Innovative Projects Realized

Infrared-Visible video registration

It has been shown in the past, that the combination infrared and visible images improves at short and long distance the detection and segmentation of people, which are the main object of interest in visual surveillance. Indeed, people in visible imagery can be hard to detect when the color contrast with background is low and when the illumination of the scene is poor. In contrast, IR imagery performs generally well in these situations. However, IR imagery is not perfect either as a good thermal contrast is required to detect people. As a result, it is beneficial to use both modalities jointly to capitalize on both of their strength. In our case, we use long wavelength IR and visible cameras in a stereo configuration, which can also provide coarse depth information. We consider both planar and non-planar scenes.
Combining visible and IR to improve the detection of people requires solving many problems. It is especially difficult to find features that match in both modalities to perform registration. Even after registration, both aligned silhouettes (IR/Vis) will be imperfectly segmented. Which parts correspond to the human, which parts do not? My research objectives match these research problems towards the solution for combining IR/Vis human silhouettes for improve detection and tracking of people.

Studying Developer Coordination Patterns in OS Distributions

The success of open source (OS) projects like the Linux kernel and the Apache web server is not only due to their high technical quality, but also because of the unique way in which these projects are delivered to end users. OS distributions like Debian, Ubuntu and countless derivatives customize and bundle thousands of OS projects to make them readily available to millions of end users worldwide.

Distributions play in fact the role of middlemen between “downstream” end users and “upstream” OS projects. This means that distributions do not just deliver software to end users, they also (and especially) filter the user feedback for the upstream projects. For example, if users report a bug or have a question on how to use a particular software project, they typically first communicate with the distribution.

However, since distributions typically only have limited man power and do not have the same level of in-depth expertise about a software project as the original developers, they need to coordinate at some point with the upstream developers. At other times, distributions do not directly contact the upstream project, but first coordinate with the parent distribution they are based on (e.g., Ubuntu’s parent is Debian). Again, the parent distribution can decide to handle a particular piece of feedback itself or to forward it upstream.

Given the many different ways in which end users, OS distributions and upstream OS projects need to coordinate, there is a clear need to analyze the various communication paths and improve their effectiveness. This would decrease the time to fix bugs, increase software quality and user satisfaction, and reduce redundant maintenance effort.

Hence, this project aims to analyze and document the patterns of coordination in OS distributions. In particular, we will mine data from the bug repositories, support fora and mailing list messages of Debian and Ubuntu distributions, and 10 to 20 major OS projects, then analyze this data in both a quantitative and qualitative way. The quantitative analysis will allow us to measure how much feedback is processed by a distribution, its parent distribution and the upstream projects. The qualitative analysis will allow us to study how feedback is treated, i.e., what are the best and worst practices for propagating feedback to the parent distribution? What kinds of tools and methodologies are needed? What are the possible risks?

The output of the project will be a catalogue of patterns that describe the best practices for coordinating user feedback between end users, distributions and upstream projects. Such a catalogue will be beneficial to various stakeholders. First, end users, distributions and upstream projects will better understand each other’s needs, enabling them to reduce the turnaround time of bugs and support requests. Second, new volunteers of a distribution or upstream project can get started more quickly. Third, similar ecosystems like Facebook apps or Eclipse plugins can leverage the experience gathered by distributions to improve their quality.

A graphical interface for semi-automated road user tracking

Although tracking moving objects in video data is more and more common in many applications, for example for surveillance and transportation, no algorithm can perform perfectly in all real world conditions, because of crowding, mixed traffic, varying lighting and weather conditions, etc. In applied fields such as transportation, the quality of the output data is critical, in particular for specific event detection. It is therefore interesting to be able to quickly review tracking results and correct them for subsequent analysis.

The goal of this project is to develop a graphical user interface (GUI) to visualize the results of a tracker applied to road traffic, review the quality of the trajectories and correct them as efficiently (with as little manual user input) as possible. An open source feature-based tracking tool has been developed in our research group (http://bitbucket.org/Nicolas/trafficintelligence) and will produce the results to be reviewed and cleaned in the tool. Tests will be conducted on real world datasets. An extension of the work will be to learn typical tracking errors to automatically identify tracker errors and event correct them automatically.

High Precision Mapping with Mobile Robots

The goal of this project is to explore active sensing strategies to map an environment as precisely and as fast as possible using teams of mobile robots, such as quadrotors and ground robots. Available sensors are assumed to include various types of cameras and RGB-D sensors (kinect), as well as standard navigation sensors (inertial, high precision differential GPS, etc.) for localization. The robotic networked team should take into account at each time the current uncertainty about the map, which depends on occlusions, shadows, etc., in order to replan the trajectories in real-time. For this project, we will develop algorithms, a simulation environment, a proof-of-concept indoor system involving cheap platforms such as AR drones, and finally an advanced outdoor system relying on state of the art hardware.

