Güleser Kalayci Demir

Patient oriented and robust automatic liver segmentation for pre-evaluation of liver transplantation

Identifying liver region from abdominal computed tomography-angiography (CTA) data sets is one of the essential steps in evaluation of transplantation donors prior to the hepatic surgery. However, due to gray level similarity of adjacent organs, injection of contrast media and partial volume effects; robust segmentation of the liver is a very difficult task. Moreover, high variations in liver margins, different image characteristics with different CT scanners and atypical liver shapes make the segmentation process even harder. In this paper, we propose a three stage (i.e. pre-processing, classification, post-processing); automatic liver segmentation algorithm that adapts its parameters according to each patient by learning the data set characteristics in parallel to segmentation process to address all the challenging aspects mentioned above. The efficiency in terms of the time requirement and the overall segmentation performance is achieved by introducing a novel modular classification system consisting of a K-Means based simple classification system and an MLP based complex one which are combined with a data-dependent and automated switching mechanism that decides to apply one of them. Proposed approach also makes the design of the overall classification system fully unsupervised that depends on the given CTA series only without requiring any given training set of CTA series. The segmentation results are evaluated by using area error rate and volume calculations and the success rate is calculated as 94.91% over a data set of diverse CTA series of 20 patients according to the evaluation of the expert radiologist. The results show that, the proposed algorithm gives better results especially for atypical liver shapes and low contrast studies where several algorithms fail.

Online local learning algorithms for linear discriminant analysis

Online local learning algorithms for a laterally-connected single-layer neural network for performing linear discriminant analysis have been proposed. A convergence proof is provided for the algorithm based on Hebbian learning. The algorithms are simulated and applied to the face recognition problem.

Bidtree Ordering in IDA* Combinatorial Auction Winner-Determination with Side Constraints

We extend Sandholms bidtree-based IDA* algorithm for combinatorial auction winner determination to deal with negotiation over tasks with precedence constraints. We evaluate its performance, and show that the order of items in the bidtree has a major impact on performance. Specifically, performance is enhanced if the items with the largest numbers of bids are at the top of the bidtree. This is due to the fact that the effective branching factor in the search tree is controlled by the number of bids returned from a query to the bidtree, which in turn is strongly related to its construction.

Terrain classification through weakly-structured vehicle/terrain interaction

We present a new terrain classification technique both for effective, autonomous locomotion over natural, unknown terrains and for the qualitative analysis of terrains for exploration and mapping. Our straight-forward approach requires a single camera with little processing of visual information. Specifically, we derived a gait bounce measure from visual servoing errors that result from vehicle-terrain interactions during normal locomotion. Characteristics of the terrain, such as roughness and compliance, manifest themselves in the spatial patterns of this signal and can be extracted using pattern classification techniques. For legged robots, different limb-terrain interactions generate gait bounce signals with different information content, thus deliberate limb motions can effect higher information content (i.e. the robot is an active sensor of terrain class). Segmentation of the gait cycle based on the limb-terrain interaction isolates portions of the gait bounce signal with high information content. The decoding of, then sequencing of, this content from each cycle segment yields a robust classification of terrain type from known benchmarks. To extract this spatio-temporal pattern of the gait bounce signal, we developed a meta-classifier using discriminant analysis and hidden Markov model. We present the gait bounce derivation. We demonstrate the viability of terrain classification for legged vehicles using gait bounce with a rigorous study of more than 700 trials, obtaining 84% accuracy. We describe how terrain classification can be used for gait adaptation, particularly in relation to an efficiency metric. We also demonstrate that our technique is generally applicable to other locomotion mechanisms such as wheels and treads.

