1. Introduction
Caries lesions can be arrested by the preferential deposition of mineral at the lesion surface that inhibits diffusion of fluids [1-4]. Since arrested lesions do not need further intervention, the assessment of lesion activity is essential for clinical diagnosis. Conventional visual and tactile methods of lesion activity assessment are not reliable [5]. Gold standards for lesion assessment such as transverse microradiography (TMR) and polarized light microscopy (PLM) either require destruction of the tooth, or in the case of microcomputed tomography (microCT) are not suitable for use in-vivo . Non-destructive diagnostic tools are needed that can assess lesion activity in a single visit. New optical diagnostic technologies that can monitor optical changes in the light scattering of lesion structures have great potential for assessing activity. Optical coherence tomography (OCT) can be used to assess lesion activity in a single measurement nondestructively in vivo; however such systems are expensive and there are no systems currently available for clinical use. OCT is capable of detecting the presence of a highly mineralized surface zone near the lesion surface which appears as a transparent surface zone due to the reduced reflectivity [4, 6, 7]. Short wavelength infrared (SWIR) imaging offers a novel and complementary approach that exploit changes in light scattering and absorption that occur in the lesion during the loss of water from the lesion structure [8, 9].
When lesions become arrested by mineral deposition, or remineralization of the outer layers of the lesion, the diffusion of fluids into the lesion are inhibited. Hence, the rate of water diffusion out of the lesion or the evaporation kinetics or dynamics reflects the degree of lesion activity. Since arrested lesions are less permeable to water due to the highly mineralized surface layer, changes in the rate of water loss can be related to changes in lesion structure and porosity. Changes in fluorescence loss [10-12], thermal emission, and SWIR reflectance [8, 9, 13-17] during lesion dehydration have been investigated as methods for assessing lesion activity. Sound enamel is transparent at SWIR wavelengths, whereas early demineralization causes increased SWIR reflectance due to scattering [18]. Water in the pores at the surface of the lesion absorbs the incident SWIR light, particularly at wavelengths greater than 1400 nm, reducing surface scattering and lesion contrast [19, 20]. Loss of that water due to evaporation produces a marked increase in reflectivity and lesion contrast. In-vivostudies have been published utilizing the fluorescence loss of white spot lesions on coronal surfaces [12] and thermal imaging to assess root caries during dehydration [21]. There is a negative association between the surface zone thickness and lesion permeability; a small increase in the surface layer thickness of less than 20 µm can lead to a marked decrease in permeability [22]. In addition, in a closely related study the surface zone was removed from arrested lesions producing a corresponding rise in the permeability providing further confirmation of the role of the surface zone in arresting lesions [23]. Recent studies have shown that the water absorption bands at 1450 and 1950 nm are advantageous for SWIR reflectance dehydration measurements and produce higher contrast between sound and demineralized enamel [24, 25].
The first clinical study to utilize SWIR reflectance dehydration measurements was recently carried out on the occlusal surfaces of primary teeth using a compact InGaAs camera and 1400-1750 nm light [26]. The presence of a highly mineralized transparent surface zone was also assessed using cross polarization optical coherence tomography (CP-OCT). The SWIR imaging results were promising showing that the delay between the application of the forced air and the rise in reflectivity is a good indicator of lesion activity. However, there was difficulty fully dehydrating the lesions in the pits and fissures and acquiring curves suitable for complete analysis of the dehydration kinetics. That study indicated that additional work is needed to improve the acquisition of dehydration curves in vivo and further develop methods for analysis of dehydration curves acquired from the pits and fissures of the occlusal surfaces. In this study, we choose to use the 1450 nm water absorption band for imaging as opposed to the 1950 nm water absorption band because compact extended range InGaAs cameras suitable for clinical use operating beyond 1750 nm are not yet available and our goal was to assess the performance of a handheld system that can be used clinically at this time.
Multiple parameters describing the shape of the curves representing the rate of intensity change during dehydration can be derived from these curves including delay, ΔI, %Ifin, and rate [25]. Delay represents the initial delay between the time air is applied and the rise in intensity, ΔI is the change in intensity before and after dehydration, the rate represents the shape of the rise in intensity and once the rise begins that typically takes a sigmoidal shape for active lesions, and %Ifin represents the fraction of the intensity rise that occurs after the most rapid change in intensity. For active lesions there is typically an initial delay before the rise in intensity occurs because the water in the pores near the lesion surface absorb the incident SWIR light. The curves for active lesions experience a rapid/steep rise in intensity (high rate) to quickly plateau in intensity (low %Ifin) since the water in the body of the lesion can more rapidly escape through the open pores. For arrested lesions there is a minimal delay since the highly mineralized surface layer lacks porosity and water. Dehydration occurs slowly (low rate) because the highly mineralized surface layer blocks the pores in the lesion and inhibits the loss of water and much of the intensity change occurs slowly in the tail of the curve (high %Ifin). The rate of increase in intensity was assessed using the Hill equation that was developed to assess ligand binding in biochemistry [27]. Previous studies used the sigmoid function to calculate an overall growth rate (OGR) calculated from two coefficients [8, 9] but we found that better fits could be achieved using the Hill equation. Both functions were designed to fit sigmoidal shaped curves. The Hill equation is given by equation 1.