Qatar Foundation Annual Research Forum Volume 2011 Issue 1

Perfusion-decellularized native lungs seeded with human BJ

RNA-induced pluripotent stem (BJ-RiPS) and umbilical vein endothelial cells failed to regenerate functional lung tissue as quantified by immunohistochemistry (no detection of TTF1,CC10,and Pro-SPB), gene expression (non-significant differences in lung-specific gene expression as compared to cells cultured under standard conditions), and in vitro lung gas exchange properties. Histological analysis of orthotopically transplanted BJ-RiPS lungs revealed a teratoma (detection of ectoderm: TuJ1, mesoderm: SMA, and endoderm: TTF1).

The prevalence of type 2 diabetes mellitus (T2D) is increasing rapidly worldwide with figures being projected to reach 700 million and 366 million by 2030 respectively, according to the recent reports by the World Health Organization 2010 and the International Diabetes Federation 2010. T2D development has been shown to be driven by both environmental and genetic factors.

Consanguinity among Middle-Eastern population, especially the Gulf region, has proved to play a major role in predisposing to multiple hereditary conditions such as cancer, hypertension, and T2D, the latter showing a moderately high prevalence (16.7%) among Qataris.

In this study we aim to identify novel genetic variants and clarify new molecular pathways of T2D in the Qatari population. We will take the advantage of the advanced technologies in genome wide scan and next generation sequencing to investigate a large three generation Qatari family with a history of early onset T2D for the initial stage of the study. More consanguineous families will be recruited for this project and will undergo the same investigational steps in order to identify shared novel mutations between the different family members.

To date, several approaches, such as candidate gene studies, linkage analysis, and genome-wide association studies (GWAS) have been used to identify genetic variants involved in the pathophysiology of T2D and glucose homeostasis. Among these, GWAS has been the most successful approach at the moment to uncovering common genetic variants involved in the disease susceptibility. Next generation whole exome sequencing is a new promising approach to gather novel insights into genes and pathways involved in T2D susceptibility, also allowing the discovery of potential rare mutations.

This study will use next generation sequencing technology to discover potential causative mutations segregating in diabetic inbred Qatari families, and possibly relevant to the Qatari population.

Chronic stress is a leading risk factor for heart diseases, diabetes, asthma and depression. However, physicians find it difficult to continuously track a person's stress levels throughout the day, as current techniques of electrocardiogram and blood pressure monitoring are not practical. There is thus a critical need for a non-invasive, ambulatory device to track physiological stress over extended periods of time. Such information would allow physicians to assess precisely the affect of stress and determine the most appropriate interventions.

The primary objective of this study was to investigate the relationship between non-invasive, physiological signal parameters and the stress level as perceived by the subject. The mapping of physiological parameters onto stress levels to accurately monitor the stress levels in a subject under various conditions will assist the diagnosis of subjects at risk of various stress related disorders.

An ambulatory, wireless device was developed with respiratory rate, galvanic skin response and heart rate sensors, which the subjects can wear comfortably while performing their everyday tasks. An experiment involving 13 activities with different stress levels was conducted on 22 subjects during which physiological data was collected using the developed device. While participating in the experiment, subjects had to record the stress level of each activity on a scale of 1 and 7. The data collected was processed in MATLAB, appropriate signal parameters extracted and then correlated with the subject's perceived stress levels.

Analysis of the data showed that the stress levels varied as the subjects progressed into different activities due to their varying current mental states. The perceived nature of stress varied considerably amongst the individuals with certain activities able to induce a stronger variation in stress compared to others. The derived features were shown to be effective in tracking the variation in stress induced in the subjects.

Results of the experiment showed that the developed device was effective in recording non-invasive physiological for use in tracking the stress levels and mental state of the subjects. Further work is being done to develop an effective model that accurately predicts stress levels based on the physiological data collected.

