A reduced-space basis function neural network method for diffuse optical tomography

Hyun Keol Kim, Jacqueline Gunther, Jennifer Hoi, Andreas H. Hielscher

Research output: Chapter in Book/Report/Conference proceedingConference contribution


We propose here a reduced space image reconstruction method that makes use of basis function neural network (BFNN) within a framework of PDE-constrained algorithm. This method reduces the solution space using the basis function approach, and finds the optimal solution through the learning process of neural network. The basis function approach improves the ill-posed nature of an original inverse problem, reducing the number of unknowns as well as regularizing the solution automatically. The proposed method was applied to breast cancer imaging, and the reconstruction performance was evaluated on how well the method can identify the tumor location in breast tissue. The results show that the BFNN method gives better results in the identification of tumor location than the traditional element-based reconstruction method.

Original languageEnglish (US)
Title of host publicationOptical Tomography and Spectroscopy of Tissue XI
EditorsRobert R. Alfano, Eva M. Sevick-Muraca, Bruce J. Tromberg, Arjun G. Yodh
ISBN (Electronic)9781628414097
StatePublished - 2015
EventOptical Tomography and Spectroscopy of Tissue XI - San Francisco, United States
Duration: Feb 9 2015Feb 11 2015

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
ISSN (Print)1605-7422


ConferenceOptical Tomography and Spectroscopy of Tissue XI
Country/TerritoryUnited States
CitySan Francisco


  • DCT
  • PDE-constrained optimization
  • basis function
  • diffuse optical tomography
  • neural network

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Atomic and Molecular Physics, and Optics
  • Radiology Nuclear Medicine and imaging
  • Biomaterials


Dive into the research topics of 'A reduced-space basis function neural network method for diffuse optical tomography'. Together they form a unique fingerprint.

Cite this