Our preliminary experience suggested that the optical fiber connection to the console did not significantly restrict manipulation of the needle, as it was light and flexible. Future needle designs could incorporate larger inner cannula diameters if smaller optical fibers are used. Future optical epidural needles could also incorporate different geometries to derive spectra from tissue regions at different distances from the needle tip. All other factors being equal, needles with larger inter-fiber distances have correspondingly larger look-ahead distances. While that may be desirable for early identification of tissues ahead of the needle tip, problems with interpreting spectra may arise if the inter-fiber distance is too large so that multiple tissue layers are sampled. If the inter-fiber distance is too small, the depth of the absorption peaks may be too small to provide robust detection of certain chromophores that exhibit relatively low absorption.
The optical epidural needle in this study was designed to derive spectroscopic contrast from tissues at the point of injection. The needle design could be modified to derive spectra from tissues surrounding the sharp tip, which could be valuable for determining whether the needle is close to puncturing a blood vessel. Such models could extend the paradigm of sensing at the beveled tip to include side-facing optical fibers that detect light incident on the NC [24
Given that the data were acquired from only a few locations in a single cadaveric specimen, the observed differences in the blood and lipid fractions provide only an initial indication of the variability of these parameters that could be encountered in different clinical contexts. We anticipate that measurements acquired in vivo will reveal considerable intra-tissue variability. Tissue heterogeneity can be expected to contribute to this variability, as can the occasional presence of additional tissues within the look-ahead distance of the needle. Additionally, minor bleeding events could result in the presence of small volumes of blood that remain close to the bevel surface and alter the spectroscopic signals and blood fractions. The extent to which this would occur during in vivo insertions is unknown. In this study, variations in the pressure imparted by the needle on tissues likely resulted in corresponding variations in the shapes and intensities of the spectra, but it is unclear whether this effect can completely account for the large variations in the spectra acquired from subcutaneous fat and from the surface of the spinal cord.
The spectral processing algorithm that was used in this study has several limitations. First, it implicitly involves the assumption that the region of tissue from which a spectrum derives is homogeneous. This assumption is likely to be violated in several cases, including situations where spectra are acquired from thin tissue layers such as the dura mater. Second, it accounts for only four chromophores that are known to give rise to prominent absorption peaks. Follow-up studies are required to determine how accurately the volume fractions of additional chromophores such as carotenes, which are known to be present in epidural fat [25
], can be estimated. The spectral processing algorithm was designed to account for differences in the magnitude and wavelength-dependence of the reduced scattering coefficient, under the assumption that the reduced scattering coefficient is much greater than the absorption coefficient [16
]. This assumption is valid for most biological tissues in the visible and near-infrared regions of the spectrum [26
]; in cases when it is not valid, the derived blood and lipid parameters may be inaccurate.
Accurately measuring the blood fraction involves several challenges. In the spectral processing algorithm, small differences in the optical absorption of hemoglobin relative to human forms [27
] were discounted. As hemoglobin is located in erythrocytes, its distribution in tissues is inhomogeneous at the cellular and vascular level. These inhomogeneities likely had small but measurable effects on the spectra [28
] that are not accounted for in the spectral processing algorithm. With optical reflectance spectra obtained from skeletal muscle, absorption from myoglobin may be prominent; as such, it could confound estimation of the blood fraction [29
]. Future spectral processing algorithms that account for the aforementioned challenges, including ones that incorporate hemoglobin packaging parameters [30
], could potentially provide accurate estimates of hemoglobin oxygen saturation and could therefore distinguish between arterial and venous blood. Another challenge encountered in the context of this ex vivo
study was the potential contamination of tissues with blood released during the course of the laminectomy; the presence of this additional blood may have elevated the measured blood fractions.
Currently, there are several methods for optically imaging within the epidural space to acquire information that is not available from the LOR technique or from fluoroscopy. With epiduroscopy, tissue ahead of the needle is visualized directly, typically by means of a flexible fiber bundle and fiber-optic illumination. This method is time-consuming and cumbersome, requiring placement of a large introducer cannula and continuous infusion of saline to distend the epidural space and create a clear path for visualization through the epiduroscope. Epiduroscopy images can be difficult to interpret, and the method is completely impractical for routine purposes like providing epidural anesthesia. Optical coherence tomography (OCT), an optical analog of ultrasound, provides micron-level spatial resolution and does not require a saline bolus for imaging within the epidural space; images can be generated with probes in direct contact with tissue [31
]. In a preliminary study, Raphael et al. demonstrated that a forward-looking OCT probe could resolve structures such as arteries and nerves in a porcine model and a dural puncture in a rabbit [33
]. At present, the histological correlates of OCT images acquired from the spinal region are not well understood, and more studies are required to determine whether OCT can provide sufficient image contrast for structures relevant to decreasing complications resulting from epidural injection.