30-03-2012, 12:29 PM
A VLSI-inspired Image Reconstruction Algorithm for
Continuous-wave Diffuse Optical Tomography Systems
[attachment=19247]
INTRODUCTION
Using diffuse optical tomography (DOT) to image inside
the human body in vivo has developed rapidly in the past
decades. DOT measures the change of absorption coefficient
which is primarily influenced by hemoglobin in wavelength
at near-infrared region (700 − 900nm) [3].
By using the different types of measurements of diffused
light to reconstruct image, DOT can be generally divided into
three main categories: the Continuous Wave (CW),
Frequency Domain and Time Domain. Table I from [1] shows
the characteristics of different DOT systems.
The CW system provides advantages such as low cost, high
portability, low power consumption and computation
overhead [2]. Therefore, there exists possibility of
implementing hardware DOT platforms of CW systems. To
progress toward this goal, this paper proposes a practical
image reconstruction algorithm. To reconstruct images from
sensor measurements, the solution of a linear inverse problem
is required [3][10].
IMAGE RECONSTRUCTION OF DOT
The forward model and inverse resolution are two critical
processes to reconstruct image. Forward model is to describe
how the photons are scattered and diffused in highly
scattering medium. Inverse problem is to resolve optical
characteristics of the medium.
CONCLUSION
In this paper, we apply the truncated Jacobi SVD
algorithm to do the inverse resolution in CW-DOT systems
and propose a sub-frame mode in order to reduce the
computational cost. We simulate inhomogeneous media with
different shapes and locations and study the impact of
different reconstruction modes on the quality of image. This
study demonstrates that low computational cost is possible
without harming severely the image quality. In short, TJSVD
is a highly efficient technique for reconstruction of good
quality images and is suitable for CW-DOT system.
Continuous-wave Diffuse Optical Tomography Systems
[attachment=19247]
INTRODUCTION
Using diffuse optical tomography (DOT) to image inside
the human body in vivo has developed rapidly in the past
decades. DOT measures the change of absorption coefficient
which is primarily influenced by hemoglobin in wavelength
at near-infrared region (700 − 900nm) [3].
By using the different types of measurements of diffused
light to reconstruct image, DOT can be generally divided into
three main categories: the Continuous Wave (CW),
Frequency Domain and Time Domain. Table I from [1] shows
the characteristics of different DOT systems.
The CW system provides advantages such as low cost, high
portability, low power consumption and computation
overhead [2]. Therefore, there exists possibility of
implementing hardware DOT platforms of CW systems. To
progress toward this goal, this paper proposes a practical
image reconstruction algorithm. To reconstruct images from
sensor measurements, the solution of a linear inverse problem
is required [3][10].
IMAGE RECONSTRUCTION OF DOT
The forward model and inverse resolution are two critical
processes to reconstruct image. Forward model is to describe
how the photons are scattered and diffused in highly
scattering medium. Inverse problem is to resolve optical
characteristics of the medium.
CONCLUSION
In this paper, we apply the truncated Jacobi SVD
algorithm to do the inverse resolution in CW-DOT systems
and propose a sub-frame mode in order to reduce the
computational cost. We simulate inhomogeneous media with
different shapes and locations and study the impact of
different reconstruction modes on the quality of image. This
study demonstrates that low computational cost is possible
without harming severely the image quality. In short, TJSVD
is a highly efficient technique for reconstruction of good
quality images and is suitable for CW-DOT system.