2D multi-layer thickness measurement with reconstructed spectrum

#Dry eye has become one of the most common causes for ophthalmological doctor visits. Dry eye is a multifactorial disease of the ocular surface that is related to the tear film. One method to diagnosing dry eye is to assess the quantity of tears by measuring the thickness of the tear film. As illustrated in FIG. 1, the tear film 100 comprises the outer layers of the eye including a lipid layer 102 that is about 50 nm thick, and a muco-aqueous layer 104 (also referred to as a mucous and/or aqueous layer) that is collectively about 1.5 μm thick. The eye further includes the cornea 106 following the muco-aqueous layer, which is about 0.5 mm thick. Currently, few technologies exist for accurately and efficiently imaging and analyzing the layers of the tear film 100 (e.g., determining layer thickness), for example, to objectively assist dry eye diagnosis.

#Interferometric techniques are among the currently available non-invasive measurements. Of these, one approach relies on a correlation between an image color and the lipid layer thickness, either quantitatively or qualitatively. Theoretically, the analysis is performed based on two-dimensional (2D) images, while typically only the average thickness within a fairly large area is presented. However, this approach is usually limited for relative lipid layer thickness estimates, and may be susceptible to phase ambiguity and uncertainty when performing absolute thickness measurements.

#According to a first example of the present disclosure, a method for measuring layer thickness of a structure comprises: acquiring a multi-spectral interference pattern of the structure; performing a hyperspectral reconstruction on the multi-spectral interference pattern, thereby generating a reconstructed full- or hyper-spectral interference pattern; and estimating the layer thickness based on the reconstructed full- or hyper-spectral interference pattern.

#FIG. 1 schematically illustrates the tear film and cornea.

#FIG. 2 illustrates an example thickness estimation method of the present disclosure that uses hyper-spectral reconstruction of multi-spectral band images.

#FIG. 3 graphically illustrates the method of FIG. 2.

#As presented herein, interferometry may be used to measure layer thicknesses of an object by imaging the object with multiple spectral bands. For example, in some embodiments, a plurality of spectral bands (and in particular, “narrow bands,” which have a narrower spectral bandwidth than conventional RGB imaging) may be used to provide a sufficient signal strength for 2D layer thickness measurement. Further, narrow spectral bandwidths can be used to help enhance the imaging contrast, particularly for muco-aqueous layer measurement. In some embodiments, these multi-spectral bands may be non-overlapping bands or overlapping bands from a RGB camera. Tear film thickness at each imaged location can then be estimated with look-up-table, curve-fitting, or machine learning methods, based on a full- or hyper-spectral interference pattern reconstructed from the imaged spectral patterns.

#The number of spectral bands can affect the accuracy and applications of such imaging. Using a relatively larger number of spectral bands can increase accuracy, but is time consuming and thus can limit the ability to measure dynamic objects such as the tear film. On the other hand, a relatively fewer number of spectral bands can allow live-speed measurement, but may not provide sufficient information for accurate curve-fitting. Hyperspectral reconstruction of the spectral bands allows for increased accuracy with a decreased number of spectral bands by reconstructing a full- or hyper-spectrum response from limited spectrum band measurements. In other words, the hyper-spectral reconstruction produces an interference pattern similar to one that might have been captured with a full-spectrum camera (and thus having the associated benefits of additional data points yielding increased thickness estimation accuracy), while maintaining the benefits of multi-spectral imaging.

#It should be appreciated that as used herein, the term “full spectrum” refers to a continuous spectral distribution, “hyper-spectral” typically refers to at least 100 discrete spectral bands, and “multi-spectral” typically refers to about 20 discrete spectral bands. In many instances, the prefixes ‘full’ and ‘hyper’ are used interchangeably to generically refer to spectral reconstructions across many frequencies and bandwidths. Further, it is noted that multi-spectral signals may have more or less than 20 discrete spectral bands, and hyper-spectral signals may have more or less than 100 discrete spectral bands, so long as the hyper-spectral signals have more discrete spectral bands than the multi-spectral signals.