In this study, we present a new algorithm based on an artificial neural network (ANN) for reducing
speckle noise from optical coherence tomography (OCT) images. The noise is modeled for different parts
of the image using Rayleigh distribution with a noise parameter, sigma, estimated by the ANN. This is
then used along with a numerical method to solve the inverse Rayleigh function to reduce the noise in the
image. The algorithm is tested successfully on OCT images of retina, demonstrating a significant increase
in the signal-to-noise ratio (SNR) and the contrast of the processed images.
Basal cell carcinoma (BCC) is the most common form of skin cancer. To improve the diagnostic accuracy,
additional non-invasive methods of making a preliminary diagnosis have been sought. We have implemented an
En-Face optical coherence tomography (OCT) for this study in which the dynamic focus was integrated into it.
With the dynamic focus scheme, the coherence gate moves synchronously with the peak of confocal gate
determined by the confocal interface optics. The transversal resolution is then conserved throughout the depth
range and an enhanced signal is returned from all depths. The Basal Cell Carcinoma specimens were obtained
from the eyelid a patient. The specimens under went analysis by DF-OCT imaging. We searched for remarkable
features that were visualized by OCT and compared these findings with features presented in the histology
slices.
In this paper, we describe the procedures to make epoxy resin and agarose phantoms designed using Mie scattering
calculations. The phantoms are constructed to be used in the estimation of point spread function (PSF) of an optical
coherence tomography (OCT) and evaluation of optical properties extraction (OPE) algorithm.
The imperfection of optical devices in an optical imaging system deteriorates wavefront which results in aberration. This reduces the optical signal to noise ratio of the imaging system and the quality of the produced images. Adaptive optics composed of wavefront sensor (WFS) and deformable mirror (DM) is a straightforward solution for this problem. The need for a WFS in an AO system, raises the cost of the overall system, and there are also instances when they cannot be used, such as in microscopy. Moreover stray reflections from lens surfaces affect the performance of the WFS. In this paper, we describe a blind optimization technique with an in-expensive electronics without using the WFS to correct the aberration in order to achieve better quality images. The correction system includes an electromagnetic DM from Imagine, Mirao52d, with 52 actuators which are controlled by particle swarm optimization (PSO) algorithm. The results of the application of simulated annealing (SA), and genetic algorithm (GA) techniques that we have implemented in the sensor-less AO are used for comparison.
KEYWORDS: Detection and tracking algorithms, Stochastic processes, Diffusion, Algorithm development, Data modeling, Signal to noise ratio, Diffusion tensor imaging, Brain, Model-based design, Data acquisition
Fibre tractography using diffusion tensor imaging allows the study of anatomical connectivity of the brain, and is an
important diagnostic tool for a range of neurological diseases. Deterministic tractography algorithms assume that the
fibre direction coincides with the principal eigenvector of a diffusion tensor. This is, however, not the case for regions
with crossing fibres. In addition noise introduces uncertainty and makes the computation of fibre directions difficult.
Stochastic tractography algorithms have been developed to overcome the uncertainties of deterministic algorithms.
However, generally, both parametric and non-parametric stochastic algorithms require longer computational time and
large amounts of memory. Multi-tensor fibre tracking methods can alleviate the problems when crossing fibres are
encountered. In this study simple and computationally efficient random-walk algorithms are described for estimating
anatomical connectivity in white matter. These algorithms are then applied to a two-tensor model to compute the
probabilities of connections between regions with complex fibre configurations. We analyze the random-walk models
quantitatively using simulated data and estimate the optimal parameter values of the models. The performance of the
tracking algorithms is verified using a physical phantom and an in vivo dataset with a wide variety of seed points. The
results confirm the effectiveness of the proposed approach, which gives comparable results to other stochastic methods.
Our approach is however significantly faster and requires less memory. The results of two-tensor random-walk
algorithms demonstrate that our algorithms can accurately identify fibre bundles in complex fibre regions.
The images obtained from confocal imaging systems present less resolution than the theoretical limit due to imperfection
of the optical components and their arrangement. This imperfection deteriorates the wavefront and introduces aberrations
to the optical system. Adaptive optics (AO) systems composed of a wavefront sensor (WFS) and a deformable mirror
represent the most used solution to this problem. Such adaptive optics systems are expensive. In addition, in microscopy,
WFSs cannot be used due to stray reflections in the system and high aberrations introduced by the specimen. For these
reasons, sensor-less AO systems have been developed to control the deformable mirror (DM) using an optimization
algorithm in an iterative manner. At each iteration, the algorithm produces a new set of voltage and sends it to the mirror
so as to optimize its shape, in such a way, as to maximize the strength of the photodetector current in the imaging
system. In this paper the results of the application of three optimization techniques in the sensor-less AO are compared.
The three optimization techniques are simulated annealing (SA), genetic algorithm (GA) and particle swarm
optimization (PSO). SA and GA have been previously implemented and PSO is explained in this paper.
This paper presents a neural network based technique to denoise speckled images in optical coherence tomography (OCT). Speckle noise is modeled as Rayleigh distribution, and the neural network estimates the noise parameter, sigma. Twenty features from each image are used as input for training the neural network, and the sigma value is the single output of the network. The certainty of the trained network was more than 91 percent. The promising image results were assessed with three No-Reference metrics, with the Signal-to-Noise ratio of the denoised image being considerably increased.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.