Right ventricle extraction by low level and model-based algorithm

In this paper we present an algorithm as the combination of a low level morphological operation and model based Global Circular Shortest Path scheme to explore the segmentation of the Right Ventricle. Traditional morphological operations were employed to obtain the region of interest, and adjust it to generate a mask. The image cropped by the mask is then partitioned into a few overlapping regions. Global Circular Shortest Path algorithm is then applied to extract the contour from each partition. The final step is to re-assemble the partitions to create the whole contour. The technique is deemed quite reliable and robust, as this is illustrated by a very good agreement between the extracted contour and the expert manual drawing output.

Biomedical events extraction using the hidden vector state model

Objective: Biomedical events extraction concerns about events describing changes on the state of bio-molecules from literature. Comparing to the protein-protein interactions (PPIs) extraction task which often only involves the extraction of binary relations between two proteins, biomedical events extraction is much harder since it needs to deal with complex events consisting of embedded or hierarchical relations among proteins, events, and their textual triggers. In this paper, we propose an information extraction system based on the hidden vector state (HVS) model, called HVS-BioEvent, for biomedical events extraction, and investigate its capability in extracting complex events. Methods and material: HVS has been previously employed for extracting PPIs. In HVS-BioEvent, we propose an automated way to generate abstract annotations for HVS training and further propose novel machine learning approaches for event trigger words identification, and for biomedical events extraction from the HVS parse results. Results: Our proposed system achieves an F-score of 49.57% on the corpus used in the BioNLP'09 shared task, which is only 2.38% lower than the best performing system by UTurku in the BioNLP'09 shared task. Nevertheless, HVS-BioEvent outperforms UTurku's system on complex events extraction with 36.57% vs. 30.52% being achieved for extracting regulation events, and 40.61% vs. 38.99% for negative regulation events. Conclusions: The results suggest that the HVS model with the hierarchical hidden state structure is indeed more suitable for complex event extraction since it could naturally model embedded structural context in sentences.

Extracting protein-protein interactions from MEDLINE using the hidden vector state model

A major challenge in text mining for biomedicine is automatically extracting protein-protein interactions from the vast amount of biomedical literature. We have constructed an information extraction system based on the Hidden Vector State (HVS) model for protein-protein interactions. The HVS model can be trained using only lightly annotated data whilst simultaneously retaining sufficient ability to capture the hierarchical structure. When applied in extracting protein-protein interactions, we found that it performed better than other established statistical methods and achieved 61.5% in F-score with balanced recall and precision values. Moreover, the statistical nature of the pure data-driven HVS model makes it intrinsically robust and it can be easily adapted to other domains.

Feature Extraction for Classification in the Data Mining Process

Dimensionality reduction is a very important step in the data mining process. In this paper, we consider feature extraction for classification tasks as a technique to overcome problems occurring because of “the curse of dimensionality”. Three different eigenvector-based feature extraction approaches are discussed and three different kinds of applications with respect to classification tasks are considered. The summary of obtained results concerning the accuracy of classification schemes is presented with the conclusion about the search for the most appropriate feature extraction method. The problem how to discover knowledge needed to integrate the feature extraction and classification processes is stated. A decision support system to aid in the integration of the feature extraction and classification processes is proposed. The goals and requirements set for the decision support system and its basic structure are defined. The means of knowledge acquisition needed to build up the proposed system are considered.

A Statistical Approach for Multilingual Document Clustering and Topic Extraction from Clusters

2000 Mathematics Subject Classification: 62H30

Model Mining and Efficient Verification of Software Product Lines

Software product line modeling aims at capturing a set of software products in an economic yet meaningful way. We introduce a class of variability models that capture the sharing between the software artifacts forming the products of a software product line (SPL) in a hierarchical fashion, in terms of commonalities and orthogonalities. Such models are useful when analyzing and verifying all products of an SPL, since they provide a scheme for divide-and-conquer-style decomposition of the analysis or verification problem at hand. We define an abstract class of SPLs for which variability models can be constructed that are optimal w.r.t. the chosen representation of sharing. We show how the constructed models can be fed into a previously developed algorithmic technique for compositional verification of control-flow temporal safety properties, so that the properties to be verified are iteratively decomposed into simpler ones over orthogonal parts of the SPL, and are not re-verified over the shared parts. We provide tool support for our technique, and evaluate our tool on a small but realistic SPL of cash desks.

