| 520 | |
▼a Burns are some of the most common injuries in both civilian and combat scenarios. Prompt and proper treatments largely depend on an accurate diagnosis of burn depth. Various techniques have been developed to facilitate this process. Experimentally, mostly imaging-based techniques are developed for such purpose, which focus on observing the changes in tissue morphology due to burns, and not on the associated mechanical characteristics. Numerically, bioheat transfer-based models in conjunction with Arrhenius type equations have been proposed to predict thermal damage to soft tissue. However, such models only account for the thermal physics and provide a heuristic understanding of tissue damage based on temperature profiles. A major knowledge gap exists in quantitative characterization and understanding of changes in intrinsic mechanical properties, as a result of the burning process.In this thesis, we propose a combined experimental and computational study that provides insight into the changes in mechanical properties of skin tissue due to thermal injury. Specifically, four burn conditions have been created: (i) 200쨘F for 10s, (ii) 200쨘F for 30s, (iii) 450쨘F for 10s, and (iv) 450쨘F for 30s. An ultrasound elastography-based technique is applied to characterize the mechanical properties of the unburnt and burnt tissues. By iteratively solving an inverse optimization problem that minimizes the difference between experimentally measured and simulated force-displacement data, the two parameters (C10 and C20) of a reduced second-order polynomial hyperelastic material model are constructed. The results indicate that while the C10 parameter does not show a statistically significant difference between the test conditions, the C20 parameter reliably identifies three (ii-iv) of the four cases (p<0.05) when comparing burnt with unburnt tissues with a classification accuracy of 60-87%. In addition, softening of the tissue is observed, which can be attributed to the changes in the structures of the collagen fibers.In order to elucidate the effects of various burn conditions on the altered mechanical characteristics of ex vivo porcine skin tissues, uniaxial tensile tests are performed. Twelve burn conditions have been created by contact heating of skin tissues at 200쨘F and 450쨘F for 10s, 15s, 20s, 30s, 40s, and 50s. Mechanical properties, such as ultimate tensile (UT) stress, ultimate tensile stretch, toughness, and parameters of a Veronda-Westmann hyperelastic material model are obtained from the experimental stress-strain curves. While the reduced second-order polynomial model adequately describes the stress-strain constitutive response within 20% strain, the Veronda-Westmann model provides a better fit for the skin tissues stretched over 100% strain. It is observed that higher temperature (450쨘F) burns can lead to an overall decreasing trend in UT stress, toughness, and hyperelastic material parameters, as well as an increase in UT strain. The lower temperature (200 쨘F) burns, however, result in non-monotonic changes in UT stress, toughness and hyperelastic material parameters, which can be an indication of meta-stable state in collagen fibrils. Statistical analysis is employed to study the altered mechanical properties of skin tissues subjected to burn conditions that induce varying degrees of burns. Pair-wise binary classifications of burn groups are performed using kernel support vector machine (KSVM) with mechanical properties as the input feature set. A leave-one-out cross-validation (LOOCV) is used for the independent assessment of the classifiers. The average accuracy, sensitivity, and specificity for pairwise separation of burn groups are 95.65%, 96.06%, and 94.88%. The results highlight that the set of mechanical properties including ultimate strength and toughness can distinguish between different burn severities, which is otherwise not possible with just the hyperelastic material properties.To further understand the changes in mechanical properties of the burnt tissue, Raman spectroscopy is used to examine the changes in collagen structures. Raman spectra of wavenumbers 500-2000 cm-1 are collected from samples of the four burn conditions (i) - (iv) and the unburnt condition. The overall spectra reveal that protein and amino acids-related bands manifest structural changes including the destruction of protein-related functional groups and changes from 慣-like helical to disordered structures that are correlated with increasing burn severity. The deconvolution of the amide I region (1580-1720 cm-1) and analysis of the sub-bands reveal a change of secondary structure of collagens from 慣-like helix dominated to 棺-aggregate dominated ones.Finally, the Raman spectra are used to classify different burn severities with two machine learning technique: KSVM in conjunction with principal component analysis (PCA), and partial least-square (PLS) method. An independent assessment of the classifiers using LOOCV approach yields an average classification accuracy, sensitivity, and specificity of ~92-93% for both techniques. The variable importance in the projection (VIP) scores computed from the PLS model reveal that bands corresponding to proteins and lipids, amide III, and amino acids are important indicators in separating unburnt or mild burns (200쨘F), while amide I has more pronounced impact in separating severe burns (450쨘F).In sum, this thesis provides insights into the mechanics and morphology of tissues subjected to thermal injury by (1) characterization of changes in mechanical properties of burnt tissues |