Split-and-merge image segmentation based on localized feature analysis and statistical tests

The space paradox in graphic representation

The negative space drawing technique refers to drawing the transparent space around and between objects, rather than drawing the objects themselves. This space-based instruction is thought to attenuate object-specific visual attention and to enhance perception of a spatial expanse. Developmentally, it is equivalent to the Piagetian dichotomic space concept of filled and empty space. A sample of 96 children from 5 to 12 years of age and 24 adults (N = 120) drew on a computer tablet a real-life model spacebox placed in front of the participant, with three cubes placed inside the model. Children followed two instructions, a Visual Realism (VR) Instruction “Please draw the three cubes and the box as you can see them” and a Negative Space (NSp) Instruction “Please draw the space around the objects,” with the sequence counterbalanced. NSp outline drawings began to show from 9 years onwards. A positive effect of the NSp technique showed for occlusion drawing because of the depiction of common contour of objects which could create a cohesive scene feature such as a horizon. The VR instruction focused attention toward the space box and enhanced 3D drawing of both the spacebox and the cubes. Thus, it could be concluded—rather paradoxically—that drawing in 3D is better based on object- than on space-based attention, while drawing occlusion is better based on space-based than object-based attention. We suggest, however, that a better definition of VR as attention to object appearances is that VR unifies objects and spatial context into one global plane.

EMDS-6: Environmental Microorganism Image Dataset Sixth Version for Image Denoising, Segmentation, Feature Extraction, Classification, and Detection Method Evaluation

Environmental microorganisms (EMs) are ubiquitous around us and have an important impact on the survival and development of human society. However, the high standards and strict requirements for the preparation of environmental microorganism (EM) data have led to the insufficient of existing related datasets, not to mention the datasets with ground truth (GT) images. This problem seriously affects the progress of related experiments. Therefore, This study develops the Environmental Microorganism Dataset Sixth Version (EMDS-6), which contains 21 types of EMs. Each type of EM contains 40 original and 40 GT images, in total 1680 EM images. In this study, in order to test the effectiveness of EMDS-6. We choose the classic algorithms of image processing methods such as image denoising, image segmentation and object detection. The experimental result shows that EMDS-6 can be used to evaluate the performance of image denoising, image segmentation, image feature extraction, image classification, and object detection methods. EMDS-6 is available at the https://figshare.com/articles/dataset/EMDS6/17125025/1.

Decision support system for the detection and grading of hard exudates from color fundus photographs

Diabetic retinopathy is a major cause of blindness, and its earliest signs include damage to the blood vessels and the formation of lesions in the retina. Automated detection and grading of hard exudates from the color fundus image is a critical step in the automated screening system for diabetic retinopathy. We propose novel methods for the detection and grading of hard exudates and the main retinal structures. For exudate detection, a novel approach based on coarse-to-fine strategy and a new image-splitting method are proposed with overall sensitivity of 93.2% and positive predictive value of 83.7% at the pixel level. The average sensitivity of the blood vessel detection is 85%, and the success rate of fovea localization is 100%. For exudate grading, a polar fovea coordinate system is adopted in accordance with medical criteria. Because of its competitive performance and ability to deal efficiently with images of variable quality, the proposed technique offers promising and efficient performance as part of an automated screening system for diabetic retinopathy.

Detection of exudates in retinal images using a pure splitting technique

Diabetic retinopathy is a major cause of blindness. Earliest signs of diabetic retinopathy are damage to blood vessels in the eye and then the formation of lesions in the retina. This paper presents an automated method for the detection of bright lesions (exudates) in retinal images. In this work, an adaptive thresholding based on a novel algorithm for pure splitting of the image is proposed. A coarse segmentation based on the calculation of a local variation for all image pixels is used to outline the boundaries of all candidates which have clear borders. A morphological operation is used to refine the adaptive thresholding results based on the coarse segmentation results. Using a clinician reference standard (ground truth), images with exudates were detected with 91.2% sensitivity, 99.3% specificity, and 99.5% accuracy. Due to its results the proposed method can achieve superior performance compared to existing techniques and is robust to image quality variability.

Kinetic quantitation of cerebral PET-FDG studies without concurrent blood sampling: statistical recovery of the arterial input function

Kinetic quantitation of dynamic positron emission tomography (PET) studies via compartmental modeling usually requires the time-course of the radio-tracer concentration in the arterial blood as an arterial input function (AIF). For human and animal imaging applications, significant practical difficulties are associated with direct arterial sampling and as a result there is substantial interest in alternative methods that require no blood sampling at the time of the study. A fixed population template input function derived from prior experience with directly sampled arterial curves is one possibility. Image-based extraction, including requisite adjustment for spillover and recovery, is another approach. The present work considers a hybrid statistical approach based on a penalty formulation in which the information derived from a priori studies is combined in a Bayesian manner with information contained in the sampled image data in order to obtain an input function estimate. The absolute scaling of the input is achieved by an empirical calibration equation involving the injected dose together with the subject's weight, height and gender. The technique is illustrated in the context of (18)F -Fluorodeoxyglucose (FDG) PET studies in humans. A collection of 79 arterially sampled FDG blood curves are used as a basis for a priori characterization of input function variability, including scaling characteristics. Data from a series of 12 dynamic cerebral FDG PET studies in normal subjects are used to evaluate the performance of the penalty-based AIF estimation technique. The focus of evaluations is on quantitation of FDG kinetics over a set of 10 regional brain structures. As well as the new method, a fixed population template AIF and a direct AIF estimate based on segmentation are also considered. Kinetics analyses resulting from these three AIFs are compared with those resulting from radially sampled AIFs. The proposed penalty-based AIF extraction method is found to achieve significant improvements over the fixed template and the segmentation methods. As well as achieving acceptable kinetic parameter accuracy, the quality of fit of the region of interest (ROI) time-course data based on the extracted AIF, matches results based on arterially sampled AIFs. In comparison, significant deviation in the estimation of FDG flux and degradation in ROI data fit are found with the template and segmentation methods. The proposed AIF extraction method is recommended for practical use.

