Triage of in vivo burn injuries and prediction of wound healing outcome using neural networks and modeling of the terahertz permittivity based on the double Debye dielectric parameters

The initial assessment of the depth of a burn injury during triage forms the basis for determination of the course of the clinical treatment plan. However, severe skin burns are highly dynamic and hard to predict. This results in a low accuracy rate of about 60 - 75% in the diagnosis of partial-thickness burns in the acute post-burn period. Terahertz time-domain spectroscopy (THz-TDS) has demonstrated a significant potential for non-invasive and timely estimation of the burn severity. Here, we describe a methodology for the measurement and numerical modeling of the dielectric permittivity of the in vivo porcine skin burns. We use the double Debye dielectric relaxation theory to model the permittivity of the burned tissue. We further investigate the origins of dielectric contrast between the burns of various severity, as determined histologically based on the percentage of the burned dermis, using the empirical Debye parameters. We demonstrate that the five parameters of the double Debye model can form an artificial neural network classification algorithm capable of automatic diagnosis of the severity of the burn injuries, and predicting its ultimate wound healing outcome by forecasting its re-epithelialization status in 28 days. Our results demonstrate that the Debye dielectric parameters provide a physics-based approach for the extraction of the biomedical diagnostic markers from the broadband THz pulses. This method can significantly boost dimensionality reduction of THz training data in artificial intelligence models and streamline machine learning algorithms.

Accurate and early prediction of the wound healing outcome of burn injuries using the wavelet Shannon entropy of terahertz time-domain waveforms

Significance

Severe burn injuries cause significant hypermetabolic alterations that are highly dynamic, hard to predict, and require acute and critical care. The clinical assessments of the severity of burn injuries are highly subjective and have consistently been reported to be inaccurate. Therefore, the utilization of other imaging modalities is crucial to reaching an objective and accurate burn assessment modality.

Aim

We describe a non-invasive technique using terahertz time-domain spectroscopy (THz-TDS) and the wavelet packet Shannon entropy to automatically estimate the burn depth and predict the wound healing outcome of thermal burn injuries.

Approach

We created 40 burn injuries of different severity grades in two porcine models using scald and contact methods of infliction. We used our THz portable handheld spectral reflection (PHASR) scanner to obtain the in vivo THz-TDS images. We used the energy to Shannon entropy ratio of the wavelet packet coefficients of the THz-TDS waveforms on day 0 to create supervised support vector machine (SVM) classification models. Histological assessments of the burn biopsies serve as the ground truth.

Results

We achieved an accuracy rate of 94.7% in predicting the wound healing outcome, as determined by histological measurement of the re-epithelialization rate on day 28 post-burn induction, using the THz-TDS measurements obtained on day 0. Furthermore, we report the accuracy rates of 89%, 87.1%, and 87.6% in automatic diagnosis of the superficial partial-thickness, deep partial-thickness, and full-thickness burns, respectively, using a multiclass SVM model.

Conclusions

The THz PHASR scanner promises a robust, high-speed, and accurate diagnostic modality to improve the clinical triage of burns and their management.

In Vivo Assessment and Monitoring of Burn Wounds Using a Handheld Terahertz Hyperspectral Scanner

The accuracy of clinical assessment techniques in diagnosing partial-thickness burn injuries has remained as low as 50-76%. Depending on the burn depth and environmental factors in the wound, such as reactive oxygen species, inflammation, and autophagy, partial-thickness burns can heal spontaneously or require surgical intervention. Herein, it is demonstrated that terahertz time-domain spectral imaging (THz-TDSI) is a promising tool for in vivo quantitative assessment and monitoring of partial-thickness burn injuries in large animals. We used a novel handheld THz-TDSI scanner to characterize burn injuries in a porcine scald model with histopathological controls. Statistical analysis (n= 40) indicates that the THz-TDSI modality can accurately differentiate between partial-thickness and full-thickness burn injuries (1-way ANOVA, p< 0.05). THz-TDSI has the potential to improve burn care outcomes by helping surgeons in making objective decisions for early excision of the wound.

Deep neural network classification of in vivo burn injuries with different etiologies using terahertz time-domain spectral imaging

Thermal injuries can occur due to direct exposure to hot objects or liquids, flames, electricity, solar energy and several other sources. If the resulting injury is a deep partial thickness burn, the accuracy of a physician's clinical assessment is as low as 50-76% in determining the healing outcome. In this study, we show that the Terahertz Portable Handheld Spectral Reflection (THz-PHASR) Scanner combined with a deep neural network classification algorithm can accurately differentiate between partial-, deep partial-, and full-thickness burns 1-hour post injury, regardless of the etiology, scanner geometry, or THz spectroscopy sampling method (ROC-AUC = 91%, 88%, and 86%, respectively). The neural network diagnostic method simplifies the classification process by directly using the pre-processed THz spectra and removing the need for any hyperspectral feature extraction. Our results show that deep learning methods based on THz time-domain spectroscopy (THz-TDS) measurements can be used to guide clinical treatment plans based on objective and accurate classification of burn injuries.

