Review and Evaluation of Medical Image Segmentation Using Methods of Optimal Filtering

Abstract

Image segmentation is an important process of image analysis. It consists of subdividing an image into its constituent parts and extracting these interesting parts. A large variety of segmentation algorithms have been developed. The evaluation and comparison of these algorithms turns out to be important and even indispensable for correctly using them. This paper presents an objective study of segmentation algorithms using the methods of optimal filtering. This study is distinguished from other studies by its consideration of both evaluation and comparison, treating both image cases (noisy and non-noisy ones). The results of implementation, an evaluation of the advantages and the drawbacks of each one of them, and a study of their immunity towards three types of noise are also presented. All these characteristics make this study a general and effective one for revealing the performance of segmentation algorithms using the methods of optimal filtering.

Hippocampal volumetry for lateralization of temporal lobe epilepsy: Automated versus manual methods

“…There are several review articles on automatic segmentation techniques for a variety of brain structures (Angelini et al, 2005; Cremers et al, 2007; Rouainia and Doghmane, 2008; Suri et al, 2002; Trichili et al, 2003). Generally, in many of the segmentation methods, there is an energy function, a shape model and an optimization algorithm.…”

Bayesian image segmentation using local iso-intensity structural orientation

“…4 In all the experiments on the performance comparison, the following configurations have been used. In OT, a 5 5 5 filter window with relative bandwidth equals 2 and center frequency equals has been used (relative bandwidth and center frequency control the characteristics of the quadrature filters, see [32] for further details).…”

“…To extract high-level information from an image, digital image analysis systems, such as industrial inspection systems, autonomous object recognition systems, and medical image analysis systems, may need to accomplish image segmentation prior to any post-processing algorithms [1]. As such, a tremendous amount of thorough research has taken place on image segmentation [2]- [4].…”

“…Edge pixel is denoted by intensity discontinuity in the image. It can be detected by a differential operator such as Sobel, Roberts, Prewitt, and Laplacian operators [2], [4]. An edge linking algorithm has been employed to assemble broken edge pixels into interconnected boundaries, and the image is partitioned into disjointed regions (i.e., segmented) according to these boundaries [1].…”

“…This type of object in the real-world image is usually assumed to be piecewise constant with random white Gaussian noise contamination [2], [4], [22]. In other words, in the noiseless image (i.e., the truth image), pixels that belong to the same object should have the same intensity value.…”