Quantitative phase imaging through an ultra-thin lensless fiber endoscope

Quantitative phase imaging (QPI) offers a label-free, non-invasive method for 3D imaging of transparent samples with nanoscale sensitivity to morphology and dynamics. It extracts quantitative biophysical parameters like refractive index, dry mass, and density from the phase shift, providing valuable information for digital pathology. Recent advancements combine QPI with deep learning for virtual staining and blood examination, even showing potential for high-throughput SARS-CoV-2 detection. However, current QPI methods rely on bulky and expensive microscope platforms with limitations in working distance and penetration depth, hindering in vivo applications, particularly in early cancer diagnosis. Multi-core fiber bundles (MCFs) offer ultra-thin probes for minimally invasive in vivo imaging, but existing MCF techniques are limited to amplitude imaging. This research addresses the need for a simple, cost-effective 3D microendoscope with nanoscale sensitivity by utilizing the MCF as a phase encoder. The proposed computational approach, named far-field amplitude-only speckle transfer (FAST), decodes the incident light field from far-field speckles, enabling 3D QPI reconstruction with nanoscale axial sensitivity and microscale lateral resolution. The system overcomes the limitations of conventional fiber facet imaging, offering significant advancements for in vivo clinical applications.

The introduction extensively reviews existing QPI techniques and their limitations in in vivo applications due to bulky equipment and limited working distance. It highlights the potential of multi-core fiber bundles (MCFs) for minimally invasive imaging but notes the current limitations of MCF-based imaging to amplitude modalities. The authors discuss previous attempts to recover 3D information from MCFs, including techniques using multi-mode fibers with transmission matrix measurements or wavefront shaping. These methods, however, often involve bulky optical systems, spatial light modulators, complicated calibration processes, and slow scanning, limiting their clinical applicability. The literature review underscores the lack of an endoscope with nanoscale optical path length sensitivity, emphasizing the need for a simpler, more cost-effective solution.

The study proposes a novel computational lensless microendoscope using an ultra-thin MCF. The MCF acts as a phase encoder, transforming the incident complex light field into a far-field speckle pattern at the detection side. The core innovation is the FAST (far-field amplitude-only speckle transfer) method, which reconstructs the incident light field from the intensity-only far-field speckles. The process begins with a reference measurement using a collimated laser or point light source to capture the intrinsic optical path length (OPL) differences of the fiber cores. Subsequently, the sample (e.g., a resolution test target, glass beads) is placed at a distance from the fiber facet. The far-field speckle pattern is recorded by a camera, and the FAST algorithm reconstructs the phase and amplitude information on the fiber facet. The original phase incident on the fiber bundle is decoded using equation (1): φ' = φ₁ - φ₀, where φ₁ and φ₀ represent the phase with and without the sample, respectively. The incident light field is then numerically back-propagated to the sample plane using the angular spectrum method to reconstruct the sample's 3D image at various axial distances. This digital refocusing capability eliminates the need for mechanical movement. The accuracy and resolution of the technique are validated using various experiments.

The research successfully demonstrates quantitative phase imaging through an ultra-thin fiber endoscope using the FAST method. The technique allows for digital refocusing, enabling 3D imaging of multi-layered samples without mechanical adjustments. Experiments with a USAF resolution target demonstrate the system's ability to resolve fine details, achieving a lateral resolution of up to 1 µm and nanoscale axial sensitivity. Imaging of glass beads flowing through a microchannel showcases the system's dynamic imaging capabilities. The quantitative phase imaging of a phase target demonstrates the accurate reconstruction of phase values, enabling the calculation of optical path differences (OPDs). The OPDs correlate with the refractive index and thickness of the sample, providing quantitative biophysical information. The 3D optical path difference map further illustrates the high fidelity of the phase reconstruction. The lateral resolution is measured across a range of phase values providing data on the reliability of the system. The successful 3D imaging of human cancer cells through the ultra-thin fiber endoscope highlights the significant potential of this technology for in vivo clinical applications.

The successful implementation of the FAST method for quantitative phase imaging through an ultra-thin fiber endoscope represents a significant advancement in minimally invasive biomedical imaging. The digital refocusing capability and nanoscale sensitivity of the system overcome limitations of previous QPI and MCF-based imaging techniques. The ability to acquire both morphological and quantitative biophysical information offers considerable advantages for disease diagnosis. The demonstrated application of the technique to image human cancer cells strongly suggests its potential for in vivo clinical applications such as early cancer detection and minimally invasive biopsies. This technology could enable more precise and less invasive diagnostic procedures, potentially improving patient outcomes. The high accuracy and resolution achieved indicate the robustness of the FAST method.

This research presents a novel computational lensless microendoscope for quantitative phase imaging using an ultra-thin MCF. The FAST method enables accurate reconstruction of the incident complex light field from far-field speckles, facilitating digital refocusing and 3D imaging with high resolution and nanoscale axial sensitivity. The validation experiments demonstrate the accuracy and potential of this technology for biomedical imaging. The successful imaging of human cancer cells through the fiber endoscope highlights the significant potential of this technology for minimally invasive in vivo clinical applications. Future research could focus on optimizing the algorithm for faster processing and exploring further applications in different biomedical settings.

While the study demonstrates impressive results, potential limitations exist. The current system may be sensitive to environmental factors affecting the stability of the fiber bundle and the speckle patterns. Further research is needed to assess the long-term stability and robustness of the system in complex in vivo environments. The computational demands of the FAST algorithm might limit the real-time imaging speed; further optimization is warranted. Finally, extensive in vivo studies are needed to fully validate the clinical utility of this technology.