Diffuse terahertz spectroscopy in turbid media using a wavelet-based bimodality spectral analysis

Current terahertz (THz) spectroscopy techniques only use the coherent light beam for spectral imaging. In the presence of electromagnetic scattering, however, the scattering-mitigated incoherent beams allow for flexible emitter-detector geometries, which enable applications such as seeing through turbid media. Despite this potential, THz spectroscopy using diffuse waves has not been demonstrated. The main obstacles are the very poor signal to noise ratios of the diffused fields and the resonance-like spectral artifacts due to multiple Mie scattering events that obscure the material absorption signatures. In this work, we demonstrate diffuse THz spectroscopy of a heterogeneous sample through turbid media using a novel technique based on the wavelet multiresolution analysis and the bimodality coefficient spectrum, which we define here for the first time using the skewness and kurtosis of the spectral images. The proposed method yields broadband and simultaneous material characterization at detection angles as high as 90° with respect to the incident beam. We determined the accuracy of the wavelet-based diffuse spectroscopy at oblique detection angles, by evaluating the area under the receiver operating characteristic curves, to be higher than 95%. This technique is agnostic to any a priori information on the spectral signatures of the sample materials or the characteristics of the scattering medium, and can be expanded for other broadband spectroscopic modalities.

Differentiation of burn wounds in an in vivo porcine model using terahertz spectroscopy

The accuracy of current burn triage techniques has remained between 50-70%. Accordingly, there is a significant clinical need for the quantitative and accurate assessment of partial-thickness burn injuries. Porcine skin represents the closest animal model to human skin, and is often used in surgical skin grafting procedures. In this study, we used a standardized in vivo porcine burn model to obtain terahertz (THz) point-spectroscopy measurements from burns with various severities. We then extracted two reflection hyperspectral parameters, namely spectral area under the curve between approximately 0.1 and 0.9 THz (-10 dB bandwidth in each spectrum), and spectral slope, to characterize each burn. Using a linear combination of these two parameters, we accurately classified deep partial- and superficial partial-thickness burns (p = 0.0159), compared to vimentin immunohistochemistry as the gold standard for burn depth determination.

Comparison of comparable scald and contact burns in a porcine model: A preliminary report

Whether the depth and healing of scalds and contact burns are similar is controversial. Due to water's greater heat capacity, we hypothesized that when exposed to similar temperatures and durations of exposure, burns caused by hot water would be deeper than those caused by contact with hot metal. Forty standardized burns were created in two anesthetized female domestic pigs using a brass bar or circulating heated water. In one pig, the temperature was kept constant (95°C) while the duration of exposure varied (5, 10, 15 seconds) In the second pig, the exposure time was kept constant (10 seconds) while the temperature of exposure varied (70°C, 80°C, 98°C). Periodic punch biopsies were taken to determine burn depth immediately after injury, percentage burns reepithelialized within 21 days, and depth of scar at 28 days. The analysis was performed using analysis of variance. When the temperature was held constant, duration of exposure (5, 10, and 15 seconds) was associated with scar depth (2.1 vs 3.8 vs 5.0 mm, respectively, P = 0.001) but not with burn depth (2.0 vs 2.2 vs 2.3 mm, respectively, P = 0.10). When exposure duration was held constant, temperature (70°C, 80°C, 98°C) was associated with scar depth (0.6 vs 1.7 vs 3.6, P < 0.001) but not with burn depth (1.2 vs 1.5 vs 1.7 mm, respectively, P = 0.21). Burn depths were greater for scald than contact burns although not significantly greater. After controlling for temperature, the difference in scar depth between scalds and contact burns was statistically significant (marginal means 3.0 for contact burns, 4.3 for scalds, P = 0.008). We conclude that burns created in swine with circulating hot water result in deeper scars than those created by contact with a brass bar when controlling for temperature and duration of exposure.

Investigation of water diffusion dynamics in corneal phantoms using terahertz time-domain spectroscopy

Perturbation of normal corneal water content is a common manifestation of many eye diseases. Terahertz (THz) imaging has the potential to serve as a clinical tool for screening and diagnosing such corneal diseases. In this study, we first investigate the diffusive properties of a corneal phantom using simultaneous THz time-domain spectroscopy (THz-TDS) and gravimetric measurements. We will then utilize a variable-thickness diffusion model combined with a stratified composite-media model to simulate changes in thickness, hydration profile, and the THz-TDS signal as a function of time. The simulated THz-TDS signals show very good agreement with the reflection measurements. Results show that the THz-TDS technique can be used to understand water diffusion dynamics in corneal phantoms as a step towards future in vivo quantitative hydration sensing.

Mitigating the effects of granular scattering using cepstrum analysis in terahertz time-domain spectral imaging

Terahertz (THz) imaging is a widely used technique in the study and detection of many chemicals and biomolecules in polycrystalline form because the spectral absorption signatures of these target materials often lie in the THz frequencies. When the size of dielectric grain boundaries are comparable to the THz wavelengths, spectral features can be obscured due to electromagnetic scattering. In this study, we first investigate this granular scattering effect in identification of chemicals with THz spectral absorption features. We then will propose a signal processing technique in the so-called "quefrency" domain to improve the ability to resolve these spectral features in the diffuse scattered THz images. We created a pellet with α-lactose monohydrate and riboflavin, two biologically significant materials with well-known vibrational spectral resonances, and buried the pellet in a highly scattering medium. THz transmission measurements were taken at all angles covering the half focal plane. We show that, while spectral features of lactose and riboflavin cannot be distinguished in the scattered image, application of cepstrum filtering can mitigate these scattering effects. By employing our quefrency-domain signal processing technique, we were able to unambiguously detect the dielectric resonance of lactose in the diffused scattering geometries. Finally we will discuss the limitation of the new proposed technique in spectral identification of chemicals.