Unsupervised Deep Learning Approach For Photoacoustic Spectral Unmixing.
Open Access
Author:
Durairaj, Deepit Abhishek
Graduate Program:
Electrical Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
November 07, 2019
Committee Members:
Sri-Rajasekhar Kothapalli, Thesis Advisor/Co-Advisor David Jonathan Miller, Thesis Advisor/Co-Advisor William Evan Higgins, Committee Member Sri-Rajasekhar Kothapalli, Program Head/Chair
Accurate spectral unmixing is required for revealing functional and molecular information of the tissue using multispectral photoacoustic imaging data. A significant challenge in deep-tissue photoacoustic imaging is the nonlinear dependence of the received photoacoustic signals on the local molecular distribution. To overcome this, we have deployed an end-to-end unsupervised neural network based on autoencoders. The proposed method employs the physical properties as the constraints to the neural network which effectively performs the unmixing and outputs the molecular concentration maps without a-priori knowledge of absorption spectra. The algorithm is tested on a set of simulated mixtures and experimental data. The proposed approach shows promising results in spectral unmixing given its completely unsupervised nature.
