Image Segmentation of Brain MRI Based on LTriDP and Superpixels of Improved SLIC

Abstract

Non-uniform gray distribution and blurred edges often result in bias during the superpixel segmentation of medical images of magnetic resonance imaging (MRI). To this end, we propose a novel superpixel segmentation algorithm by integrating texture features and improved simple linear iterative clustering (SLIC). First, a 3D histogram reconstruction model is used to reconstruct the input image, which is further enhanced by gamma transformation. Next, the local tri-directional pattern descriptor is used to extract texture features of the image; this is followed by an improved SLIC superpixel segmentation. Finally, a novel clustering-center updating rule is proposed, using pixels with gray difference with original clustering centers smaller than a predefined threshold. The experiments on the Whole Brain Atlas (WBA) image database showed that, compared to existing state-of-the-art methods, our superpixel segmentation algorithm generated significantly more uniform superpixels, and demonstrated the performance accuracy of the superpixel segmentation in both fuzzy boundaries and fuzzy regions.

Automatic image segmentation based on label propagation

“…Non-coloured medical images such as ultrasounds [14] and X-rays [15,16] may also be a target. The proposed methods could also be used for the segmentation of 3D magnetic resonance medical images [17]. They can be applied in pre-segmentation tasks for semantic segmentation of natural scenes [18,19].…”

“…FIGURE17 Qualitative results. From left to right: ground truth, LPCI-SP, LV-SP, FG-SUTP and ILP-GAU-CM (our)…”

Superpixel Segmentation Based on Grid Point Density Peak Clustering

“…Superpixel segmentation is often used as an important preprocessing method for image algorithms in different research fields, such as the image segmentation [ 1 , 2 ] and object recognition [ 3 , 4 ]. The superpixel segmentation is used to improve the diagnostic accuracy of medical images and the fruit and road recognition in agricultural automatic systems.…”

“…According to the superpixel boundary and shape, there are two main segmentation results as Figure 1 shows. (1) The superpixel shapes are more regular and evenly distributed in Figure 1a-d [11], but the superpixel boundaries cannot describe the texture and contour features well. (2) The boundary adherence in Figure 1e,f [9,13] are generally better than Figure 1a-d, but the boundaries and shapes are complex and very irregular.…”

Latest Advances in Human Brain Dynamics

“…A methodological study by Wang et al [ 11 ] proposed a novel superpixel segmentation algorithm by integrating texture features that further improved simple linear iterative clustering (SLIC). Segmentation of MRI images is of paramount importance, and it is an open task that deserves further exploration.…”