An evaluation of the pedestrian classification in a multi-domain multi-modality setup

The objective of this article is to study the problem of pedestrian classification across different light spectrum domains (visible and far-infrared (FIR)) and modalities (intensity, depth and motion). In recent years, there has been a number of approaches for classifying and detecting pedestrians in both FIR and visible images, but the methods are difficult to compare, because either the datasets are not publicly available or they do not offer a comparison between the two domains. Our two primary contributions are the following: (1) we propose a public dataset, named RIFIR , containing both FIR and visible images collected in an urban environment from a moving vehicle during daytime; and (2) we compare the state-of-the-art features in a multi-modality setup: intensity, depth and flow, in far-infrared over visible domains. The experiments show that features families, intensity self-similarity (ISS), local binary patterns (LBP), local gradient patterns (LGP) and histogram of oriented gradients (HOG), computed from FIR and visible domains are highly complementary, but their relative performance varies across different modalities. In our experiments, the FIR domain has proven superior to the visible one for the task of pedestrian classification, but the overall best results are obtained by a multi-domain multi-modality multi-feature fusion.

Content-independent embedding scheme for multi-modal medical image watermarking

Background As the increasing adoption of information technology continues to offer better distant medical services, the distribution of, and remote access to digital medical images over public networks continues to grow significantly. Such use of medical images raises serious concerns for their continuous security protection, which digital watermarking has shown great potential to address. Methods We present a content-independent embedding scheme for medical image watermarking. We observe that the perceptual content of medical images varies widely with their modalities. Recent medical image watermarking schemes are image-content dependent and thus they may suffer from inconsistent embedding capacity and visual artefacts. To attain the image content-independent embedding property, we generalise RONI (region of non-interest, to the medical professionals) selection process and use it for embedding by utilising RONI’s least significant bit-planes. The proposed scheme thus avoids the need for RONI segmentation that incurs capacity and computational overheads. Results Our experimental results demonstrate that the proposed embedding scheme performs consistently over a dataset of 370 medical images including their 7 different modalities. Experimental results also verify how the state-of-the-art reversible schemes can have an inconsistent performance for different modalities of medical images. Our scheme has MSSIM (Mean Structural SIMilarity) larger than 0.999 with a deterministically adaptable embedding capacity. Conclusions Our proposed image-content independent embedding scheme is modality-wise consistent, and maintains a good image quality of RONI while keeping all other pixels in the image untouched. Thus, with an appropriate watermarking framework (i.e., with the considerations of watermark generation, embedding and detection functions), our proposed scheme can be viable for the multi-modality medical image applications and distant medical services such as teleradiology and eHealth.

A visualisation framework for collaborative virtual environment usage information

Due to the popularity of modern Collaborative Virtual Environments, there has been a related increase in their size and complexity. Developers therefore need visualisations that expose usage patterns from logged data, to understand the structures and dynamics of these complex environments. This chapter presents a new framework for the process of visualising virtual environment usage data. Major components, such as an event model, designer task model and data acquisition infrastructure are described. Interface and implementation factors are also developed, along with example visualisation techniques that make use of the new task and event model. A case study is performed to illustrate a typical scenario for the framework, and its benefits to the environment development team.

Toxicity profile of the immunomodulatory thalidomide analogue, lenalidomide : Phase I clinical trial of three dosing schedules in patients with solid malignancies

Thalidomide is an anti-angiogenic agent currently used to treat patients with malignant cachexia or multiple myeloma. Lenalidomide (CC-5013) is an immunomodulatory thalidomide analogue licensed in the United States of America (USA) for the treatment of a subtype of myelodysplastic syndrome. This two-centre, open-label phase I study evaluated dose-limiting toxicities in 55 patients with malignant solid tumours refractory to standard chemotherapies. Lenalidomide capsules were consumed once daily for 12 weeks according to one of the following three schedules: (I) 25 mg daily for the first 7 d, the daily dose increased by 25 mg each week up to a maximum daily dose of 150 mg; (II) 25 mg daily for 21 d followed by a 7-d rest period, the 4-week cycle repeated for 3 cycles; (III) 10 mg daily continuously. Twenty-six patients completed the study period. Two patients experienced a grade 3 hypersensitivity rash. Four patients in cohort I and 4 patients in cohort II suffered grade 3 or 4 neutropaenia. In 2 patients with predisposing medical factors, grade 3 cardiac dysrhythmia was recorded. Grade 1 neurotoxicity was detected in 6 patients. One complete and two partial radiological responses were measured by computed tomography scanning; 8 patients had stable disease after 12 weeks of treatment. Fifteen patients remained on treatment as named patients; 1 with metastatic melanoma remains in clinical remission 3.5 years from trial entry. This study indicates the tolerability and potential clinical efficacy of lenalidomide in patients with advanced solid tumours who have previously received multi-modality treatment. Depending on the extent of myelosuppressive pre-treatment, dose schedules (II) or (III) are advocated for large-scale trials of long-term administration. © 2006 Elsevier Ltd. All rights reserved.