Genetic factors underlying the survival of Legionella in water

One of the best-studied water-borne bacterial pathogens, in term of its interaction with phagocytic protozoans, is Legionella pneumophila. This species is an important, but often underestimated, cause of community-acquired and nosocomial pneumonia. Transmission occurs primarily by inhalation of contaminated water droplets, but the exact mechanism and other factors influencing virulence remain unclear. Once in the lungs, Legionella infects and replicates inside alveolar macrophages and causes widespread tissue damage. Legionella is also able to replicate within a wide variety of phagocytic protozoans, and can survive for up to a year in the water environment in the absence of any nutrients or susceptible phagocytic protozoans. We have identified a few genes potentially involved in the long-term survival in water by using microarray analysis. The goal of this project is to generate mutant strains for those genes and test the ability of the mutant strains to survive in water.

Structural Investigation of an antibiotic-producing NRPS

Pacidamycins are group of peptide antibiotics which act against the Pseudomonas human pathogens by inhibiting cell wall biosynthesis. Pacidamycins are produced in Streptomyces coeruleorubidus by proteins of a nonribosomal peptide synthetase (NRPS) gene cluster. NRPS are a class of modular proteins that synthesize small peptide products with a wide variety of chemical and biological properties. Interestingly, one protein in the gene cluster, the transferase PacB, is not a typical NRPS protein, and has been shown to function by catalyzing the transfer of an alanine residue from alanyl-tRNA to the amino terminus of the NRPS-tethered peptide intermediate. Thus, PacB departs from normal NRPS logic to appropriate an amino acid substrate from the protein synthesis machinery. The aim of this project is to study how the unusual transferase PacB selects and adds alanine from tRNA to an NRPS peptide. Elucidating this mechanism has important implications in health sciences because it may enable the development of antibacterial drugs by producing pacidamycin derivatives, or using this PacB functionality in bioengineering of other peptide therapeutics.

Enhanced Remote Viewing Capabilities from a Camera Array

The project will entail one or more of the following activities, depending in part by which elements are completed in the summer of 2014.

– adding “last mile support” for rendering 3D stereoscopic display from the dual-interpolated views we’re able to generate
– implementing a dynamic camera-swapping selection mechanism based on the current set of requested viewpoints to achieve maximum quality and smoothness as users move between views
– evaluating the limits of scalability of competing rendering architectures, in which interpolation is carried out client-side rather than server-side
– incorporating recording functionality so that a sufficient subset of the camera outputs can be saved for on-demand viewing of the sessions at a later date

Design and Synthesis of Covalent Enzyme Inhibitors

Introduction. Fibroblast activation protein-? (FAP), an enzyme of the S9 protein family, was identified as a major player in the activation of the stroma required for tumor growth and proliferation and progression of cancer cells. In parallel, prolyl oligopeptidase (POP), an enzyme of the same family, has been found to be responsible for the production of a potent stimulator of angiogenesis and that its inhibition arrests the growth of cancer cells. Recently it was suggested that inhibiting both FAP, blocking stromal invasion, and POP, reducing angiogenesis may lead to growth arrest in different cancer cells types. This report turned our attention towards the design of dual inhibitors or drug combinations as potentially effective cancer therapies.
We have achieved inhibition of POP through the combined expertise of the Moitessier group in drug design and synthesis, of Dr. Juillerat-Jeanneret’s group (Lausanne, Switzerland) in biology and of the Mittermaier group in biophysics to better understand the binding mechanism of these covalent inhibitors and to develop higher throughput biological assays. Our lead molecules will now be further modified into dual POP/FAP inhibitors.

Objectives. In order to validate POP/FAP dual inhibitors or combination of selective inhibitors as cancer therapeutics and develop and test improved POP/FAP inhibitors, we propose to: use our computational drug design tools (Forecaster platform) to design novel dual inhibitors and to synthesize them in the laboratory. Biolgocial assays in other other labs will follow.
Awaiting more information from the professor. Please check back soon. Do not contact Globalink Research Internships.

Perioperative use of opioids and risk of ovarian cancer recurrence and metastasis

Objective. The primary objective of this cohort study is to determine if use of opioids increases the risk of ovarian cancer recurrence and metastasis. The specific objectives are to determine whether: 1) use of IMM opioids during surgery increases the risk of cancer recurrence or metastasis when compared to NIMM opioid use, among patients undergoing general anesthesia; 2) use of postoperative IMM opioids during surgery increases the risk of cancer recurrence or metastasis when compared to NIMM opioid use, among patients undergoing general anesthesia; 3) use of postoperative IMM opioids during surgery increases the risk of cancer recurrence or metastasis when compared to NIMM opioid use, among patients undergoing general anesthesia and epidural, and 4) there is a dose and duration-response relationship between IMM and NIMM opioids at recovery room and the risk of cancer recurrence or metastasis. Knowledge that opioids may prevent cellular immune function, increase angiogenesis and promote breast tumour growth in rodents supports our hypothesis that increased preference of opioid analgesics could be related to increased risk of cancer recurrence and metastasis. Paradoxically, little research has addressed the association of opioids use and cancer recurrence or metastasis, considering that preference for opioid analgesics has increased substantially worldwide.

