ABSTRACT
Virtual tissues (phantoms) are widely used for performance evaluation of imaging systems. Specific design of the phantom is necessary for the correct assessment of a system's parameters. In an effort to reduce the amount of time and energy spent making application-oriented phantoms, we describe procedures to make epoxy-resin solid phantoms based on Mie scattering theory, with two different scatterers: polystyrene and gold microspheres. The phantoms are specifically designed to be used in two applications: (a) the gold microspheres solid phantoms are used to estimate the point-spread function (PSF) of an optical coherence tomography (OCT) system, and (b) the polystyrene solid phantom are used to evaluate the performance of an OCT-images optical properties extraction (OPE) algorithm. Phantoms with differing combination of materials have been tested to achieve the most suitable combination for producing an accurate PSF for application (a) and a valid evaluation/parameter optimization of the algorithm in application (b). An en face time-domain dynamic focus OCT is used for imaging.
ABSTRACT
Optical coherence tomography (OCT) has the potential for skin tissue characterization due to its high axial and transverse resolution and its acceptable depth penetration. In practice, OCT cannot reach the theoretical resolutions due to imperfections of some of the components used. One way to improve the quality of the images is to estimate the point spread function (PSF) of the OCT system and deconvolve it from the output images. In this paper, we investigate the use of solid phantoms to estimate the PSF of the imaging system. We then utilize iterative Lucy-Richardson deconvolution algorithm to improve the quality of the images. The performance of the proposed algorithm is demonstrated on OCT images acquired from a variety of samples, such as epoxy-resin phantoms, fingertip skin and basaloid larynx and eyelid tissues.
Subject(s)
Algorithms , Image Enhancement/methods , Image Interpretation, Computer-Assisted/methods , Tomography, Optical Coherence/methods , Humans , Reproducibility of Results , Sensitivity and SpecificityABSTRACT
This paper describes a region-growing algorithm for the segmentation of large lesions in T1-weighted magnetic resonance (MR) images of the head. The algorithm involves a gray level similarity criterion to expand the region and a size criterion to prevent from over-growing outside the lesion. The performance of the algorithm is evaluated and validated on a series of pathologic three-dimensional MR images of the head.
Subject(s)
Algorithms , Brain Diseases/pathology , Brain/pathology , Magnetic Resonance Imaging/methods , Fourier Analysis , Image Processing, Computer-Assisted , Imaging, Three-Dimensional , Sensitivity and SpecificityABSTRACT
A new region growing method for finding the boundaries of blobs is presented. A unique feature of the method is that at each step, at most one pixel exhibits the required properties to join the region. The method uses two novel discontinuity measures, average contrast and peripheral contrast, to control the growing process.