Ensemble Stream Model for Data-Cleaning in Sensor Networks

Ensemble Stream Modeling and Data-cleaning are sensor information processing systems have different training and testing methods by which their goals are cross-validated. This research examines a mechanism, which seeks to extract novel patterns by generating ensembles from data. The main goal of label-less stream processing is to process the sensed events to eliminate the noises that are uncorrelated, and choose the most likely model without over fitting thus obtaining higher model confidence. Higher quality streams can be realized by combining many short streams into an ensemble which has the desired quality. The framework for the investigation is an existing data mining tool. First, to accommodate feature extraction such as a bush or natural forest-fire event we make an assumption of the burnt area (BA*), sensed ground truth as our target variable obtained from logs. Even though this is an obvious model choice the results are disappointing. The reasons for this are two: One, the histogram of fire activity is highly skewed. Two, the measured sensor parameters are highly correlated. Since using non descriptive features does not yield good results, we resort to temporal features. By doing so we carefully eliminate the averaging effects; the resulting histogram is more satisfactory and conceptual knowledge is learned from sensor streams. Second is the process of feature induction by cross-validating attributes with single or multi-target variables to minimize training error. We use F-measure score, which combines precision and accuracy to determine the false alarm rate of fire events. The multi-target data-cleaning trees use information purity of the target leaf-nodes to learn higher order features. A sensitive variance measure such as ƒ-test is performed during each node's split to select the best attribute. Ensemble stream model approach proved to improve when using complicated features with a simpler tree classifier. The ensemble framework for data-cleaning and the enhancements to quantify quality of fitness (30% spatial, 10% temporal, and 90% mobility reduction) of sensor led to the formation of streams for sensor-enabled applications. Which further motivates the novelty of stream quality labeling and its importance in solving vast amounts of real-time mobile streams generated today.

Development of An Enhanced Oil Sands Fluid Tailings Generation Model

Bitumen extraction from surface-mined oil sands results in the production of large volumes of Fluid Fine Tailings (FFT). Through Directive 085, the Province of Alberta has signaled that oil sands operators must improve and accelerate the methods by which they deal with FFT production, storage and treatment. This thesis aims to develop an enhanced method to forecast FFT production based on specific ore characteristics. A mass relationship and mathematical model to modify the Forecasting Tailings Model (FTM) by using fines and clay boundaries, as the two main indicators in FFT accumulation, has been developed. The modified FTM has been applied on representative block model data from an operating oil sands mining venture. An attempt has been made to identify order-of-magnitude associated tailings treatment costs, and to improve financial performance by not processing materials that have ultimate ore processing and tailings storage and treatment costs in excess of the value of bitumen they produce. The results on the real case study show that there is a 53% reduction in total tailings accumulations over the mine life by selectively processing only lower tailings generating materials through eliminating 15% of total mined ore materials with higher potential of fluid fines inventory. This significant result will assess the impact of Directive 082 on mining project economic and environmental performance towards the sustainable development of mining projects.

Work Extraction and Energy Storage in the Dicke Model

We study work extraction from the Dicke model achieved using simple unitary cyclic transformations keeping into account both a non optimal unitary protocol, and the energetic cost of creating the initial state. By analyzing the role of entanglement, we find that highly entangled states can be inefficient for energy storage when considering the energetic cost of creating the state. Such surprising result holds notwithstanding the fact that the criticality of the model at hand can sensibly improve the extraction of work. While showing the advantages of using a many-body system for work extraction, our results demonstrate that entanglement is not necessarily advantageous for energy storage purposes, when non optimal processes are considered. Our work shows the importance of better understanding the complex interconnections between non-equilibrium thermodynamics of quantum systems and correlations among their subparts.

A numerical model for prediction of oil, trajectory due to extraction activity in the north coasts of the Persian Gulf

This thesis considers a three- dimensional numerical model based on 3-D Navier— Stokes and continuity equations involving various wind speeds (North west), water surface levels, horizontal shier stresses, eddy viscosity, densities of oil and gas condensate- water mixture flows. The model is used to simulate the prediction of the surface movement of oil and gas condensate slicks from spill accident in the north coasts of Persian Gulf.

Study of salinity gradients in the Persian Gulf and an experimental model for electric energy extraction from them using nano-membranes

