Machine learning has been used to classify melanoma from 2D dermoscopic images, with diagnosis accuracy comparable to that of dermatologists. However, there are few approaches for reconstructing the Breslow depth of the melanoma, which is a crucial metric for skin cancer diagnosis and stage identification. To estimate the depth of tissue lesions, we propose a multiplexed Diffused Optical Imaging Generative Adversarial Network (mDOI-GAN) solution. mDOI-GAN is composed of two main parts: the reconstruction and the projection sections. The reconstruction section estimates the optimal structure of 3D tissue volume depending on the presented 2D reflectance images. The projection section simulates the light-tissue interaction, generating 2D diffusion reflectance images corresponding to the given composition of 3D tissue volume. These two sections work together to form an optimization loop. The network utilizes structurally illuminated 2D reflectance images captured from a low-cost and contact-free Diffuse Optical Imaging acquisition platform as inputs. By iteratively minimizing the distance between experimental and synthetic 2D reflectance images, we will obtain a global optimal 3D reconstruction of tissue volume. We train the network with a customized 3D synthetic dermatology dataset derived from the HAM10000 dataset. Our design has the advantage of introducing the physics modeling of steady-state diffuse optical imaging and 3D spatial probability density function into the deep learning network. Our model is posed to be more flexible, interpretable, and predictable than the current end-to-end, black-box neural network benchmarks.
The availability and the cost of 3D imaging systems are still a problem nowadays, which brings us to the need and urgency of a new way to democratize optical biopsy. Inspired by structural illumination and diffuse optical imaging, we propose a 3D-multiplexed diffused optical imaging (3D-mDOI) solution, a technique to reconstruct the 3D optical properties of the tissue from 2D diffuse images and estimate the depth of tissue lesions. 3D-mDOI uses a low-cost and contact-free design of the imaging acquisition platform, integrating a digital micromirror device (DMD) and an infrared-enhanced CCD camera. The imaging setup that creates custom sampling patterns for tissue photon migration enables spatial multiplexing to overcome low photon signals. We design a hybrid reconstruction pipeline for harvesting the benefits from existing mathematical solutions. The analytical solution of the steady-state radiation transfer equation is utilized to compute each pixel's optical properties in 2D. Monte Carlo simulation provides the stochastic solution for 3D photon diffusion patterns on the discretized tissue volume. We then map the 2D optical properties to the corresponding 3D photon diffusion patterns between a light source and a detector. To better correct the instrumental noises, we design multiple calibrations. 3D-mDOI is versatile, non-invasive, and cost-effective, containing 3D insights to subsurface molecular composition. The technique reconstructs lesions up to 5mm below the surface with 0.2mm axial spatial resolution. We could apply the solution to broad applications in the scientific and medical fields, including the rapid estimation of melanoma staging.
Hyperpsectral fluorescence imaging has been gaining its popularity in life-science field for its simultaneous multiplexing capability of multiple fluorescent labels. Traditional diffraction grating-based hyperspectral acquisition has limited photon-throughput due to the loss at the diffractive optics. The uniform spectral sampling using multiple narrow spectral bands also limits the detectable photons for each channel, which limits the imaging speed as longer exposure is required to achieve sufficient signal to noise ratios. Here we present a Fourier transform-based spectral sampling strategy based on high efficiency dichroic mirrors, enabling video-speed snapshot acquisition with the capability of multiplexing more than five fluorescent signatures.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.