Amir HajiRassouliha

Subpixel Measurement of Living Skin Deformation Using Intrinsic Features

Accurate measurement of skin deformation is essential to study and understand its behaviour under mechanical load. Digital image correlation (DIC) techniques are commonly used for in-vivo subpixel measurements of deformation using camera-based devices. However, most of the existing DIC methods have modest accuracy and require the addition of feature-rich textures in order to measure the deformations. These limitations have made it challenging to measure skin deformations using DIC, especially where the skin does not have a rich texture and high measurement accuracies are required. Recently, an accurate and robust algorithm, named phase-based Savitzky–Golay gradient correlation (P-SG-GC), has been proposed for subpixel image registration. This algorithm addresses many of the limitations of existing DIC algorithms, and its advantages could lead to new advances in measuring skin deformations. In this paper, we test the accuracy and applicability of P-SG-GC for measuring subpixel deformations of living skin.

Motion Correction Using Subpixel Image Registration

Several methods have been proposed to correct motion in medical and non-medical applications, such as optical flow measurements, particle filter tracking, and image registration. In this paper, we designed experiments to test the accuracy and robustness of a recently proposed algorithm for subpixel image registration. In this case, the algorithm is used to correct the relative motion of the object and camera in pairs of images. This recent algorithm (named phase-based Savitzky-Golay gradient-correlation (P-SG-GC)) can achieve very high accuracies in finding synthetically applied translational shifts.

Phonated speech reconstruction using twin mapping models

Computational speech reconstruction algorithms have the ultimate aim of returning natural sounding speech to aphonic and dysphonic individuals. These algorithms can also be used by unimpaired speakers for communicating sensitive or private information. When the glottis loses function due to disease or surgery, aphonic and dysphonic patients retain the power of vocal tract modulation to some degree but they are unable to speak anything more than hoarse whispers without prosthetic aid. While whispering can be seen as a natural and secondary aspect of speech communications for most people, it becomes the primary mechanism of communications for those who have impaired voice production mechanisms, such as laryngectomees. In this paper, by considering the current limitations of speech reconstruction methods, a novel algorithm for converting whispers to normal speech is proposed and the efficiency of the algorithm is discussed. The proposed algorithm relies upon twin mapping models and makes use of artificially generated whispers (called whisperised speech) to regenerate natural phonated speech from whispers. Through a training-based approach, the mapping models exploit whisperised speech to overcome frame to frame time alignment problem in the speech reconstruction process.

Surface deformation tracking and modeling of soft materials

Characterizing the mechanical properties of skin may lead to improvements in surgical scarring, burns treatments, artificial skin science, and disease detection. We present a method of validating a phase-based crosscorrelation method of material point tracking, used to measure surface deformations in soft tissues, using a silicone gel phantom. Tracking of a high spatial-resolution speckle pattern was validated using independent fluorescent microsphere markers. A finite element mesh was deformed according to the tracked speckle pattern, and used to predict the location of the markers. Predictions of microsphere location were compared to stereo-reconstructions. Under a 2900 μm indentation, markers under rms displacements of 125 μm produced a discrepancy between prediction and reconstruction of 23 μm. The same deformation conditions were used to illustrate the use of surface tracking for identifying mechanical properties. A force-driven finite element mesh, using a Neo-Hookean constitutive model, reproduced the surface deformation with an rms error of 172 μm.

A Low-cost, hand-held stereoscopic device for measuring dynamic deformations of skin in vivo

Measuring the deformation of skin in vivo is useful in a number of applications. For example, the response of skin to a variety of mechanical loadings can provide information about the health of the underlying tissue. A number of devices have been developed for measuring the surface deformation of in vivo skin. However, existing devices are typically incapable of covering large areas of skin, or are expensive. To address these issues, we present the design and evaluation of a hand-held low-cost stereoscopic device for in vivo measurement of the dynamic surface deformation of skin. A camera rig with sufficient mechanical strength was designed to hold four high-speed synchronised cameras. The field of view (FOV) of the cameras is approximately 20 mm × 20 mm. A sample 2D deformation measurement of the surface of skin was performed to show the application of this device.

3D surface profiling using arbitrarily positioned cameras

3D surface measurements are important for studying the biomechanical properties of deformable tissues. For 3D surface profiling and reconstruction, corresponding points on an object should be matched in different camera views. This process is traditionally performed in systems that use stereo camera pairs or multiple cameras with aligned optical axes. To measure the deformation in soft tissues, it may be more appropriate to arbitrarily position the cameras. For instance, cameras can be placed to overcome obstructions that may be caused by measurement apparatus, such as a surface indenter. A truly arbitrary placement system requires the development of a new algorithm for finding corresponding points during surface reconstruction, as existing methods cannot handle large incompatibilities due to the perspective effects between rotated camera views. In this study, we have proposed a procedure for feature matching that can be used with arbitrarily positioned cameras. This proposed method is then used to generate a 3D surface profile of a silicone gel phantom.

