1 Transillumination Imaging for Blood Oxygen Saturation Estimation of Skin Lesions Brian D’Alessandro, Atam P. Dhawan, Fellow, IEEE Abstract—Detecting the early stages of melanoma can be greatly assisted by an accurate estimate of subsurface blood volume and blood oxygen saturation, indicative of angiogenesis. Visualization of this blood volume present beneath a skin lesion can be achieved through transillumination of the skin. As the absorption of major chromophores in the skin is wavelength dependent, multispectral imaging can provide the needed infor- mation to separate out relative amounts of each chromophore. However, a critical challenge to this strategy is relating the pixel intensities observed in a given image to the wavelength dependent total absorption existing at each spatial location. Consequently, in this paper we develop an extension to Beer’s law, estimated through a novel voxel-based, parallel processing Monte Carlo simulation of light propagation in skin which takes into account the specific geometry of our transillumination imaging apparatus. We then use this relation in a linear mixing model, solved using a multispectral image set, for chromophore separation and oxygen saturation estimation of an absorbing object located at a given depth within the medium. Validation is performed through Monte Carlo simulation, as well as by imaging on a skin phantom. Results show that subsurface oxygen saturation can be reasonably estimated with good implications for the reconstruction of 3D skin lesion volumes using transillumination towards early detection of malignancy. Index Terms—transillumination, multispectral imaging, melanoma, skin lesions, skin cancer I. I NTRODUCTION S KIN cancer is the most common form of malignancy, but the survival rate is very high if the cancer is detected, diagnosed, and treated early. To assist in this early detection, a hand held imaging device known as the Nevoscope [1] utilizes the principles of optical imaging and transillumination to better observe the subsurface structure of a skin lesion, such as the depth of the lesion and the blood volume surrounding the lesion. This device directs light into the skin at a 45 degree angle through a fiber optic ring placed directly against the skin surface. Light diffuses through the skin tissue, and photons which scatter back through the lesion up to the surface are captured by a charge-coupled device (CCD) image sensor attached to the Nevoscope (see Fig. 1). Since surface illumination is also blocked, the geometry of this illumination device essentially creates a virtual light source behind the skin lesion, thus providing a transilluminated image which contains important subsurface features for the early detection of skin cancer. By filtering the light source, a multispectral image set of the lesion can also be obtained, providing additional information which can be exploited for analysis [2]. Brian D’Alessandro (email: bmd5@njit.edu) and Atam P. Dhawan (email: dhawan@njit.edu) are with the Department of Electrical and Computer Engineering, New Jersey Institute of Technology, Newark, NJ 07102 USA. Skin CCD Fiber Optic Ring Light Fig. 1. Nevoscope Geometry One challenge associated with this method of imaging skin lesions is accurately measuring the concentration of subsurface skin features from the 2D images. The pixel intensity seen in a certain pixel location in an image depends on the concentration, mixture, and depth of the main subsurface chro- mophores in skin. Concentration mixture is represented by the overall absorption coefficient, from which the concentration of individual chromophores can be separated by multispectral imaging [3]. The absorption of main chromophores in the skin (oxyhemoglobin (HbO 2 ), deoxyhemoglobin (Hb), and melanin) varies depending on wavelength (see Fig. 2). Thus, by comparing the localized absorption difference between two or more wavelengths, it is possible to obtain a measure of how much spectral distortion is caused by blood and melanin absorption. Of course, because of the diffusive nature of light in tissue, the actual absorption coefficient at specific locations within the volume is difficult to obtain. A precise quantitative result is further confounded by scattering and the photon path of Nevoscope illumination. Hence, as a step toward estimating the spatially dependent total absorption for multispectral chromophore separation, in this paper we present an extension to Beer’s law for light transmission through an object in a medium, which is tailored to the geometry of the Nevoscope device and is valid for objects located at a given depth within the medium (a preliminary version of which has appeared in [4]). The extension is developed by use of a novel voxel-based, parallel processing Monte Carlo simulation of light propagation from the Nevoscope ring through the skin. First, we discuss the development and improvements related to our simulation. Then, we use the adjusted Beer’s law equation, along with the multispectral imaging set and a linear mixing model, to estimate the relative concentration of oxyhemoglobin and deoxyhemoglobin in an object embedded in the skin, first using simulated images and later using real images with the Nevoscope and a skin phantom. II. MONTE CARLO SIMULATION Monte Carlo simulation is a statistical technique for simu- lating random processes, and has been applied to light-tissue