Overview and Prospectus
The principal focus of The Skin Appearance Laboratory is to study the ways in which the interaction of light and tissue can be used to assess the health of the skin and the presence of disease. In particular, it is of interest to examine how these interactions can be captured with images to support dermatologic diagnosis. Research interests cover the investigation and modeling of the underlying anatomical, physiological and optical processes that transform incident into reflected light, the range of techniques and devices used to capture the characteristics of healthy and diseased skin, and methods to improve the distribution of dermatologic care through imaging standards development and telemedicine.
Capturing an accurate representation of skin is deceptively difficult. Skin is a living organ that distends and deforms as it envelops an articulated frame. What must be captured is only a few hundred micrometers thick, but that thin barrier provides a surface that scatters, absorbs and reflects light with a complex matrix of structures, many as small as the wavelengths of the incident light. It is also a dynamic system whose appearance varies under tension and pressure in a manner at times both manifest and subtle. Tension can alter the structure of the skin’s optical interface across scale, while pressure can change the spectrum of the backscatter by redistributing the blood within the vascular system. Although the skin is renewed through cell turnover every few weeks, it retains the consequences of damage and disease for a much longer duration. These, and a host of other complications, ensure that capturing and displaying images of a region of skin over time for comparison, or even once with sufficient diagnostic accuracy, will require many things done right.
How do you determine what it takes to do skin imaging right? Images of the skin, the anatomy of the skin, simulations of the skin – these complementary measurements and representations can be brought together to form a mutually constraining triad. The constraints are built up from reflectance data that are obtained from surface imaging in all its variations using polarization, narrowband multispectral, IR, and UV filtering, as well as from dermoscopy; from anatomical data using histological sections to produce serial reconstructions that help define 3D skin models, and other views of tissue such as with in vivo confocal scanning microscopy; and from simulations that are generated using models of light transport in tissue models and rendering algorithms. When these measures and representations are brought into agreement, they can be used to contribute to the definition of the diagnostic imaging chain. These values, obtained through experimentation and analysis, can be used to define diagnostic imaging done right.
Using these results, GUIs can be constructed that correct photographic mistakes, optimize image quality, and tease out what might not otherwise be apparent under normal visual examination. Advances in computational photography can provide functionality that goes beyond the simple documentation of backscattered light. For example, while imaging and histology can be combined initially to constrain and to validate simulations, the goal ultimately is to produce simulations by taking the imaging data and models of light transport through tissue and to use them to construct a representation of the underlying skin tissue. It is not the case that the dendritic arborization of every melanocyte can be reconstructed from such noninvasive imaging. Nonetheless, the potential exists for spatial models of tissue obtained in this manner to have significantly greater diagnostic value than that provided by traditional surface images alone. Consequently, it will require advances in tool design similar to those proposed here to provide the needed support to midlevel clinical staff in extending access to specialty services such as dermatology. Applied in face-to-face examinations or through the use of telemedicine, a single dermatologist can also use these tools to supervise a much more effective and efficient workflow and workforce.
Finally, in this town where George Eastman did so much for imaging, it would be a dereliction not to extend the research in this laboratory to the study of film-based photography. It is of particular interest to study photographic portraiture, a genre so often brought to life by the quality of the representation of skin. This quality is due in no small part to the aesthetic appeal designed into any of the classic or modern film processes. It is derived from the different film’s distinctive fusion of the capture of fine detail, the representation of variations in tonal range and distribution, and the coloration of image structure through the addition of characteristic noise patterns. It is of interest to study the manner in which the properties of these media interact with the features of the skin as the latent image is altered by the chemistry of film and paper. It is also timely to capture the rich depth and dimension of these photographic processes now before they disappear entirely into museums and over a century and a half of imaging lore is lost to the digital age.
Brian C. Madden