A Plasmonic Gold Nanostar Theranostic Probe for In Vivo Tumor Imaging and Photothermal Therapy.

Nanomedicine has attracted increasing attention in recent years, because it offers great promise to provide personalized diagnostics and therapy with improved treatment efficacy and specificity. In this study, we developed a gold nanostar (GNS) probe for multi-modality theranostics including surface-enhanced Raman scattering (SERS) detection, x-ray computed tomography (CT), two-photon luminescence (TPL) imaging, and photothermal therapy (PTT). We performed radiolabeling, as well as CT and optical imaging, to investigate the GNS probe's biodistribution and intratumoral uptake at both macroscopic and microscopic scales. We also characterized the performance of the GNS nanoprobe for in vitro photothermal heating and in vivo photothermal ablation of primary sarcomas in mice. The results showed that 30-nm GNS have higher tumor uptake, as well as deeper penetration into tumor interstitial space compared to 60-nm GNS. In addition, we found that a higher injection dose of GNS can increase the percentage of tumor uptake. We also demonstrated the GNS probe's superior photothermal conversion efficiency with a highly concentrated heating effect due to a tip-enhanced plasmonic effect. In vivo photothermal therapy with a near-infrared (NIR) laser under the maximum permissible exposure (MPE) led to ablation of aggressive tumors containing GNS, but had no effect in the absence of GNS. This multifunctional GNS probe has the potential to be used for in vivo biosensing, preoperative CT imaging, intraoperative detection with optical methods (SERS and TPL), as well as image-guided photothermal therapy.

Non-invasive in vivo imaging in small animal research

Non-invasive real time in vivo molecular imaging in small animal models has become the essential bridge between in vitro data and their translation into clinical applications. The tremendous development and technological progress, such as tumour modelling, monitoring of tumour growth and detection of metastasis, has facilitated translational drug development. This has added to our knowledge on carcinogenesis. The modalities that are commonly used include Magnetic Resonance Imaging (MRI), Computed Tomography (CT), Positron Emission Tomography (PET), bioluminescence imaging, fluorescence imaging and multi-modality imaging systems. The ability to obtain multiple images longitudinally provides reliable information whilst reducing animal numbers. As yet there is no one modality that is ideal for all experimental studies. This review outlines the instrumentation available together with corresponding applications reported in the literature with particular emphasis on cancer research. Advantages and limitations to current imaging technology are discussed and the issues concerning small animal care during imaging are highlighted.

Partial volume correction using bimodal contrast agents in PET-MRI

Tese de mestrado integrado em Engenharia Biomédica e Biofísica, apresentada à Universidade de Lisboa, através da Faculdade de Ciências, 2014

A Diagnostic Assessment Of Evolutionary Multiobjective Optimization For Water Resources Systems

This study contributes a rigorous diagnostic assessment of state-of-the-art multiobjective evolutionary algorithms (MOEAs) and highlights key advances that the water resources field can exploit to better discover the critical tradeoffs constraining our systems. This study provides the most comprehensive diagnostic assessment of MOEAs for water resources to date, exploiting more than 100,000 MOEA runs and trillions of design evaluations. The diagnostic assessment measures the effectiveness, efficiency, reliability, and controllability of ten benchmark MOEAs for a representative suite of water resources applications addressing rainfall-runoff calibration, long-term groundwater monitoring (LTM), and risk-based water supply portfolio planning. The suite of problems encompasses a range of challenging problem properties including (1) many-objective formulations with 4 or more objectives, (2) multi-modality (or false optima), (3) nonlinearity, (4) discreteness, (5) severe constraints, (6) stochastic objectives, and (7) non-separability (also called epistasis). The applications are representative of the dominant problem classes that have shaped the history of MOEAs in water resources and that will be dominant foci in the future. Recommendations are provided for which modern MOEAs should serve as tools and benchmarks in the future water resources literature.

Precise Near-to-Head Acoustics with Binaural Synthesis

For enhanced immersion into a virtual scene more than just the visual sense should be addressed by a Virtual Reality system. Additional auditory stimulation appears to have much potential, as it realizes a multisensory system. This is especially useful when the user does not have to wear any additional hardware, e.g., headphones. Creating a virtual sound scene with spatially distributed sources requires a technique for adding spatial cues to audio signals and an appropriate reproduction. In this paper we present a real-time audio rendering system that combines dynamic crosstalk cancellation and multi-track binaural synthesis for virtual acoustical imaging. This provides the possibility of simulating spatially distributed sources and, in addition to that, near-to-head sources for a freely moving listener in room-mounted virtual environments without using any headphones. A special focus will be put on near-to-head acoustics, and requirements in respect of the head-related transfer function databases are discussed.

Long-term safety and feasibility of three-vessel multimodality intravascular imaging in patients with ST-elevation myocardial infarction: the IBIS-4 (integrated biomarker and imaging study) substudy.

