INTRODUCTIONConventional nerve histomorphometry is resource intensive, necessitating complex sample preparation and error-prone axon quantification. Stimulated Raman scattering (SRS) microscopy and third harmonic generation (THG) are label-free, nonlinear optical processes with minimal sample preparation requirements that enable high-contrast and high-resolution imaging of histologic samples. Herein, we demonstrate the utility of these emerging imaging techniques for label-free resolution of myelin in rodent and human peripheral nerve. We further train a machine learning algorithm for rapid quantification of myelinated axons from imaged sections and demonstrate its applicability to surgical practice.

METHODSFresh frozen sections of healthy rat sciatic nerve and human motor branch of obturator nerve were employed. Sections were left unstained and imaged by SRS and THG (using a custom-assembled multiphoton microscope), or stained with FluoroMyelin® and imaged by widefieldfluorescent microscopy. Myelinated axon counts were automatically quantified across entire cross-sections using a trainable random forest machine learning algorithm.

RESULTSRobust visualization of myelinated axons in unstained sections was possible with SRS and THG imaging. Rapid quantification of myelinated axon counts in healthy rodent and human nerve was achieved using a machine learning algorithm.

CONCLUSIONSA rapid protocol for quantification of myelinated axon counts from peripheral nerves employing label-free imaging techniques with minimal sample preparation has been described.

Iván Coto Hernández, PhD1; Wenlong Yang, PhD2; Lars Rishø, PhD3; Suresh Mohan, MD1; Siddharth Ramachandran, PhD3; Nate Jowett, MD1

1Surgical Photonics & Engineering Laboratory, Massachusetts Eye and Ear Infirmary, Harvard Medical School. 2Center for Advanced Imaging, Harvard University. 3Boston University

REFERENCES

ABSTRACT

CONTACT

Novel techniques for high-throughput peripheral nerve histomorphometry

Iván Coto Hernández, PhDMassachusetts Eye and Ear Infirmary Harvard Medical School, Boston, MA, USAEmail: ivan_cotohernandez@meei.harvard.edu

Fig 1. (A-D) SRS imaging of sciatic nerve from Thy1-YFP mice. 2PE image of axons and SRS of myelin in alongitudinal sciatic nerve section (30µm depth). (E-G) 2PE image of axons and SRS of myelin in a sciaticnerve cross-section (2µm). Scale bar 40 μm.

1. Tian, F. et al. Monitoring peripheral nerve degeneration in ALS by label-free stimulated Raman scattering imaging. Nat. Commun. (2016). doi:10.1038/ncomms13283

2. Lim, H. et al. Label-free imaging of Schwann cell myelination by third harmonic generation microscopy. Proc. Natl. Acad. Sci. 111, 18025–18030 (2014).Fig 3. Automated quantification of human nerve with SRS aided by machine learning. Scale bar 40 μm.

Fig 2. Chemical labeling versus label-free imaging of human nerve. (A-B) Motor branch of obturator nerve.(C-D) Cross-face nerve graft (CFNG) tip. Scale bar 40 μm.

Fig 4. Multiphoton imaging of sciatic nerve of a Thy-YFP mouse. Third-harmonic generation (blue) shows the myelin fibers;second-harmonic generation (red) images collagen fibers; and three-photon excited fluorescence (green) show the axons.

Fig 6. Label-free imaging of ex vivo human obturator nerve. (A) Second-harmonic generation (red) images collagen fibers. (B)Third-harmonic generation (green) shows the myelin fibers. (C-D) Merge. The scale bar is 40 µm.

Fig 5. Power dependence for 3P, THG, and SHG, respectively.