Ryali Srikanth

Contextual encoding in uniform and adaptive mesh-based lossless compression of MR images

We propose and evaluate a number of novel improvements to the mesh-based coding scheme for 3-D brain magnetic resonance images. This includes: 1) elimination of the clinically irrelevant background leading to meshing of only the brain part of the image; 2) content-based (adaptive) mesh generation using spatial edges and optical flow between two consecutive slices; 3) a simple solution for the aperture problem at the edges, where an accurate estimation of motion vectors is not possible; and 4) context-based entropy coding of the residues after motion compensation using affine transformations. We address only lossless coding of the images, and compare the performance of uniform and adaptive mesh-based schemes. The bit rates achieved (about 2 bits per voxel) by these schemes are comparable to those of the state-of-the-art three-dimensional (3-D) wavelet-based schemes. The mesh-based schemes have been shown to be effective for the compression of 3-D brain computed tomography data also. Adaptive mesh-based schemes perform marginally better than the uniform mesh-based methods, at the expense of increased complexity.

WAVELET-BASED ESTIMATION OF HEMODYNAMIC RESPONSE FUNCTION FROM fMRI DATA

We present a new algorithm to estimate hemodynamic response function (HRF) and drift components of fMRI data in wavelet domain. The HRF is modeled by both parametric and nonparametric models. The functional Magnetic resonance Image (fMRI) noise is modeled as a fractional brownian motion (fBm). The HRF parameters are estimated in wavelet domain by exploiting the property that wavelet transforms with a sufficient number of vanishing moments decorrelates a fBm process. Using this property, the noise covariance matrix in wavelet domain can be assumed to be diagonal whose entries are estimated using the sample variance estimator at each scale. We study the influence of the sampling rate of fMRI time series and shape assumption of HRF on the estimation performance. Results are presented by adding synthetic HRFs on simulated and null fMRI data. We also compare these methods with an existing method,(1) where correlated fMRI noise is modeled by a second order polynomial functions.

MR image coding using content-based mesh and context

Existing schemes for 3-D magnetic resonance (MR) images, such as block matching method and uniform mesh-based scheme, are inadequate to model the motion field of MR sequence because deformation within a mesh element may not all be similar. We propose a scheme consisting of (a) content-based mesh generation using optic flow between two consecutive images (b) forward motion tracking (c) motion compensation using affine transformation and (d) context-based modeling. We also propose a simple scheme to overcome aperture problem at edges where an accurate estimation of motion vectors is not possible. By using context-based modeling, motion compensation yields a better estimate of the next frame and hence a lower entropy of the residue. The obtained average compression ratio of 4.3 is better than the values of 4, achieved by CALIC, and 3, by the existing uniform mesh-based interframe coding scheme.

Shape adaptive integer wavelet transform based coding scheme for 2-D/3-D MR images

This paper proposes a new shape adaptive integer wavelet transform (SAIWT) based progressive transmission coding scheme for 2-D and 3-D brain MRI. This scheme consists of: (a) extraction of shape information (mask) by morphological operations; (b) 2D/3D separable biorthogonal 9-7 SAIWT; (c) intraband wavelet encoding of the foreground object. This object based coding greatly improves both overall progressive transmission performance and lossless compression; (d) entropy coding of resulting bit stream and differential coding of the boundary of binary mask. The performance of shape adaptive scheme (both 2D and 3D) is superior to the conventional rectangular based schemes at all bit rates.

A new coding scheme for 2-D and 3-D MR images using shape adaptive integer wavelet transform

We propose a new shape adaptive integer wavelet transform based progressive transmission coding scheme for 2-D and 3-D MRI. The scheme consists of (a) extraction of shape information (b) shape adaptive integer wavelet transform (c) intraband wavelet encoding and (d) entropy coding. The proposed scheme results in improved performance of progressive transmission as compared to the conventional rectangular wavelet transform coding schemes. The main contribution of this paper is its unique approach which rejects the noisy background rather than considering it for lossy compression as proposed in most of the recent schemes. This rejection of unwanted information results in considerable decrease in bit rate. Another contribution is object based 3-D coding which is out of scope of baseline JPEG2000.

