In vivo automated quantification of thermally damaged human tissue using polarization sensitive optical coherence tomography

Abstract

Over the decades numerous technologies have been performed for the quantification of skin injuries, but their poor sensitivity, specificity and accuracy limits their applications. Optical coherence tomography (OCT) can be potential tool for the identification but the modern high-speed OCT system acquires huge amount of data, which will be very time-consuming and tedious process for human interpretation. Our proposed method opens the possibility of fully automated quantitative analysis based on morphological features of thermally damaged tissue, which will become biomarker for the removal of non-viable skin. The proposed method is based on multi-level ensemble classifier by dissociating morphological features (A-line, B-scan, phase images) extracted from Polarization Sensitive Optical Coherence Tomography (PS-OCT) images. Our proposed classifier attained the average sensitivity, specificity and accuracy is 92.22%, 87.2% and 92.5%, respectively, in detecting the thermally damaged human skin. Moreover, we show that our classifier is one of the best possible classifier based on features extracted from PS-OCT images, which demonstrates the significance of PS-OCT data in detecting abnormality in human skin.

Introduction

Thermal damage in human tissue is defined as an injury to the tissues caused by a net flux of pathologic energy (Rab et al., 2005). Regardless of the source, whether it is chemical, electrical or radiation, thermal damage leads to a common pathway, which is the disruption of skin integrity (Rab et al., 2005). Thermal injury is one of the most painful injuries ever experienced by human being. When it occurs to any part of the body, it damages the nerve endings causing intense pain sensations. In normal case, it causes external damage on the body surface. Severe injury may become the cause of certain physical disabilities because the damaged skin has decreased ability to fight against infection. Therefore, the skin is no longer able to function as a protective barrier to the environment (Cole et al., 1990; Liddington and Shakespeare, 1996). Presently, the accuracy of detecting the burn/thermal injuries of skin is exclusively depends upon the experience of the doctor or clinician, which is maximum upto 80% (Li et al., 2015). Due to thermal injuries, the internal property of the skin changes, hence the proper removal of the non-viable skin is a crucial task for the doctors. Under-debridement increases the risk of infection while over-debridement reduces the skin’s regenerative capacity hence scarring effect (Liddington and Shakespeare, 1996). Thus, evaluation of thermal damaged tissue and wound depth remains an important clinical goal in the accomplishment of intensely injured patients. Although, the clinical assessment of damaged tissue by specialists is reported to be only 80% and is limited due to the lack of an accurate, quantitative technique for measuring tissue stiffness and function (Brown et al., 1998). Thus, there is an urgent need of automated device that can quantitatively discriminate the normal and thermal damaged human skin. Such system would be of a great importance for taking decisions about further medical intervention and for doing proper diagnosis. Some attempts were made to introduce different modalities for the assessment of thermal damage, e.g. R. Cole (Eisenbeiß et al., 1999) Liddington (Afromowitz et al., 1988), from which modality like thermography benefits from its relative technical ease, but was limited by its long-time of investigation and diagnostic procedure, which is qualitative (Still et al., 2001). In the clinical practice of-depth assessment modalities were currently used, in which laser direct imaging (LDI) and indocyanine green (ICG) video angiography estimates best accuracy (Boushel et al., 2000). ICG depicts a dynamic behaviour, which fluctuates in real time (O’reilly et al., 1989) where as LDI provides a static perfusion map with validity as high as 99% (Anselmo and Zawacki, 1977). So, it was considered to be the best technique for assessing the thermal damage at the critical and most difficult to diagnosis level. But its major limitation is that their accuracy is contingent on optics and can adversely affect depth measurements (Gambichler et al., 2011).

Now days, thermal damage assessment depends upon the non-invasive techniques that can be employed for rapid diagnostic studies with minimum oppression to patient. Penetration depth and resolution generally having an inverted relationship in non-invasive imaging modalities (Crisan et al., 2013). In comparison with other imaging modalities, Optical Coherence Tomography (OCT) is a non-contact, non-invasive, high-resolution imaging technique with the depth of penetration ∼1 mm, and resolution ∼4–10 μm (Han et al., 2011). Thus, for skin imaging OCT provides peerless balance between penetration depth and resolution. It offers many other advantages in research and clinical practice (Mogensen et al., 2009; Pierce et al., 2004a), which provide valuable information to physicians in real-time.

The two main reasons that have limited the scope of visually determining the changes in structural features, associated with tissue information are first, it is not possible, to evaluate visually the huge amount of data acquired from a high-speed OCT system, and second, it is hard to state some properties of tissue texture, like coarseness, non-uniformity by direct visualization. Significant improvement to OCT technology have been realized, including improvement in imaging depths which extends beyond 2 mm, resolution, and considerably reduced imaging time (Gladkova et al., 2000). When birefringent tissues are of interest, infusion of polarization sensitivity (Trasischker et al., 2014; Everett et al., 1998; Park et al., 2001), is significant, that able to detect and quantify the polarization state of light. It is ascertained that the birefringence of the damaged tissue changes due to the change in the alignment of epidermis and dermal layer and can be detected and quantified using Polarisation Sensitive OCT (PS-OCT). In an in vivo animal model, PS-OCT demonstrates a relationship between birefringence and thermal damage tissue (Pierce et al., 2004b; Iftimia et al., 2013). The two-dimensional images acquired in OCT are called B-scans (X–Z), which is further combined to form a 3-D OCT volume. Thus, it is essential for quantitative diagnosis to evolve an automated image processing method to characterize tissue on the basis of PS-OCT images.

To acquire intensity and birefringence based images of normal and thermal damaged tissue in vivo, we used PS-OCT. We then proposed an algorithm for extracting A-line features, B-scan textural features from amplitude images and birefringence based features from phase images acquired from PS-OCT system of normal and thermally damage tissue to quantify structural changes. Finally, a set of most relevant OCT features of normal and thermal damaged tissue were identified. Results suggest that PS-OCT can be considered as a strong aspirant for robust and automated diagnosis of thermal damaged tissue.