reconstruction process uses a known measurement matrix derived from the illumination patterns and solves an optimization problem to recover the image.
• The approach is experimentally validated using a 6 mm thick chicken breast sample as a scattering medium, with visible-wavelength illumination and a single-pixel detector.

Experimental results

Research questions

• RQ1 Can high-fidelity imaging of deeply embedded objects be achieved in highly scattering biological tissue using incoherent light and a single-pixel detector?
• RQ2 To what extent can compressive sensing reduce the number of measurements required for image reconstruction in turbid media without coherent illumination?
• RQ3 Can a single-pixel camera system achieve reliable image recovery in thick, scattering samples without time-gated detection or raster scanning?
• RQ4 How does the performance of the method compare to conventional imaging techniques in terms of resolution, signal-to-noise ratio, and practicality in biological settings?
• RQ5 What is the maximum imaging depth achievable with this incoherent, single-pixel, compressive sensing approach in ex vivo tissue?

Key findings

• The method successfully reconstructed a high-fidelity image of a transilluminated target embedded in a 6 mm thick chicken breast sample using only a single-pixel detector and incoherent illumination.
• The image reconstruction was achieved without the use of coherent light sources, time-gated detection, or mechanical raster scanning, significantly simplifying the experimental setup.
• The technique leveraged compressive sensing to recover the image from a number of measurements substantially lower than the number of pixels in the target image.
• The experimental results demonstrated that the single-pixel camera system could resolve fine structural details of the target despite strong light scattering in the tissue.
• The approach maintained image quality and contrast even at depths where conventional imaging would fail due to multiple scattering and signal degradation.
• The study confirms the feasibility of using incoherent structured illumination and bucket detection for deep-tissue imaging in a noninvasive, cost-effective, and biocompatible manner.