Translator Disclaimer
6 March 2013 Improved DOT reconstruction by estimating the inclusion location using artificial neural network
Author Affiliations +
Proceedings Volume 8668, Medical Imaging 2013: Physics of Medical Imaging; 86684C (2013)
Event: SPIE Medical Imaging, 2013, Lake Buena Vista (Orlando Area), Florida, United States
Diffuse optical tomography (DOT), a noninvasive imaging modality, uses near infrared light to illuminate the tissue and reconstructs the optical parameters of the tissue from the intensity measurements at the surface. Here continuous wave measurement with improved localization is proposed to make the overall instrument inexpensive. Due to the non-unique solution of the inverse problem, prior information improves the resolution of the reconstructed image. An artificial neural network (ANN) based approach is developed to obtain the location of the inclusion. The peak amplitude, 50% and 10% bandwidth and their corresponding source-detector angles of the difference intensity plot with and without the inclusion are taken as the input. The offset distance between the source and centre of inclusion, the angle with x-axis, sample and inclusion radii are the output of the 2 layered error back propagation neural network. Least square optimization with regularization term is used to minimize the mean squared error for image reconstruction. The optical parameters are updated using the prior information from the ANN. The parameters present in double the region of detected area only are updated. The performance of the proposed method has been assessed quantitatively by computing the mean square error, object centroid error and misclassification ratio. The use of prior improves the convergence and reduces the presence of ghost or noise. Hence the proposed method shows potential to improve DOT reconstruction.
© (2013) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Rusha Patra and Pranab K. Dutta "Improved DOT reconstruction by estimating the inclusion location using artificial neural network", Proc. SPIE 8668, Medical Imaging 2013: Physics of Medical Imaging, 86684C (6 March 2013);

Back to Top