Three-dimensional (3D) videos are getting quite popular, and equipment for recording and processing them are becoming affordable. Creating 3D videos is expensive. Thus, protecting 3D videos against illegal copying is an important problem. We present a novel system for finding 3D video copies. Our system also identifies the location of the copied part in the reference video. The system can be used, for example, by video content owners, video hosting sites, and third-party companies to find illegally copied 3D videos. To the best of our knowledge, this is the first complete 3D video copy detection system in the literature.

Detecting 3D video copies is a challenging problem. First, comparing numerous numbers of frames from potential copies against reference videos is computationally intensive. Second, many modifications occur on copied videos; some of them are intentional to avoid detection and others are side effects of the copying process. For example, a copied video can be scaled, rotated, cropped, transcoded to a lower bit rate, or embedded into another video. The contrast, brightness, and colors of a video can also be manipulated. Furthermore, 3D videos come in various encoding formats, including stereo, multiview, video plus depth, and multiview plus depth. Changing the format is possible during copying, which complicates the detection process. Finally, new views can be synthesized from existing ones. These views display the scene from different angles, and thus reveal different information than in original views. For example, an object occluded in one view could appear in another.

We implemented the proposed system and evaluated its performance using many 3D videos. We created a large set of query videos with 284 videos to represent all practical scenarios. Our results show that the proposed system achieves high precision and recall values in all scenarios. Specifically, our system results in 100% precision and recall when copied videos are unmodified parts of original videos, and it produces more than 90% precision and recall when copied videos are subjected to various transformations. Even in extreme cases where each video is subjected to five different transformations, our system yields more than 75% precision and recall.

Mobile devices such as smart-phones and tablets are becoming ubiquitous, with ever increasing communication capabilities. In situations where the necessary infrastructure is unavailable, costly, or overloaded, opportunistically connecting theses devices becomes a challenging area of research. Data is disseminated using nodes that store-carry-and-forward messages across the network. In such networks, node cooperation is fundamental for the message delivery process. Therefore, the lack of node cooperation (e.g., a node may refuse to act as a relay and settle for sending and receiving its own data) causes considerable degradation in the network. In order to ensure node cooperation in such networks, we investigate three main challenges: (i) ensuring fair resource utilization among participating mobile devices, (ii) enabling trustful communication between users, and (iii) guaranteeing scalable solutions for large number of devices.

(i) Fairness is particularly important for mobile opportunistic networks since it acts as a major incentive for node cooperation. We propose and evaluate FOG - a real-time distributed framework that ensures efficiency-fairness trade-off for users participating in the opportunistic network.

(ii) Since users may not accept to forward messages in opportunistic networks without incentives, we introduce a set of trust-based filters to provide the user with an option of choosing trustworthy nodes in coordination with personal preferences, location priorities, contextual information, or encounter-based keys.

(iii) Mobile opportunistic solutions should scale to large networks. Our hypothesis is that in large-scale networks, mobile-to-mobile communication has its limitations. We therefore introduce CAF, a Community Aware Forwarding framework, which can easily be integrated with most state-of-the-art algorithms, in order to improve their performance in large-scale networks. CAF uses social information to break down the network into sub-communities, and forward message within and across sub-communities.

In the three contributions we propose above, we adopt a real-trace driven approach to study, analyze, and validate our algorithms and frameworks. Our analysis is based on different mobility traces including the San Francisco taxicab trace, traces collected from conferences such as Infocom’06 and CoNext’07, and Dartmouth campus wireless data set.

Qatar is accumulating substantial local expertise in biomedical data analytics. In particular, QCRI is forming a scientific computing multidisciplinary group with a particular interest in machine learning, statistical modeling and bioinformatics. We are now in a strong position to address the computational needs of biomedical researchers in Qatar, and to prepare a new generation of scientists with a multidisciplinary expertise.

The goal of genomics, proteomics and metabolomics is to identify, and characterize the function of genes, proteins, and small molecules that participate in chemical reactions, and are essential for maintaining life. This research area expands rapidly and holds a great promise in the discovery of risk factors and potential biomarkers of diseases such as obesity and diabetes, the two areas of increasing concern in Qatar population.