Suitability of recent hardware accelerators (DSPs, FPGAs, and GPUs) for computer vision and image processing algorithms

Abstract Computer vision and image processing algorithms form essential components of many industrial, medical, commercial, and research-related applications. Modern imaging systems provide high resolution images at high frame rates, and are often required to perform complex computations to process image data. However, in many applications rapid processing is required, or it is important to minimise delays for analysis results. In these applications, central processing units (CPUs) are inadequate, as they cannot perform the calculations with sufficient speed. To reduce the computation time, algorithms can be implemented in hardware accelerators such as digital signal processors (DSPs), field-programmable gate arrays (FPGAs), and graphics processing units (GPUs). However, the selection of a suitable hardware accelerator for a specific application is challenging. Numerous families of DSPs, FPGAs, and GPUs are available, and the technical differences between various hardware accelerators make comparisons difficult. It is also important to know what speed can be achieved using a specific hardware accelerator for a particular algorithm, as the choice of hardware accelerator may depend on both the algorithm and the application. The technical details of hardware accelerators and their performance have been discussed in previous publications. However, there are limitations in many of these presentations, including: inadequate technical details to enable selection of a suitable hardware accelerator; comparisons of hardware accelerators at two different technological levels; and discussion of old technologies. To address these issues, we introduce and discuss important considerations when selecting suitable hardware accelerators for computer vision and image processing tasks, and present a comprehensive review of hardware accelerators. We discuss the practical details of chip architectures, available tools and utilities, development time, and the relative advantages and disadvantages of using DSPs, FPGAs, and GPUs. We provide practical information about state-of-the-art DSPs, FPGAs, and GPUs as well as examples from the literature. Our goal is to enable developers to make a comprehensive comparison between various hardware accelerators, and to select a hardware accelerator that is most suitable for their specific application.

Towards a Real-Time Full-Field Stereoscopic Imaging System for Tracking Lung Surface Deformation Under Pressure Controlled Ventilation

The normal decline in lung function that occurs with age is virtually indistinguishable from early disease, leading to frequent misdiagnosis in the elderly. Computational modelling promises to be a useful tool for improving our understanding of lung mechanics. However, there is currently no unified structure-function computational model that explains how age-dependent structural changes translate to decline in whole lung function. Furthermore, existing models suffer from weak parameterisation due to lack of available data. To begin addressing this issue, we have developed a real-time full-field stereoscopic imaging system for tracking surface deformation of the rat lung during pressure-controlled ventilation. The system will enable the acquisition of novel physiological data on lung tissue mechanics. This study presents preliminary lung surface tracking results from experiments on Sprague-Dawley rats under pressure controlled ventilation. This rich data will provide us with previously unavailable information for constructing and validating more realistic computational models of the lung to help us better understand the mechanisms behind decline in lung function with aging and help guide the development of new diagnostic methods to distinguish age from lung disease.

Quantifying Carotid Pulse Waveforms Using Subpixel Image Registration

Cardiovascular diseases are a common cause of death. Symptoms of cardiovascular disease often arise at a stage of the disease where treatments are ineffective. Hence, methods that can help early diagnosis of heart problems are essential for preventing heart failure. Assessing the shape of the carotid artery waveforms is one of the methods that clinicians use to diagnose heart and valvular diseases, such as hypertrophic obstructive cardiomyopathy, aortic stenosis, and aortic regurgitation. The carotid artery waveforms may be estimated using pulsed-Doppler ultrasound devices or quantified using catheterisation. However, both of these solutions have limitations. Currently, among available solutions, there is no inexpensive, non-invasive objective method, or diagnostic tool for estimating or quantifying the carotid waveforms. To address these limitations, we have designed a portable non-contact camera-based device to quantify the carotid arterial waveforms. The proposed device calculates the vessel-induced deformation of skin from videos taken from the neck to estimate the carotid artery pressure waveforms. This device takes advantage of our precise and sensitive subpixel image registration algorithm to measure skin deformations from sequential frames of the videos. The skin deformations obtained using our device were compared against a laser displacement measurement device with a resolution of 0.2 μm, and a correlation score of 0.95 was achieved for five subjects.

A training-based speech regeneration approach with cascading mapping models

Computational speech reconstruction algorithms have the ultimate aim of returning natural sounding speech to aphonic and dysphonic patients as well as those who can only whisper. In particular, individuals who have lost glottis function due to disease or surgery, retain the power of vocal tract modulation to some degree but they are unable to speak anything more than hoarse whispers without prosthetic aid. While whispering can be seen as a natural and secondary aspect of speech communications for most people, it becomes the primary mechanism of communications for those who have impaired voice production mechanisms, such as laryngectomees. In this paper, by considering the current limitations of speech reconstruction methods, a novel algorithm for converting whispers to normal speech is proposed and the efficiency of the algorithm is explored. The algorithm relies upon cascading mapping models and makes use of artificially generated whispers (called whisperised speech) to regenerate natural phonated speech from whispers. Using a training-based approach, the mapping models exploit whisperised speech to overcome frame to frame time alignment problems that are inherent in the speech reconstruction process. This algorithm effectively regenerates missing information in the conventional frameworks of phonated speech reconstruction, and is able to outperform the current state-of-the-art regeneration methods using both subjective and objective criteria.