It is that specific need that Skin-DOCTor proposes to cover by developing new theoretical approaches implementing diffusive light propagation and building an optical prototype system delivering three-dimensionally accurate images of pigmented skin lesions, advancing Optical Computed Tomography (Figure 2) beyond the current state of the art. This will enable the extraction of accurate, quantitative results at depths of several mm, exceeding the limit of currently available technologies. The specific target of this technological development is the realization of a clinical system for the in situ evaluation and classification of pigmented skin lesions biopsy samples. Characterization and discrimination of benign from malignant melanocytic lesions will be achieved with higher sensitivity and specificity than current technologies, matching the diagnostic value of histopathological examination. In addition, accurate calculation of the depth of the lesion will be achieved and advanced expert learning algorithms will be employed to accomplish optimal diagnostic and prognostic capabilities based on the ultimate concepts of heterogeneous data integration. Considering the wide array of multimodal and multiscale biomedical data separately available for disease characterization (from 1D to 3D), Skin-DOCTor proposes the integration of heterogeneous biomedical data in order to construct an accurate diagnostic support tool. This system will be functional within the operating room and will provide immediate results to the surgeon.
Clinical studies report that the diagnosis by simple inspection has a sensitivity of 65%-80%, depending on the dermatologist’s experience and training [Kittler et. al., Lancet Oncol., 3, 159-165, (2002)]. Considering that the clinical progression of patients suffering from skin melanoma is directly associated with the early diagnosis of the tumor, the establishment of an objective, accurate, quantitative method is of high priority. The necessity for improvement in diagnostic accuracy of melanocytic lesions led to the development of imaging methods such as in vivo confocal laser scanning microscopy, second harmonic generation imaging, spectroscopy and optical coherence tomography, none of which, so far, have gained wide acceptance in clinical practice as a diagnostic tool. Optical Coherence Tomography has been suggested as the optical analogous of histology and while it provides very high resolution images it suffers (similarly to the other techniques) from light diffusion inside tissue and reaches, at best, depths of 1mm. So far the most advanced approaches to optical imaging of tissue function and structure are separated depending on the amount of scattering present in the media. At the two ends of the scattering regime lie Optical Projection Tomography (OPT) [Sharpe, Ahlgren, et al., Science 5567, 541-544, (2002)], which is capable of imaging non scattering samples, such as optically cleared animal embryos and pupae, and Diffusion Optical Tomography (DOT), [Arridge, Inverse Problems, 15, R41-R93, (1999)], a technique applied to imaging small animals and humans where light scattering is not disregarded and the propagation of light is considered diffusive. The region lying between these two extremes, defined mesoscopic, remains however without a specific theoretical approach that can accurately describe weakly diffusive light propagation. It is in this region that small animal specimens, such as zebra fish, and small tissue samples, such as biopsies, lie in respect to absorption and scattering of transmitted light. The current Optical Computed Tomography systems represent projections as line integrals generated from multiple sources that emit light along parallel rays, which are transformed according to Beer’s law to calculate the sums of attenuation coefficients (Figure 2). However, when light scattering becomes significant and therefore it cannot be ignored, ballistic propagation required by Beer’s Law fails to deliver accurate images.