In this paper, we develop new statistical modeling techniques of clustering based on mixture models with model selection of large biomedical datasets (proteomics and metabolomics). Deterministic and Bayesian approach are used. The new approach is formulated within the multivariate mixture-model cluster analysis to handle both normal (Gaussian) and non-normal (non-Gaussian) large dimensional data.

To choose the number of component mixture clusters we develop the model selection with information measure of complexity (ICOMP) criterion of the estimated inverse-Fisher information matrix. We have promising preliminary results which, suggest the use of our algorithm to identify obesity susceptibility genes in humans in a genome-wide association study and in Mass spectrum data generated for adipocyte tissue for an obesity study.

In this research, we propose novel techniques to improve automatic speech recognition (ASR) and statistical machine translation (SMT) for dialectal Arabic. Since dialectal Arabic speech resources are very sparse, we describe how existing Modern Standard Arabic (MSA) speech data can be applied to dialectal Arabic acoustic modeling. Our assumption is that MSA is always a second language for all Arabic speakers, and in most cases we can identify the original dialect of a speaker even though he is speaking MSA. Hence, an acoustic model trained with sufficient number of MSA speakers will implicitly model the acoustic features for the different Arabic dialects. Since, MSA and dialectal Arabic do not share the same phoneme set, we propose phoneme sets normalization in order to crosslingually use MSA in dialectal Arabic ASR. After normalization, we applied state-of-the-art acoustic model adaptation techniques to adapt MSA acoustic models with little amount of dialectal speech. Results indicate significant decrease in word error rate (WER). Since it is hard to phonetically transcribe large amounts of dialectal Arabic speech, we studied the use of graphemic acoustic models where phonetic transcription is approximated to be word letters instead of phonemes. A large number of Gaussians in the Gaussian mixture model is used to model missing vowels. In the case of graphemic adaptation, significant decrease in WER was also observed. The approaches were applied with Egyptian Arabic and Levantine Arabic. The reported experimental work was performed while the first author was at the German University in Cairo in collaboration with Ulm University. This work will be extended at Qatar University in collaboration with the University of Illinois to cover ASR and SMT for Qatari broadcast TV. We propose novel algorithms for learning the similarities and differences between Qatari Arabic (QA) and MSA, for purposes of automatic speech translation and speech-to-text machine translation, building on our own definitive research in the relative phonological, morphological, and syntactic systems of QA and MSA, and in the application of translation to interlingual semantic parse. Furthermore, we propose a novel efficient and accurate speech-to-text translation system, building on our research in landmark-based and segment-based ASR.

It is often a problem for a physically challenged person to perform a simple routine task such as eating, moving around, and picking up things on a shelf. Usually, these tasks require assistance from a capable person. However, this total or partial reliance on others for daily routines may be bothersome to the physically challenged and diminishes their self-esteem. Moreover, getting around in a wheel chair, for example, requires the use of some form of a joystick, which is usually not so user-friendly. At Texas A&M Qatar, we developed two vision-based motion detection and actuation systems, which can be used to control the motion of a wheel chair without the use of a joystick and remotely control the motion of a service robot for assistance with daily routines. The first vision system detects the orientation of the face whereas the second detects the motion of color tags placed on the person's body. The orientation of the face and the motion of the color tags are detected using a CCD camera and could be used to command the wheel chair and the remote robot wirelessly. The computation of the color tags’ motion is achieved through image processing using eigenvectors and color system morphology. Through inverse dynamics and coordinate transformation, the motion of the operator's head, limbs, and face orientation could be computed and converted to the appropriate motor angles on the wheelchair and the service robot. Our initial results showed that it takes, on average, 65 milliseconds per calculation. The systems performed well even in complex environments with errors that did not exceed 2 pixels with a response time of about 0.1 seconds. The results of the experiments are available at:

http://www.youtube.com/watch?v=5TC0jqlRe1U, http://www.youtube.com/watch?v=3sJvjXYgwVo, http://www.youtube.com/watch?v=yFxLaVWE3f8, and http://www.youtube.com/watch?v=yFxLaVWE3f8.