Subspace and hypothesis based effective segmentation of co-articulated basic-units for concatenative speech synthesis

In this paper, we present two new methods for vowel-consonant segmentation of a co-articulated basic-units employed in our Thirukkural Tamil text-to-speech synthesis system (G. L. Jayavardhana Rama et al, IEEE workshop on Speech Synthesis, 2002). The basic-units considered in this are CV, VC, VCV, VCCV and VCCC, where C stands for a consonant and V for any vowel. In the first method, we use a subspace-based approach for vowel-consonant segmentation. It uses oriented principal component analysis (OPCA) where the test feature vectors are projected on to the V and C subspaces. The crossover of the norm-contours obtained by projecting the test basic-unit onto the V and C subspaces give the segmentation points which in turn helps in identifying the V and C durations of a test basic-unit. In the second method, we use probabilistic principal component analysis (PPCA) to get probability models for V and C. We then use the Neymen-Pearson (NP) test to segment the basic-unit into V and C. Finally, we show that the hypothesis testing turns out to be an energy detector for V-C segmentation which is similar to the first method.

Estimation theoretic framework for comparing polarization based, continuous-wave direct imaging schemes

We report a maiden study and comparison of two important classes of polarization based continuous-wave optical imaging schemes for imaging through scattering media, namely, the polarization difference imaging (PDI) and the polarization modulation imaging (PMI). We cast the problem in an estimation theoretic framework and base the comparison on two visualization parameters, the polarization magnitude and the degree of polarization. We show that PDI is superior in estimating these two parameters in active imaging. However, we show that PMI is suitable for passive imaging and that the PDI is a specific implementation of PMI.

Polarization-rich continuous wave direct imaging:modeling and visualization

We report a study and comparison of continuous-wave, optical polarization difference imaging (PDI) and polarization modulation imaging (PMI) for imaging through scattering media. The problem is cast in the framework of a theoretical estimation, and the comparison is based on three visualization parameters, namely, the magnitude, the degree, and the orientation of the polarization. We show that PDI is superior in estimating the first two parameters in active imaging under specific conditions, while the PMI is suitable for passive imaging and is the only way to estimate polarization orientation. We also propose new schemes for rendering polarization information as a color image and for applying the newly introduced polarization-orientation imaging for segmentation. Simulation and experimental results verify the theoretical conclusions.

Wavelet-Based Estimation of Hemodynamic Response Function

We present a new algorithm to estimate hemodynamic response function (HRF) and drift component in wavelet domain. The HRF is modeled as a gaussian function with unknown parameters. The functional Magnetic resonance Image (fMRI) noise is modeled as a fractional brownian motion (fBm). The HRF parameters are estimated in wavelet domain since wavelet transform with sufficient number of vanishing moments decorrelates a fBm process. Due to this decorrelating property of wavelet transform, the noise covariance matrix in wavelet domain can be assumed to be diagonal whose entries are estimated using sample variance estimators at each scale. We study the influence of sampling time and shape assumption on the estimation performance. Results are presented by adding synthetic HRFs on null fMRI data.

Signal subspace based enhancement and MAP parameter estimation of fMRI signals

We propose a signal subspace based functional magnetic resonance image (fMRI) signal enhancement followed by maximum a posteriori (MAP) estimation of the parameters of a hemodynamic response function (HRF). The fMRI time-series, which is corrupted by physiological and scanner noise, is a low SNR signal. This signal is projected onto signal-plus-noise space and then enhanced in this space. The enhanced signal is then used to estimate the parameters of the HRF using MAP estimation. Preliminary results indicate that signal enhancement greatly improves the estimation performance.