A three-step reconstruction method for fluorescence molecular tomography based on compressive sensing

Yansong Zhu, Abhinav Kumar Jha, Jakob K. Dreyer, Hanh N.D. Le, Jin U. Kang, Per E. Roland, Dean Foster Wong, Arman Rahmim

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

Abstract

Fluorescence molecular tomography (FMT) is a promising tool for real time in vivo quantification of neurotransmission (NT) as we pursue in our BRAIN initiative effort. However, the acquired image data are noisy and the reconstruction problem is ill-posed. Further, while spatial sparsity of the NT effects could be exploited, traditional compressive-sensing methods cannot be directly applied as the system matrix in FMT is highly coherent. To overcome these issues, we propose and assess a three-step reconstruction method. First, truncated singular value decomposition is applied on the data to reduce matrix coherence. The resultant image data are input to a homotopy-based reconstruction strategy that exploits sparsity via ℓ1 regularization. The reconstructed image is then input to a maximum-likelihood expectation maximization (MLEM) algorithm that retains the sparseness of the input estimate and improves upon the quantitation by accurate Poisson noise modeling. The proposed reconstruction method was evaluated in a three-dimensional simulated setup with fluorescent sources in a cuboidal scattering medium with optical properties simulating human brain cortex (reduced scattering coefficient: 9.2 cm-1, absorption coefficient: 0.1 cm-1 and tomographic measurements made using pixelated detectors. In different experiments, fluorescent sources of varying size and intensity were simulated. The proposed reconstruction method provided accurate estimates of the fluorescent source intensity, with a 20% lower root mean square error on average compared to the pure-homotopy method for all considered source intensities and sizes. Further, compared with conventional ℓ2 regularized algorithm, overall, the proposed method reconstructed substantially more accurate fluorescence distribution. The proposed method shows considerable promise and will be tested using more realistic simulations and experimental setups.

Original languageEnglish (US)
Title of host publicationOptical Tomography and Spectroscopy of Tissue XII
PublisherSPIE
Volume10059
ISBN (Electronic)9781510605596
DOIs
StatePublished - 2017
EventOptical Tomography and Spectroscopy of Tissue XII - San Francisco, United States
Duration: Jan 30 2017Feb 1 2017

Other

OtherOptical Tomography and Spectroscopy of Tissue XII
CountryUnited States
CitySan Francisco
Period1/30/172/1/17

Fingerprint

Tomography
tomography
Fluorescence
fluorescence
Scattering
Singular value decomposition
Mean square error
Maximum likelihood
Brain
Synaptic Transmission
Optical properties
root-mean-square errors
cortexes
Detectors
scattering coefficients
estimates
matrices
brain
absorptivity
decomposition

Keywords

  • Compressive sensing
  • FMT
  • Noise modeling
  • Reconstruction

ASJC Scopus subject areas

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

Cite this

Zhu, Y., Jha, A. K., Dreyer, J. K., Le, H. N. D., Kang, J. U., Roland, P. E., ... Rahmim, A. (2017). A three-step reconstruction method for fluorescence molecular tomography based on compressive sensing. In Optical Tomography and Spectroscopy of Tissue XII (Vol. 10059). [1005911] SPIE. https://doi.org/10.1117/12.2252664

A three-step reconstruction method for fluorescence molecular tomography based on compressive sensing. / Zhu, Yansong; Jha, Abhinav Kumar; Dreyer, Jakob K.; Le, Hanh N.D.; Kang, Jin U.; Roland, Per E.; Wong, Dean Foster; Rahmim, Arman.

Optical Tomography and Spectroscopy of Tissue XII. Vol. 10059 SPIE, 2017. 1005911.

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

Zhu, Y, Jha, AK, Dreyer, JK, Le, HND, Kang, JU, Roland, PE, Wong, DF & Rahmim, A 2017, A three-step reconstruction method for fluorescence molecular tomography based on compressive sensing. in Optical Tomography and Spectroscopy of Tissue XII. vol. 10059, 1005911, SPIE, Optical Tomography and Spectroscopy of Tissue XII, San Francisco, United States, 1/30/17. https://doi.org/10.1117/12.2252664
Zhu Y, Jha AK, Dreyer JK, Le HND, Kang JU, Roland PE et al. A three-step reconstruction method for fluorescence molecular tomography based on compressive sensing. In Optical Tomography and Spectroscopy of Tissue XII. Vol. 10059. SPIE. 2017. 1005911 https://doi.org/10.1117/12.2252664
Zhu, Yansong ; Jha, Abhinav Kumar ; Dreyer, Jakob K. ; Le, Hanh N.D. ; Kang, Jin U. ; Roland, Per E. ; Wong, Dean Foster ; Rahmim, Arman. / A three-step reconstruction method for fluorescence molecular tomography based on compressive sensing. Optical Tomography and Spectroscopy of Tissue XII. Vol. 10059 SPIE, 2017.
@inproceedings{cb9e103c591c4d97af9cac56e7a3238a,
title = "A three-step reconstruction method for fluorescence molecular tomography based on compressive sensing",
abstract = "Fluorescence molecular tomography (FMT) is a promising tool for real time in vivo quantification of neurotransmission (NT) as we pursue in our BRAIN initiative effort. However, the acquired image data are noisy and the reconstruction problem is ill-posed. Further, while spatial sparsity of the NT effects could be exploited, traditional compressive-sensing methods cannot be directly applied as the system matrix in FMT is highly coherent. To overcome these issues, we propose and assess a three-step reconstruction method. First, truncated singular value decomposition is applied on the data to reduce matrix coherence. The resultant image data are input to a homotopy-based reconstruction strategy that exploits sparsity via ℓ1 regularization. The reconstructed image is then input to a maximum-likelihood expectation maximization (MLEM) algorithm that retains the sparseness of the input estimate and improves upon the quantitation by accurate Poisson noise modeling. The proposed reconstruction method was evaluated in a three-dimensional simulated setup with fluorescent sources in a cuboidal scattering medium with optical properties simulating human brain cortex (reduced scattering coefficient: 9.2 cm-1, absorption coefficient: 0.1 cm-1 and tomographic measurements made using pixelated detectors. In different experiments, fluorescent sources of varying size and intensity were simulated. The proposed reconstruction method provided accurate estimates of the fluorescent source intensity, with a 20{\%} lower root mean square error on average compared to the pure-homotopy method for all considered source intensities and sizes. Further, compared with conventional ℓ2 regularized algorithm, overall, the proposed method reconstructed substantially more accurate fluorescence distribution. The proposed method shows considerable promise and will be tested using more realistic simulations and experimental setups.",
keywords = "Compressive sensing, FMT, Noise modeling, Reconstruction",
author = "Yansong Zhu and Jha, {Abhinav Kumar} and Dreyer, {Jakob K.} and Le, {Hanh N.D.} and Kang, {Jin U.} and Roland, {Per E.} and Wong, {Dean Foster} and Arman Rahmim",
year = "2017",
doi = "10.1117/12.2252664",
language = "English (US)",
volume = "10059",
booktitle = "Optical Tomography and Spectroscopy of Tissue XII",
publisher = "SPIE",