Salinity gradient power (SGP) is the energy that can be obtained from the mixing entropy of two solutions with a different salt concentration. River estuary, as a place for mixing salt water and fresh water, has a huge potential of this renewable energy. In this study, this potential in the estuaries of rivers leading to the Persian Gulf and the factors affecting it are analysis and assessment. Since most of the full water rivers are in the Asia, this continent with the potential power of 338GW is a second major source of energy from the salinity gradient power in the world (Wetsus institute, 2009). Persian Gulf, with the proper salinity gradient in its river estuaries, has Particular importance for extraction of this energy. Considering the total river flow into the Persian Gulf, which is approximately equal to 3486 m3/s, the amount of theoretical extractable power from salinity gradient in this region is 5.2GW. Iran, with its numerous rivers along the coast of the Persian Gulf, has a great share of this energy source. For example, with study calculations done on data from three hydrometery stations located on the Arvand River, Khorramshahr Station with releasing 1.91M/ energy which is obtained by combining 1.26m3 river water with 0.74 m3 sea water, is devoted to itself extracting the maximum amount of extractable energy. Considering the average of annual discharge of Arvand River in Khorramshahr hydrometery station, the amount of theoretical extractable power is 955 MW. Another part of parameters that are studied in this research, are the intrusion length of salt water and its flushing time in the estuary that have a significant influence on the salinity gradient power. According to the calculation done in conditions HWS and the average discharge of rivers, the maximum of salinity intrusion length in to the estuary of the river by 41km is related to Arvand River and the lowest with 8km is for Helle River. Also the highest rate of salt water flushing time in the estuary with 9.8 days is related to the Arvand River and the lowest with 3.3 days is for Helle River. Influence of these two parameters on reduces the amount of extractable energy from salinity gradient power as well as can be seen in the estuaries of the rivers studied. For example, at the estuary of the Arvand River in the interval 8.9 days, salinity gradient power decreases 9.2%. But another part of this research focuses on the design of a suitable system for extracting electrical energy from the salinity gradient. So far, five methods have been proposed to convert this energy to electricity that among them, reverse electro-dialysis (RED) method and pressure-retarded osmosis (PRO) method have special importance in practical terms. In theory both techniques generate the same amount of energy from given volumes of sea and river water with specified salinity; in practice the RED technique seems to be more attractive for power generation using sea water and river water. Because it is less necessity of salinity gradient to PRO method. In addition to this, in RED method, it does not need to use turbine to change energy and the electricity generation is started when two solutions are mixed. In this research, the power density and the efficiency of generated energy was assessment by designing a physical method. The physical designed model is an unicellular reverse electro-dialysis battery with nano heterogenic membrane has 20cmx20cm dimension, which produced power density 0.58 W/m2 by using river water (1 g NaCl/lit) and sea water (30 g NaCl/lit) in laboratorial condition. This value was obtained because of nano method used on the membrane of this system and suitable design of the cell which led to increase the yield of the system efficiency 11% more than non nano ones.

Evaluation of hemosponge in promoting dental socket healing after 3rd mandibular premolar extraction in a feline model

Aim: To investigate the healing process following use of collagen sponges in the dental socket after extraction. Wound complications during the study were also evaluated. Methods: 32 cats were included in this study. IV administration of the combination of diazepam (0.22 mg/kg) and ketamine (10 mg/kg) was used to induce general anesthesia. Surgical extraction of both 3rd mandibular premolars was performed. The open dental sockets were divided in two groups. In Group A, the open dental socket on the left side was closed using 4-0 Monocryl in simple interrupted pattern. In Group B, the right dental socket was filled with lyophilized hydrolyzed collagen and the buccal and lingual flaps were sutured using 4-0 Monocryl and simple interrupted pattern. Meloxicam (0.2 mg/kg) was used to manage the post-extraction pain in all cats. Ampicilline 20 mg/kg was used as prophylaxis. The wounds were observed during the study to evaluate any signs of inflammation or dehiscence. Radiographs were taken to compare healing of the socket 3 weeks after the procedure. A 1 mm biopsy punch sample was taken from sockets in all cats for comparison of the healing in both groups. Results: Hemorrhage occurred only in the sockets of Group A. Remission of radiolucent area occurred in both groups. Mean score of inflammation was lower and mean scores of fibrotic reaction and fibroplasia were higher in Group B (p<0.05). Conclusions: Use of hemosponge in alveolar socket may accelerate fibroplasia and formation of the connective tissue and reduce inflammation after tooth extraction. Therefore, post-extraction use of the hemostatic agent in the dental socket is recommended.

Ensemble Stream Model for Data-Cleaning in Sensor Networks

Ensemble Stream Modeling and Data-cleaning are sensor information processing systems have different training and testing methods by which their goals are cross-validated. This research examines a mechanism, which seeks to extract novel patterns by generating ensembles from data. The main goal of label-less stream processing is to process the sensed events to eliminate the noises that are uncorrelated, and choose the most likely model without over fitting thus obtaining higher model confidence. Higher quality streams can be realized by combining many short streams into an ensemble which has the desired quality. The framework for the investigation is an existing data mining tool. First, to accommodate feature extraction such as a bush or natural forest-fire event we make an assumption of the burnt area (BA*), sensed ground truth as our target variable obtained from logs. Even though this is an obvious model choice the results are disappointing. The reasons for this are two: One, the histogram of fire activity is highly skewed. Two, the measured sensor parameters are highly correlated. Since using non descriptive features does not yield good results, we resort to temporal features. By doing so we carefully eliminate the averaging effects; the resulting histogram is more satisfactory and conceptual knowledge is learned from sensor streams. Second is the process of feature induction by cross-validating attributes with single or multi-target variables to minimize training error. We use F-measure score, which combines precision and accuracy to determine the false alarm rate of fire events. The multi-target data-cleaning trees use information purity of the target leaf-nodes to learn higher order features. A sensitive variance measure such as f-test is performed during each node’s split to select the best attribute. Ensemble stream model approach proved to improve when using complicated features with a simpler tree classifier. The ensemble framework for data-cleaning and the enhancements to quantify quality of fitness (30% spatial, 10% temporal, and 90% mobility reduction) of sensor led to the formation of streams for sensor-enabled applications. Which further motivates the novelty of stream quality labeling and its importance in solving vast amounts of real-time mobile streams generated today.