Machine learning-based classification system using depth-dependent variation encoding for classifying cervical two-photon excited fluorescence image stacks

Sanghavi, Foram M.; Panetta, Karen; Agaian, Sos S.; Polleys, Christopher; Thieu, Hong-Thao; Genega, Elizabeth M.; Georgakoudi, Irene

doi:10.1117/12.2623305

Cited by 2 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our probe is also capable of collecting images at different depths, owing to axial actuation via a dc servo motor. This is significant, since published examples of metabolic tissue characterization for cancer detection often relies on depth-dependent variation of the parameters of interest [ 14 , 37 , 38 ]. Other features of the probe include flexible delivery of ultrashort pulses with minimal nonlinear effects using a hollow core photonic bandgap fiber and focusing to sub-micron spot size by designing a miniaturized objective consisting of commercially available lenses.…”

Section: Introductionmentioning

confidence: 99%

Two photon imaging probe with highly efficient autofluorescence collection at high scattering and deep imaging conditions

Camli,

Andrus,

Roy

et al. 2024

Biomed. Opt. Express

Self Cite

View full text Add to dashboard Cite

In this paper, we present a 2-photon imaging probe system featuring a novel fluorescence collection method with improved and reliable efficiency. The system aims to miniaturize the potential of 2-photon imaging in the metabolic and morphological characterization of cervical tissue at sub-micron resolution over large imaging depths into a flexible and clinically viable platform towards the early detection of cancers. Clinical implementation of such a probe system is challenging due to inherently low levels of autofluorescence, particularly when imaging deep in highly scattering tissues. For an efficient collection of fluorescence signals, our probe employs 12 0.5 NA collection fibers arranged around a miniaturized excitation objective. By bending and terminating a multitude of collection fibers at a specific angle, we increase collection area and directivity significantly. Positioning of these fibers allows the collection of fluorescence photons scattered away from their ballistic trajectory multiple times, which offers a system collection efficiency of 4%, which is 55% of what our bench-top microscope with 0.75 NA objective achieves. We demonstrate that the collection efficiency is largely maintained even at high scattering conditions and high imaging depths. Radial symmetry of arrangement maintains uniformity of collection efficiency across the whole FOV. Additionally, our probe can image at different tissue depths via axial actuation by a dc servo motor, allowing depth dependent tissue characterization. We designed our probe to perform imaging at 775 nm, targeting 2-photon autofluorescence from NAD(P)H and FAD molecules, which are often used in metabolic tissue characterization. An air core photonic bandgap fiber delivers laser pulses of 100 fs duration to the sample. A miniaturized objective designed with commercially available lenses of 3 mm diameter focuses the laser beam on tissue, attaining lateral and axial imaging resolutions of 0.66 µm and 4.65 µm, respectively. Characterization results verify that our probe achieves collection efficiency comparable to our optimized bench-top 2-photon imaging microscope, minimally affected by imaging depth and radial positioning. We validate autofluorescence imaging capability with excised porcine vocal fold tissue samples. Images with 120 µm FOV and 0.33 µm pixel sizes collected at 2 fps confirm that the 300 µm imaging depth was achieved.

show abstract

Section: Introductionmentioning

confidence: 99%

Two photon imaging probe with highly efficient autofluorescence collection at high scattering and deep imaging conditions

Camli,

Andrus,

Roy

et al. 2024

Biomed. Opt. Express

Self Cite

View full text Add to dashboard Cite

show abstract

“…Our endoscope also has the capability to perform axial scanning with a dc servo motor to be able to obtain images at different depths of the epithelium in a clinical setting. This is significant, since recent work in the literature suggests depth dependent information about metabolic and morphological changes can be relevant in classification of healthy and precancerous tissues 11 .…”

Section: Introductionmentioning

confidence: 99%

2p autofluorescence imaging endoscope for clinical observation of metabolic changes in cervical tissue

Camli

Andrus²,

Roy³

et al. 2023

Endoscopic Microscopy XVIII

View full text Add to dashboard Cite

Imaging modalities capable of detecting functional changes over small areas can increase sensitivity and specificity of early cancer detection. Label-free imaging of metabolic activity at cellular level resolution over full thickness of cervix epithelium is possible with 2p imaging. However, low probability of 2p excitation and scattering nature of tissues limit autofluorescence levels in 2p imaging. We present a 2p autofluorescence imaging endoscope system for detection of metabolic changes in cervix in a clinical setting, with an increased collection efficiency in scattering media. Collection of autofluorescence signals is done with a multitude of high NA fibers arranged around a miniaturized excitation objective. By cleaving the collection fibers at a specific angle, we increase the directivity of the collection and the collection efficiency per fiber. The endoscope performs imaging at 775 nm, which is capable of exciting NAD(P)H and FAD molecules. Laser pulses of 100 fs duration are delivered to the sample with an air core photonic bandgap fiber. Fiber is scanned in spiral pattern via a piezo actuator tube. Scanning at different tissue depths is possible with the axial actuation of the endoscope via a linear stepper motor. Benchtop tests indicate that the endoscope system has lateral and axial resolutions of 0.65 μm and 4.33 μm, respectively. Fluorescence images of pollen cores are presented to demonstrate the imaging quality of the endoscope system.

show abstract

Machine learning-based classification system using depth-dependent variation encoding for classifying cervical two-photon excited fluorescence image stacks

Cited by 2 publications

References 0 publications

Two photon imaging probe with highly efficient autofluorescence collection at high scattering and deep imaging conditions

Two photon imaging probe with highly efficient autofluorescence collection at high scattering and deep imaging conditions

2p autofluorescence imaging endoscope for clinical observation of metabolic changes in cervical tissue

Contact Info

Product

Resources

About