}

TY - GEN

T1 - A three-step reconstruction method for fluorescence molecular tomography based on compressive sensing

AU - Zhu, Yansong

AU - Jha, Abhinav Kumar

AU - Dreyer, Jakob K.

AU - Le, Hanh N.D.

AU - Kang, Jin U.

AU - Roland, Per E.

AU - Wong, Dean Foster

AU - Rahmim, Arman

PY - 2017

Y1 - 2017

N2 - Fluorescence molecular tomography (FMT) is a promising tool for real time in vivo quantification of neurotransmission (NT) as we pursue in our BRAIN initiative effort. However, the acquired image data are noisy and the reconstruction problem is ill-posed. Further, while spatial sparsity of the NT effects could be exploited, traditional compressive-sensing methods cannot be directly applied as the system matrix in FMT is highly coherent. To overcome these issues, we propose and assess a three-step reconstruction method. First, truncated singular value decomposition is applied on the data to reduce matrix coherence. The resultant image data are input to a homotopy-based reconstruction strategy that exploits sparsity via ℓ1 regularization. The reconstructed image is then input to a maximum-likelihood expectation maximization (MLEM) algorithm that retains the sparseness of the input estimate and improves upon the quantitation by accurate Poisson noise modeling. The proposed reconstruction method was evaluated in a three-dimensional simulated setup with fluorescent sources in a cuboidal scattering medium with optical properties simulating human brain cortex (reduced scattering coefficient: 9.2 cm-1, absorption coefficient: 0.1 cm-1 and tomographic measurements made using pixelated detectors. In different experiments, fluorescent sources of varying size and intensity were simulated. The proposed reconstruction method provided accurate estimates of the fluorescent source intensity, with a 20% lower root mean square error on average compared to the pure-homotopy method for all considered source intensities and sizes. Further, compared with conventional ℓ2 regularized algorithm, overall, the proposed method reconstructed substantially more accurate fluorescence distribution. The proposed method shows considerable promise and will be tested using more realistic simulations and experimental setups.

AB - Fluorescence molecular tomography (FMT) is a promising tool for real time in vivo quantification of neurotransmission (NT) as we pursue in our BRAIN initiative effort. However, the acquired image data are noisy and the reconstruction problem is ill-posed. Further, while spatial sparsity of the NT effects could be exploited, traditional compressive-sensing methods cannot be directly applied as the system matrix in FMT is highly coherent. To overcome these issues, we propose and assess a three-step reconstruction method. First, truncated singular value decomposition is applied on the data to reduce matrix coherence. The resultant image data are input to a homotopy-based reconstruction strategy that exploits sparsity via ℓ1 regularization. The reconstructed image is then input to a maximum-likelihood expectation maximization (MLEM) algorithm that retains the sparseness of the input estimate and improves upon the quantitation by accurate Poisson noise modeling. The proposed reconstruction method was evaluated in a three-dimensional simulated setup with fluorescent sources in a cuboidal scattering medium with optical properties simulating human brain cortex (reduced scattering coefficient: 9.2 cm-1, absorption coefficient: 0.1 cm-1 and tomographic measurements made using pixelated detectors. In different experiments, fluorescent sources of varying size and intensity were simulated. The proposed reconstruction method provided accurate estimates of the fluorescent source intensity, with a 20% lower root mean square error on average compared to the pure-homotopy method for all considered source intensities and sizes. Further, compared with conventional ℓ2 regularized algorithm, overall, the proposed method reconstructed substantially more accurate fluorescence distribution. The proposed method shows considerable promise and will be tested using more realistic simulations and experimental setups.

KW - Compressive sensing

KW - FMT

KW - Noise modeling

KW - Reconstruction

UR - http://www.scopus.com/inward/record.url?scp=85019184733&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85019184733&partnerID=8YFLogxK

U2 - 10.1117/12.2252664

DO - 10.1117/12.2252664

M3 - Conference contribution

C2 - 28596634

AN - SCOPUS:85019184733

VL - 10059

BT - Optical Tomography and Spectroscopy of Tissue XII

PB - SPIE

ER -