We assessed the feasibility and the procedural and long-term safety of intracoronary (i.c) imaging for documentary purposes with optical coherence tomography (OCT) and intravascular ultrasound (IVUS) in patients with acute ST-elevation myocardial infarction (STEMI) undergoing primary PCI in the setting of IBIS-4 study. IBIS4 (NCT00962416) is a prospective cohort study conducted at five European centers including 103 STEMI patients who underwent serial three-vessel coronary imaging during primary PCI and at 13 months. The feasibility parameter was successful imaging, defined as the number of pullbacks suitable for analysis. Safety parameters included the frequency of peri-procedural complications, and major adverse cardiac events (MACE), a composite of cardiac death, myocardial infarction (MI) and any clinically-indicated revascularization at 2 years. Clinical outcomes were compared with the results from a cohort of 485 STEMI patients undergoing primary PCI without additional imaging. Imaging of the infarct-related artery at baseline (and follow-up) was successful in 92.2 % (96.6 %) of patients using OCT and in 93.2 % (95.5 %) using IVUS. Imaging of the non-infarct-related vessels was successful in 88.7 % (95.6 %) using OCT and in 90.5 % (93.3 %) using IVUS. Periprocedural complications occurred <2.0 % of OCT and none during IVUS. There were no differences throughout 2 years between the imaging and control group in terms of MACE (16.7 vs. 13.3 %, adjusted HR1.40, 95 % CI 0.77-2.52, p = 0.27). Multi-modality three-vessel i.c. imaging in STEMI patients undergoing primary PCI is consistent a high degree of success and can be performed safely without impact on cardiovascular events at long-term follow-up.

Theoretical foundations of neural networks

Neural networks have often been motivated by superficial analogy with biological nervous systems. Recently, however, it has become widely recognised that the effective application of neural networks requires instead a deeper understanding of the theoretical foundations of these models. Insight into neural networks comes from a number of fields including statistical pattern recognition, computational learning theory, statistics, information geometry and statistical mechanics. As an illustration of the importance of understanding the theoretical basis for neural network models, we consider their application to the solution of multi-valued inverse problems. We show how a naive application of the standard least-squares approach can lead to very poor results, and how an appreciation of the underlying statistical goals of the modelling process allows the development of a more general and more powerful formalism which can tackle the problem of multi-modality.

Parametric study of femtosecond inscription of microstructures for OCT artefact fabrication

As optical coherence tomography (OCT) becomes widespread, validation and characterization of systems becomes important. Reference standards are required to qualitatively and quantitatively measure the performance between difference systems. This would allow the performance degradation of the system over time to be monitored. In this report, the properties of the femtosecond inscribed structures from three different systems for making suitable OCT characterization artefacts (phantoms) are analyzed. The parameter test samples are directly inscribed inside transparent materials. The structures are characterized using an optical microscope and a swept-source OCT. The high reproducibility of the inscribed structures shows high potential for producing multi-modality OCT calibration and characterization phantoms. Such that a single artefact can be used to characterize multiple performance parameters such the resolution, linearity, distortion, and imaging depths. © 2012 SPIE.

A multimedia indexing and retrieval framework for multimedia database systems

The main challenges of multimedia data retrieval lie in the effective mapping between low-level features and high-level concepts, and in the individual users' subjective perceptions of multimedia content. ^ The objectives of this dissertation are to develop an integrated multimedia indexing and retrieval framework with the aim to bridge the gap between semantic concepts and low-level features. To achieve this goal, a set of core techniques have been developed, including image segmentation, content-based image retrieval, object tracking, video indexing, and video event detection. These core techniques are integrated in a systematic way to enable the semantic search for images/videos, and can be tailored to solve the problems in other multimedia related domains. In image retrieval, two new methods of bridging the semantic gap are proposed: (1) for general content-based image retrieval, a stochastic mechanism is utilized to enable the long-term learning of high-level concepts from a set of training data, such as user access frequencies and access patterns of images. (2) In addition to whole-image retrieval, a novel multiple instance learning framework is proposed for object-based image retrieval, by which a user is allowed to more effectively search for images that contain multiple objects of interest. An enhanced image segmentation algorithm is developed to extract the object information from images. This segmentation algorithm is further used in video indexing and retrieval, by which a robust video shot/scene segmentation method is developed based on low-level visual feature comparison, object tracking, and audio analysis. Based on shot boundaries, a novel data mining framework is further proposed to detect events in soccer videos, while fully utilizing the multi-modality features and object information obtained through video shot/scene detection. ^ Another contribution of this dissertation is the potential of the above techniques to be tailored and applied to other multimedia applications. This is demonstrated by their utilization in traffic video surveillance applications. The enhanced image segmentation algorithm, coupled with an adaptive background learning algorithm, improves the performance of vehicle identification. A sophisticated object tracking algorithm is proposed to track individual vehicles, while the spatial and temporal relationships of vehicle objects are modeled by an abstract semantic model. ^