Kashif Rajpoot

Local-phase based 3D boundary detection using monogenic signal and its application to real-time 3-D echocardiography images

Ultrasound images are characterized by their speckle appearance, low contrast, and poor signal-to-noise ratio. It is always challenging to extract important clinical information from these images. An important step before automatic measurement is to transform the image into significant features of interest. Intensity based methods do not perform particularly well on ultrasound images. However, it has been shown previously that ultrasound images respond well to local phase-based methods which are theoretically intensityinvariant and thus suitable for low-contrast nature of ultrasound images. We extend the local phase-based method of feature asymmetry measure computation to detect 3D features using the monogenic signal, which is an isotropic extension of the analytic signal to higher dimensional functions. The proposed method is applied to real-time 3D echocardiography images and the visual results for the endocardial and epicardial boundary detection are presented.

The evaluation of single-view and multi-view fusion 3D echocardiography using image-driven segmentation and tracking

Real-time 3D echocardiography (RT3DE) promises a more objective and complete cardiac functional analysis by dynamic 3D image acquisition. Despite several efforts towards automation of left ventricle (LV) segmentation and tracking, these remain challenging research problems due to the poor-quality nature of acquired images usually containing missing anatomical information, speckle noise, and limited field-of-view (FOV). Recently, multi-view fusion 3D echocardiography has been introduced as acquiring multiple conventional single-view RT3DE images with small probe movements and fusing them together after alignment. This concept of multi-view fusion helps to improve image quality and anatomical information and extends the FOV. We now take this work further by comparing single-view and multi-view fused images in a systematic study. In order to better illustrate the differences, this work evaluates image quality and information content of single-view and multi-view fused images using image-driven LV endocardial segmentation and tracking. The image-driven methods were utilized to fully exploit image quality and anatomical information present in the image, thus purposely not including any high-level constraints like prior shape or motion knowledge in the analysis approaches. Experiments show that multi-view fused images are better suited for LV segmentation and tracking, while relatively more failures and errors were observed on single-view images.

Multiview RT3D Echocardiography Image Fusion

Real-time three-dimensional echocardiography (RT3DE) permits the acquisition and visualization of the beating heart in 3D. However, its actual utility is limited due to missing anatomical structures and limited field-of-view (FOV). We present an automatic two-stage registration and fusion method to integrate multiple single-view RT3DE images. The registration scheme finds a rigid transformation by using a multiresolution algorithm. The fusion is based on the 3D wavelet transform, utilizing the separation of the image into low- and high-frequency wavelet subbands. The qualitative and quantitative results, from 12 subjects, demonstrate that the proposed fusion framework helps in: (i) filling-in missing anatomical information, (ii) extending the FOV, and (iii) increasing the structural information and image contrast.

SVM optimization for hyperspectral colon tissue cell classification

The classification of normal and malginant colon tissue cells is crucial to the diagnosis of colon cancer in humans. Given the right set of feature vectors, Support Vector Machines (SVMs) have been shown to perform reasonably well for the classification [4,13]. In this paper, we address the following question: how does the choice of a kernel function and its parameters affect the SVM classification performance in such a system? We show that the Gaussian kernel function combined with an optimal choice of parameters can produce high classification accuracy.

Wavelets and support vector machines for texture classification

We present a novel texture classification algorithm using 2-D discrete wavelet transform (DWT) and support vector machines (SVM). The DWT is used to generate feature images from individual wavelet subbands, and a local energy function is computed corresponding to each pixel of the feature images. This feature vector is first used for training and later on for testing the SVM classifier. The experimental setup consists of images from the Brodatz and MlT VisTeX texture databases and a combination of some images therein. The proposed method produces promising classification results for both single and multiple class texture analysis problems.

Multiview Fusion 3-d Echocardiography: Improving the Information and Quality of Real-Time 3-D Echocardiography

The advent of real-time 3-D echocardiography (RT3DE) promised dynamic 3-D image acquisition with the potential of a more objective and complete functional analysis. In spite of that, 2-D echocardiography remains the backbone of echocardiography imaging in current clinical practice, with RT3DE mainly used for clinical research. The uptake of RT3DE has been slow because of missing anatomic information, limited field-of-view (FOV) and tedious analysis procedures. This paper presents multiview fusion 3D echocardiography, where multiple images with complementary information are acquired from different probe positions. These multiple images are subsequently aligned and fused together for preserving salient structures in a single, multiview fused image. A novel and simple wavelet-based fusion algorithm is proposed that exploits the low- and high-frequency separation capability of the wavelet analysis. The results obtained show that the proposed multiview fusion considerably improves the contrast (31.1%), contrast-to-noise ratio (46.7%), signal-to-noise ratio (44.7%) and anatomic features (12%) in 3-D echocardiography, and enlarges the FOV (28.2%). This indicates that multiview fusion substantially enhances the image quality and information.