Locally constrained mixture representation of dynamic imaging data from PET and MR studies

Dynamic positron emission tomography (PET) studies provide measurements of the kinetics of radiotracers in living tissue. This is a powerful technology which can play a major role in the study of biological processes, potentially leading to better understanding and treatment of disease. Dynamic PET data relate to complex spatiotemporal processes and its analysis poses significant challenges. In previous work, mixture models that expressed voxel-level PET time course data as a convex linear combination of a finite number of dominant time course characteristics (called sub-TACs) were introduced. This paper extends that mixture model formulation to allow for a weighted combination of scaled sub-TACs and also considers the imposition of local constraints in the number of sub-TACs that can be active at any one voxel. An adaptive 3D scaled segmentation algorithm is developed for model initialization. Increases in the weighted residual sums of squares is used to guide the choice of the number of segments and the number of sub-TACs in the final mixture model. The methodology is applied to five data sets from representative PET imaging studies. The methods are also applicable to other contexts in which dynamic image data are acquired. To illustrate this, data from an echo-planar magnetic resonance (MR) study of cerebral hemodynamics are considered. Our analysis shows little indication of departure from a locally constrained mixture model representation with at most two active components at any voxel. Thus, the primary sources of spatiotemporal variation in representative dynamic PET and MR imaging studies would appear to be accessible to a substantially simplified representation in terms of the generalized locally constrained mixture model introduced.

Hybrid image segmentation using watersheds and fast region merging

A hybrid multidimensional image segmentation algorithm is proposed, which combines edge and region-based techniques through the morphological algorithm of watersheds. An edge-preserving statistical noise reduction approach is used as a preprocessing stage in order to compute an accurate estimate of the image gradient. Then, an initial partitioning of the image into primitive regions is produced by applying the watershed transform on the image gradient magnitude. This initial segmentation is the input to a computationally efficient hierarchical (bottom-up) region merging process that produces the final segmentation. The latter process uses the region adjacency graph (RAG) representation of the image regions. At each step, the most similar pair of regions is determined (minimum cost RAG edge), the regions are merged and the RAG is updated. Traditionally, the above is implemented by storing all RAG edges in a priority queue. We propose a significantly faster algorithm, which additionally maintains the so-called nearest neighbor graph, due to which the priority queue size and processing time are drastically reduced. The final segmentation provides, due to the RAG, one-pixel wide, closed, and accurately localized contours/surfaces. Experimental results obtained with two-dimensional/three-dimensional (2-D/3-D) magnetic resonance images are presented.

Metabolic images from dynamic positron emission tomography studies

A dynamic sequence of positron emission tomography (PET) images gives rise to the possibility of creating images of in vivo tissue metabolism. For this reason PET is potentially a valuable instrument in the study of human biology and medicine. The analysis of dynamic PET data to produce metabolic images is a challenging problem from a statistical point of view. For example, a typical data set arising in the study of cerebral glucose utilization has on the order of 30 time-binned images per cross-sectional slice of tissue under examination, each of dimension 128 x 128 pixels. Metabolic imaging requires that the time series at each pixel, known as the time activity curve (TAC), be analysed to produce an estimate of local metabolism. This paper describes a mixture analysis approach to the construction of such metabolic images. In the approach the TAC at a given pixel is expressed as a weighted sum of sub-TACs corresponding to homogeneous tissues represented at the pixel. Estimates of tissue metabolism at the pixel are then constructed as a weighted sum of the metabolism associated with the individual sub-TACs. The procedure is illustrated by application to a [F-18]-labelled deoxyglucose study in a patient with a brain tumour. The ability to map simultaneously a range of parameters related to the transport and biochemical transformation of the radio-tracer, demonstrates the potential power of dynamic PET.

Adaptive image region-growing

Proposes a simple, yet general and powerful, region-growing framework for image segmentation. The region-growing process is guided by regional feature analysis; no parameter tuning or a priori knowledge about the image is required. To decide if two regions should be merged, instead of comparing the difference of region feature means with a predefined threshold, the authors adaptively assess region homogeneity from region feature distributions. This results in an algorithm that is robust with respect to various image characteristics. The merge criterion also minimizes the number of merge rejections and results in a fast region-growing process that is amenable to parallelization.