A video tracking based improvement of acute toxicity test on Artemia salina

Booster biocides have negative impacts on marine organisms and it is therefore essential to monitor their toxicity prior to their use in antifouling paints. In this study, we describe a method for improving the reliability of a well-used acute toxicity test. We studied a model organism, Artemia salina, by using a novel video tracking method. Acute toxicities of a reference toxicant (potassium dichromate) and a potential toxicant (p-coumaric acid) were examined for newly hatched A. salina nauplii. Survival percentage, swimming velocity, and paths covered by nauplii were determined with the new algorithm. A. salina nauplii were affected by both tested toxicants, an outcome clearly detected by the algorithm. The results demonstrate that the video tracking algorithm could be used for testing the acute toxicities of booster biocides as well as other potential toxicants on A. salina. With slight modifications, it could be used for testing other similar aquatic micro-organisms. This report has linked video sequences.

Terrain Classification Using Weakly-Structured Vehicle/Terrain Interaction

We present a new terrain classification technique both for effective, autonomous locomotion over rough, unknown terrains and for the qualitative analysis of terrains for exploration and mapping. Our approach requires a single camera with little processing of visual information. Specifically, we derived a gait bounce measure from visual servoing errors that results from vehicle-terrain interactions during normal locomotion. Characteristics of the terrain, such as roughness and compliance, manifest themselves in the spatial patterns of this signal and can be extracted using pattern classification techniques. This vision-based approach is particularly beneficial for resource-constrained robots with limited sensor capability. In this paper, we present the gait bounce derivation. We demonstrate the viability of terrain classification for legged vehicles using gait bounce with a rigorous study of more than 700 trials, obtaining an 83% accuracy on a set of laboratory terrains. We describe how terrain classification may be used for gait adaptation, particularly in relation to an efficiency metric. We also demonstrate that our technique may be generally applicable to other locomotion mechanisms such as wheels and treads.

Motion estimation with cooperatively working multiple robots

We have investigated the performance of simultaneously estimating the 3D motion and structure for navigation when the scale information is obtained by utilizing the cooperative efforts of multiple robots. The method determines the relative positions of robots by tracking a specific geometric feature that is part of their structure, and then uses the extended Kalman filter to estimate the motion and structure. For implementation we used two CRAWLER Scouts, and performed several experiments to explore the effects of cooperative running of robots on the motion estimation.

Clustering within quantum mechanical framework

We study clustering problem within quantum mechanical framework by utilizing the Schroedinger equation written for the lowest energy state. We extend the analysis of Horn and Gottlieb [1] by providing an explicit discussion of probability distribution within full quantum mechanical context and examine the clustering performances for various probability distribution functions with numerical experiments.

Risk and user preferences in winner determination

We discuss a solution to the winner determination problem which takes into account not only costs but also risk aversion of the agent that accepts the bids and works for tasks that have time and precedence constraints. We develop an equivalent unit approach to the group of tasks to analyze the system and use Expected Utility Theory as the basic mechanism for decision-making. Our theoretical and experimental analysis shows that Expected Utility is especially useful for choosing between cheap-but-risky and costly-but-safe bids. Moreover, we show how bids with similar costs and similar probabilities of being successfully completed but different time windows can be efficiently selected or rejected.

Coexistence of Deterministic and Stochastic Bistability in a 1-D Birth-Death Process with Hill Type Nonlinear Birth Rates

The paper shows that, for a specific 1 − D birth-death process, the parameter ranges for the existence of the deterministic bistability exactly coincide with the parameter ranges for the existence of the stochastic bistability, namely bimodality. The considered 1 − D birth-death process is a reduced model of TMG induced lactose operon of Escherichia coli in which the birth rate for intra cellular TMG molecules is of Hill type. As opposed to the results reported by some works for other 1 − D birth-death processes in the literature, such as the observation called Keizer paradox, no bistability without bimodality and also no bimodality without bistability are obtained for any set of model parameters. Bistability without bimodality is observed to occur only when invalid calculations of steady-state probabilities due to i) big number problems related to very small probabilities corresponding to troughs between two peaks and/or ii) inappropriately low choice of molecule numbers are not prevented.