P. Aparna

Recent Advancements in Retinal Vessel Segmentation

Retinal vessel segmentation is a key step towards the accurate visualization, diagnosis, early treatment and surgery planning of ocular diseases. For the last two decades, a tremendous amount of research has been dedicated in developing automated methods for segmentation of blood vessels from retinal fundus images. Despite the fact, segmentation of retinal vessels still remains a challenging task due to the presence of abnormalities, varying size and shape of the vessels, non-uniform illumination and anatomical variability between subjects. In this paper, we carry out a systematic review of the most recent advancements in retinal vessel segmentation methods published in last five years. The objectives of this study are as follows: first, we discuss the most crucial preprocessing steps that are involved in accurate segmentation of vessels. Second, we review most recent state-of-the-art retinal vessel segmentation techniques which are classified into different categories based on their main principle. Third, we quantitatively analyse these methods in terms of its sensitivity, specificity, accuracy, area under the curve and discuss newly introduced performance metrics in current literature. Fourth, we discuss the advantages and limitations of the existing segmentation techniques. Finally, we provide an insight into active problems and possible future directions towards building successful computer-aided diagnostic system.

A nearest neighbor based approach for classifying epileptiform EEG using nonlinear DWT features

Epilepsy is a pathological condition characterized by spontaneous, unforeseeable occurrence of seizures, during which the perception or behaviour of a person is altered, if not disturbed. In prediction of occurance of seizures, better classification accuracies have been reported with the use of non linear features and hence they have been estimated from wavelet transformed Electro Encephalo Graph (EEG) data and used to train k Nearest Neighbour (kNN) classifier to classify the EEG into normal, background and epileptic classes. Very good accuracy performance of nearly 100% has been reported from the current work.

An efficient algorithm for textural feature extraction and detection of tumors for a class of brain MR imaging applications

In this paper, we propose an efficient textural feature extraction algorithm (TFEA) based on higher order statistical cumulant namely Kurtosis for a class of brain MR imaging applications. Using a model that represents the wavelet coefficient energies of the sub-bands of multi-level decomposition of the image as a basis, a feature set involving three parameters for each band corresponding to probability density function (PDF) of generalized Gaussian type is derived. The logical correctness and working of the proposed TFEA are first verified based on MATLAB ver.2010a tool. The algorithm is applied in conjunction with one of the popularly used canny edge detection algorithm for segmenting a class of real and synthetic magnetic resonance (MR) images to detect the region of a tumor if present. The use of the proposed approach results in reduced feature set size thus obviating the need for employing specialized feature selection/reduction algorithms. A detailed look at the experimental results clearly show an improvement in the segmentation quality compared with conventional method.

Low Complexity Distributed Video Coding with Golay Codes

In this paper we present a low complexity video coder working on the principle of distributed source coding that combines the principle of channel coding with source coding. In this work the encoder complexity is shifted to the decoder to support uplink friendly video applications, simultaneously achieving the rate-distortion performance of the conventional predictive coding system. In this work concept of syndrome coding with Golay codes is adopted for compression. The simulation results presented in this paper reveals the superior performance of this distributed video coder over the Intraframe coders in terms of rate distortion performance, simultaneously achieving low complexity when compared to predictive coders.

Distributed Video Coding Using LDPC Codes for Wireless Video

Popular video coding standards like H.264 and MPEG working on the principle of motion-compensated pre-dictive coding demand much of the computational resources at the encoder increasing its complexity. Such bulky encoders are not suitable for applications like wireless low power surveillance, multimedia sensor networks, wireless PC cameras, mobile camera phones etc. New video coding scheme based on the principle of distributed source coding is looked upon in this paper. This scheme supports a low complexity encoder, at the same time trying to achieve the rate distortion performance of conventional video codecs. Current im-plementation uses LDPC codes for syndrome coding.

Syndrome coding of video with LDPC codes

In this paper we present the simulation results of the video coding system based on the principle of distributed source coding. Unlike conventional video coding system, this system exploits source statistics at the decoder, thus reversing the complexity model. Current implementation uses LDPC codes for syndrome coding.

Adaptive Local Cosine transform for Seismic Image Compression

A typical seismic analysis involves collection of data by an array of seismometers, transmission over a narrow-band channel, and storage of data for analysis. Transmission and archiving of large volumes of data involves great cost. Hence there is a need to devise suitable methods for compressing the seismic data without compromising on the quality of the reconstructed signal. This paper presents our work on the seismic data compression based on adaptive local cosine transform and its associated multi-resolution and best-basis methodology and compares the results with wavelet based implementation.

A feasible QRS detection algorithm for arrhythmia diagnosis

This paper presents a reliable QRS detection algorithm to detect and classify Electrocardiogram (ECG) waveform abnormalities by extracting features such as heart rate and duration of QRS complex. As R peak detection is the pivotal step in automatic electrocardiogram analysis, various mathematical operations like clipping, differentiation and squaring are carried out in the preprocessing stage to enhance the section containing the QRS complex. Thresholding is performed to detect the R peaks. In order to improve the accuracy of the algorithm search back technique is implemented to determine the missing R peaks and heart beat. Once the position of R peak is detected, positions of Q and S are determined using two different search intervals to take into account the anomalous conditions as well. The heart rate and QRS width are then computed and compared with the normal values to determine the degree and type of abnormality. MIT-BIH database is used to evaluate this algorithm. The algorithm gives sensitivity of 99.34% and positive predictivity of 96.79%. In the proposed algorithm complicated mathematical operations like Fourier Transform, Hilbert Transform or crosscorrelation are not computed, hence is convenient to realize.

A comprehensive solution to road traffic accident detection and ambulance management

Delay in providing Emergency Medical Services (EMS) is the cause of the high mortality rate in road traffic accidents in countries like India. There is delay involved in each and every stage of the process, right from reporting an accident to dispatching an ambulance, till the patient is safely handed over to the casualty. Minimizing this delay can help save lives. We propose a comprehensive solution to both accident detection and ambulance management. When the in-vehicle accident detection module reports an accident, the main server automatically dispatches the nearest ambulance to the accident spot. The android application used by the ambulance driver assists the driver to reach the location quickly and safely. Automation of accident detection and ambulance dispatch, along with providing guidance to the ambulance driver, is achieved here. This can save precious time and help standardize the whole process.

Symmetry based perceptually lossless compression of 3D medical images in spatial domain

A perceptual model developed in spatial domain based on background luminance is combined with lossless compression frame work to remove visually redundant information. Block matching is applied on anatomical symmetry present in medical images to remove intra slice and inter slice correlations. The obtained results show better compression gains against lossless compression techniques, without any degradation in visual quality.