Real-time 3D fusion echocardiography.

OBJECTIVES This study assessed 3-dimensional fusion echocardiography (3DFE), combining several real-time 3-dimensional echocardiography (RT3DE) full volumes from different transducer positions, for improvement in quality and completeness of the reconstructed image. BACKGROUND The RT3DE technique has limited image quality and completeness of datasets even with current matrix transducers. METHODS RT3DE datasets were acquired in 32 participants (mean age 33.7 +/- 18.8 years; 27 men, 5 women). The 3DFE technique was also performed on a cardiac phantom. The endocardial border definition of RT3DE and 3DFE segments was graded for quality: good (2), intermediate (1), poor (0), or out of sector. Short-axis and apical images were compared in RT3DE and 3DFE, yielding 2,048 segments. The images were processed to generate 3DFE and then compared with cardiac magnetic resonance. RESULTS In the heart phantom, fused datasets showed improved contrast to noise ratio from 49.4 +/- 25.1 (single dataset) to 125.4 +/- 25.1 (6 datasets fused together). In subjects, more segments were graded as good quality with 3DFE (805 vs. 435; p < 0.0001) and fewer as intermediate (184 vs. 283; p = 0.017), poor (31 vs. 265; p < 0.0001), or out of sector (4 vs. 41; p < 0.001) compared with the single 3-dimensional dataset. End-diastolic volume (EDV) and end-systolic volume (ESV) obtained from 3-dimensional fused datasets were equivalent to those from single datasets (EDV 118.2 +/- 39 ml vs. 119.7 +/- 43 ml; p = 0.41; ESV 48.1 +/- 30 ml vs. 48.4 +/- 35 ml; p = 0.87; ejection fraction [EF] 61.0 +/- 10% vs. 61.8 +/- 10%; p = 0.44). Bland-Altman analysis showed good 95% limits of agreement for the nonfused datasets (EDV +/-46 ml; ESV +/-36 ml; EF +/-14%) and the fused datasets (EDV +/-45 ml; ESV +/-35 ml; EF +/-16%), when compared with cardiac magnetic resonance. CONCLUSIONS Fusion of full-volume datasets resulted in an improvement in endocardial borders, image quality, and completeness of the datasets.

Investigation into the Fusion of Multiple 4-D Fetal Echocardiography Images to Improve Image Quality

Recent advances in four-dimensional (4-D) ultrasound enable the acquisition and visualisation of the entire fetal heart. However, getting consistent, shadow free, data remains problematic due to the uncontrollable nature of fetal orientation. This article presents the first investigation into the utility of image fusion to improve the quality of volumetric fetal cardiac imaging. Multiple volume scans are registered using a semiautomatic approach and five fusion methods are assessed for their ability to remove artefacts and improve image quality. Image quality is assessed in terms of signal-to-noise ratio, contrast and contrast-to-noise ratio. Qualitative results are presented for the ability to remove artefacts. The fusion methods assessed were found to be divided into those that reduce noise and those that increase contrast. The effect of fusion on left ventricle segmentation using commercial state-of-the-art software is also considered. The use of image fusion is shown to reduce the variability of volume estimates by about 50% relative to measurement on a single scan.

Wavelet based segmentation of hyperspectral colon tissue imagery

Segmentation is an early stage for the automated classification of tissue cells between normal and malignant types. We present an algorithm for unsupervised segmentation of images of hyperspectral human colon tissue cells into their constituent parts by exploiting the spatial relationship between these constituent parts. This is done by employing a modification of the conventional wavelet based texture analysis, on the projection of hyperspectral image data in the first principal component direction. Results show that our algorithm is comparable to other more computationally intensive methods which exploit the spectral characteristics of the hyperspectral imagery data

Texture Segmentation using Ant Tree Clustering

Motivated by the self-assembling behavior of real ants, we present a novel algorithm for texture segmentation which is based on ant tree clustering of wavelet features. In a pattern recognition setting, wavelet features are extracted using either of the two subband filtering methods: discrete wavelet transform (DWT) or discrete wavelet packet transform (DWPT). The feature classification process is inspired by the self-assembling behavior observed in real ants where ants progressively become attached to an existing support and then successively to other attached ants thus building trees based on the similarity of feature vectors. The results thus obtained compare favorably to those of other recently published filtering based texture segmentation algorithms