It is our intent to implement the vision based sensing system on an actual wheelchair and service robot and test it using a physically challenged person.

The proliferation of wireless technologies and inexpensive network-cameras has enabled low-cost and quick deployment of cameras for several surveillance applications, such as traffic monitoring and border control. Smart Camera Networks (SCNs) are networks of cameras that self configure and adapt to improve their operation and reduce the demand on human operators. However, SCNs are constrained by the ability of the underlying wireless network. Streaming video over a network requires substantial bandwidth, and strict Quality-of-Service (QoS) guarantees. In contrast, existing wireless networks have limited bandwidth, and the protocols do not guarantee QoS. Thus, for SCNs to scale beyond a small number of cameras, it is vital to design efficient video delivery protocols that are aware of the limitations of the underlying wireless network.

We propose to use Video Aggregation, a technique that enables efficient delivery of video in SCNs by combining related video streams. Existing SCNs use traditional routing protocols where intermediate network routers simply forward the video packets from cameras towards the video analysis center (or base-station). This is inefficient in SCNs since multiple cameras often cover overlapping regions of interest, and video information of these regions are redundantly transmitted over the network. The proposed video aggregation protocol eliminates redundant transmissions by dynamically pruning the overlapping areas at the intermediate routers. The routers blend the received streams into one panoramic video stream with no overlaps. Aggregation also dynamically controls the streaming rate to avoid network congestion and packet drops; the routers adjust the rate of the outgoing video by estimating the available network bandwidth. Thus, base-stations receive video frames with minimal packet drops, thus improving the quality of received video.

Our testbed and simulation results show that aggregation outperforms traditional routing both in terms of received video quality and network bandwidth usage. Our testbed experiments show that aggregation improves the received video quality (the Peak Signal-to-Noise-Ratio metric) by 54%. In larger networks, we observed that aggregation eliminates up to 90% of packet drops that were observed in SCNs with traditional routing. In future, we plan to develop a suite of video delivery protocols, which include SCN-aware scheduling and transport protocols.

Hala is a bi-lingual (Arabic and English) robot receptionist located at Carnegie Mellon in Qatar. Hala is presented to users as a 3-D animated face on a screen. Users type to her, and she replies in speech synced with realistic facial expressions. However there are two existing problems with the robot. First, Hala's animation engine does not fully adhere to existing research on face dynamics, which makes it difficult to create natural and interesting facial expressions. Natural expressions help towards an engaging user experience by articulating non-verbal aspects (ex: confusion, glee, horror, etc). Second, while speaking in Arabic lip-movements are not realistic because they were adopted from English utterances. In this work we address these two limitations.

Similar to the movie and video-game industry, we leverage Paul Ekman's seminal work on Facial Action Coding System (FACS) to demarcate Hala's 3D face model into muscle-primitives. These primitives are used to compose complex, yet natural, facial expressions. We have also authored an in-house tool, which allows non-programmers (for ex: artists) to manipulate the face in real-time to create expressions.

The sounds humans make while talking are symbolically captured as “phonemes”. The corresponding shapes of the lips, for these sounds (i.e. phonemes), are called “visemes”. We used existing research and observed each other (and a mirror), to develop visemes that accurately capture Arabic pronunciations. Hala can thus utilize English and Arabic visemes for accurate lip-movement and syncing. We empirically tested and evaluated our work by comparing it with previous lip movements for common Arabic utterances. Nonetheless, certain pronunciations can fire less-than-ideal visemes if they are preceded by silence.

Upon identifying and addressing these limitations, Hala has 11 new facial expressions for a more natural looking and behaving robot. This work also pioneered the first implemented subset of Arabic visemes on a robot.