A geometric-sensitivity-difference method improves object
depth-localization for continuous-wave fluorescence diffuse optical tomography:
An in silico study in an axial outward-imaging geometry
Affiliations: School of Electrical and Computer Engineering,
Oklahoma State University, Stillwater, OK, USA | Department of Radiology, Medical School, University of
Michigan, Ann Arbor, MI, USA
Note: [] Corresponding author: Daqing Piao, School of Electrical and
Computer Engineering, Oklahoma State University, Stillwater, OK 74078, USA.
E-mail: daqing.piao@okstate.edu
Abstract: A geometric-sensitivity-difference (GSD) based reconstruction method
is demonstrated in fluorescence diffuse optical tomography (FDOT) for improving
the depth-localization of objects. The GSD method optimizes the data-model fit
based on paired-measurements between source-detector pairs sharing either the
source or the detector channel, as comparing to conventional methods that
optimize the data-model fit based on un-paired measurements of individual
source-detector pairs. This in silico study is limited to continuous-wave and
2-dimension, for a circular-array outward-imaging geometry of which the native
sensitivity of measurement varies strongly with respect to the depth of the
object. The outcomes of GSD method are compared to that of two conventional
methods: one is the baseline method which does not involve any scheme to
compensate the variation of native sensitivity; the other applies a
depth-adapted weight to counteract the depth-variance of the native
sensitivity. These three methods were evaluated using synthetic data
corresponding to the following conditions of the object: (1) Single object with
a 3-folds of positive contrast of fluorescence over the background was set at
edge-depths of 0.5 mm, 5 mm and 10 mm; (2) Two objects with identical 3-folds
of positive or 1/3-folds of negative contrast of fluorescence over the
background were set at a fixed edge-depth of 10 mm and different azimuthal
separations of 45 degree, 135 degree, and 180 degree; (3) Two objects with
identical 3-folds of positive or 1/3-folds of negative contrast of fluorescence
over the background were set at a fixed azimuthal separation of 90° and at
edge-depths of 0.5 mm, 5 mm and 10 mm. The GSD method outperforms the other two
methods in localizing a single anomaly and resolving two anomalies, for the
anomaly possessing either the 3 folds positive or 1/3-folds negative contrast
of fluorescence over the background. The case of objects with negative contrast
over the background has specific implications to imaging zinc-specific
fluorophore uptake in prostate.