Method. In this nested case-control study, participants aged ? 18 years who received a first ovarian cancer surgery between 2007 and 2013 will be included in the study cohort. All eligible patients will be followed until: 1) a first cancer recurrence occurs during the study period; 2) death from any cause; 3) development of one exclusion criteria during follow-up; or 4) end of the study period, whichever comes first. The primary outcomes are the incidence of cancer recurrence and metastasis during the study period. Use of IMM or NIMM opioids during surgery and at recovery room will be assessed by chart review among all ovarian cancer patients surgery cohort. We will calculate cumulative dose and time for each participant, and assess use of opioids before and after surgery. We will assess the date of prescription, the quantity prescribed, the duration, and the daily doses also by chart review of the database. Opioids will be separated into IMM and NIMM groups. If patients use opioids before the surgery they will be classified as prevalent cases, whereas if opioids are only used after the surgery, patients will be classified as incident cases. Cox proportional hazards regression will be used to estimate the hazard rate ratios of cancer recurrence and metastases related to the use of IMM or NIMM opioids.

Significance. The results of this study will advance our understanding of the effects of opioids use (IMM and NIMM) on the risk of ovarian cancer recurrence and metastasis, including the function of their dose and duration. These results may help to reduce the risk of recurrence or metastasis in ovarian cancer, as well as in other types of cancer. Pharmacotherapy optimization can play an important role in reducing the public health impact of cancer recurrence or metastasis.

Energy-efficient thermal management in deep underground mines

It is well known that underground miners are exposed to one of the most dangerous working environments on earth. As miners delve deeper into the earth, the rock pressure and temperature rise, which in turn have an adverse effect on the environment and thus lead to higher operating costs. Furthermore, in coal mines where methane gas and dust is also produced in coal beds, the combination of methane accumulation, oxygen and spark/heat can trigger explosions – which may cause fatal accidents. Generally, mining environments are dynamic and changing very rapidly during extraction process. Essentially, changes in mine environment affect the performance of ventilation system. Careful estimation and design of air flow rate, spray system, cooling load and additional ventilation auxiliary equipment are of importance to ensure safety, comfortability and productivity whilst maintaining low operating cost.
In this research, three-dimensional structures of a mine development zone together with its auxiliary ventilation system are simulated using a computational fluid dynamic approach. Several designs and ventilations as well as heating and/or cooling scenarios are evaluated with respect to methane dispersion, dust distribution, diesel emission control and thermal management in mines. Effects of design and operating parameters, e.g., fan power, placement of air duct, length of mine advancement, diesel emission and heat generated from mining machine as well as heating/cooling loads, are examined to ensure good and safe design. The advantages and limitations of each design are discussed and compared not only in terms of quality and quantity, but also in terms of the overall pressure drop and energy requirements which represent the associated cost of ventilation and air conditioning system.

Developing new viscoelastic substrates for cell mechanics studies

Traction force microscopy is a well-established methodology to measure the local contractile forces exerted by a cell, and is accomplished by measuring the deformation of a soft elastic substrate by cell forces. From this measured deformation, the strain is calculated, and by knowing the elastic modulus of the substrate, the forces are calculated. Key changes in physiology and pathology are reflected in these cell forces, such as in metastatic progression of cancer. These substrates accurately reflect the cell forces, but poorly reflect the total work done by the cell, as rather quickly a force-equilibrium is reached; in order to do so, we must be able to include a dissipative component to the substrate, allowing cells to continuously expend energy deforming the surface. Such measurements will allow new measurements of cell contractile work, as well as characterizing time-scales of cell structural relaxation. Moreover, these materials will allow us to not only examine the interaction of cells with elastic substrates, but to measure the role of dissipation in determining cell activity, from movement to stem cell differentiation.

Regulating Innovation: Law and the Creative District

We will investigate the relationship between law and innovation in our newly globalized era. We will draw on well-documented innovative districts in Montreal and Silicon Valley to assess how law promotes or hinders such innovation. We will consider emerging initiatives such as the Blueseed startup community, which is a proposed un-regulated innovation zone to be located in international waters in order to avoid national regulatory compliance regimes.

Innovation, for the purposes of this project, will be defined as the creation and development of new ideas, products, and methodology–particularly within the technology industry. This industry’s reliance on intangibles to create value has created a new paradigm for property law and has prompted both a worldwide competition for capital and the ability for corporations to essentially opt-out of taxes and legal regimes that are unfavourable to their revenue streams. The state has accordingly responded in an effort to reap the advantages of capital inflow and to concurrently restrict the benefits offered to corporations to those contemplated by the legislature.

Our main focus will be how state regulation affects innovation. States often use tax incentives to lure businesses in, with the hope of future benefits from increased employment, consumer spending and other spillovers. We will ask if tax incentives are really determinative with regard to a corporation’s decision to move to a given jurisdiction and if the anti-abuse measures in place are adequate to control the scope of the incentives. From a property perspective, the state regulates innovation directly through patent law, and indirectly by controlling the dissemination of knowledge and know-how through non-compete covenants. With respect to this aspect, we will ask if the current allocation of property rights and knowledge sharing controls is still relevant in our open-access era, and if such regulation is conducive to sustained innovation.