Olivier Salvado
CSIRO, Digital Productivity Flagship, Faculty Member
Biomedical imaging analysis typically comprises a variety of complex tasks requiring sophisticated algorithms and visualising high dimensional data. The successful integration and deployment of the enabling software to clinical (research)... more
Biomedical imaging analysis typically comprises a variety of complex tasks requiring sophisticated algorithms and visualising high dimensional data. The successful integration and deployment of the enabling software to clinical (research) partners, for rigorous evaluation and testing, is a crucial step to facilitate adoption of research innovations within medical settings. In this paper, we introduce the Simple Medical Imaging Library Interface (SMILI), an object oriented open-source framework with a compact suite of objects geared for rapid biomedical imaging (cross-platform) application development and deployment. SMILI supports the development of both command-line (shell and Python scripting) and graphical applications utilising the same set of processing algorithms. It provides a substantial subset of features when compared to more complex packages, yet it is small enough to ship with clinical applications with limited overhead and has a license suitable for commercial use. Afte...
Research Interests:
The centiloid scale was recently proposed to provide a standard framework for the quantification of β-amyloid PET images, so that amyloid burden can be expressed on a standard scale. While the framework prescribes SPM8 as the standard... more
The centiloid scale was recently proposed to provide a standard framework for the quantification of β-amyloid PET images, so that amyloid burden can be expressed on a standard scale. While the framework prescribes SPM8 as the standard analysis method for PET quantification, non-standard methods can be calibrated to produce centiloid values. We have previously developed a PET-only quantification: CapAIBL. In this study, we show how CapAIBL can be calibrated to the centiloid scale. Calibration images for C-PiB, F-NAV4694, F-Florbetaben, F-Flutemetamol and F- Florbetapir were analysed using the standard method and CapAIBL. Using these images, both methods were calibrated to the centiloid scale. Centiloid values computed using CapAIBL were compared to those computed using standard method. For each tracer, a separate validation was performed using an independent dataset from the AIBL study. Using the calibration images, there was a very strong agreement, and very little bias between the ...
Research Interests:
ABSTRACT This paper presents a validation study on the application of a novel interslice interpolation technique for musculoskeletal structure segmentation of articulated joints and muscles on human magnetic resonance imaging data. The... more
ABSTRACT This paper presents a validation study on the application of a novel interslice interpolation technique for musculoskeletal structure segmentation of articulated joints and muscles on human magnetic resonance imaging data. The interpolation technique is based on morphological shape-based interpolation combined with intensity based voxel classification. Shape-based interpolation in the absence of the original intensity image has been investigated intensively. However, in some applications of medical image analysis, the intensity image of the slice to be interpolated is available. For example, when manual segmentation is conducted on selected slices, the segmentation on those unselected slices can be obtained by interpolation. We proposed a two- step interpolation method to utilize both the shape information in the manual segmentation and local intensity information in the image. The method was tested on segmentations of knee, hip and shoulder joint bones and hamstring muscles. The results were compared with two existing interpolation methods. Based on the calculated Dice similarity coefficient and normalized error rate, the proposed method outperformed the other two methods.
Research Interests:
There is increasing evidence that educating trainee surgeons by simulation is preferable to traditional operating-room training methods with actual patients. Apart from reducing costs and risks to patients, training by simulation can... more
There is increasing evidence that educating trainee surgeons by simulation is preferable to traditional operating-room training methods with actual patients. Apart from reducing costs and risks to patients, training by simulation can provide some unique benefits, such as greater control over the training procedure and more easily defined metrics for assessing proficiency. Virtual reality (VR) simulators are now playing an increasing role in surgical training. However, currently available VR simulators lack the fidelity to teach trainees past the novice-to-intermediate skills level. Recent technological developments in other industries using simulation, such as the games and entertainment and aviation industries, suggest that the next generation of VR simulators should be suitable for training, maintenance and certification of advanced surgical skills. To be effective as an advanced surgical training and assessment tool, VR simulation needs to provide adequate and relevant levels of ...
Research Interests:
ABSTRACT
Research Interests:
APOEɛ4 genotype and aging have been identified as risk factors for Alzheimer's disease (AD). In addition, subjective memory complaints (SMC) might be a first clinical expression of the effect of AD pathology on cognitive functioning.... more
APOEɛ4 genotype and aging have been identified as risk factors for Alzheimer's disease (AD). In addition, subjective memory complaints (SMC) might be a first clinical expression of the effect of AD pathology on cognitive functioning. To assess whether APOEɛ4 genotype, age, SMC, and episodic memory are risk factors for high amyloid-β (Aβ) burden in cognitively normal elderly. 307 cognitively normal participants (72.7 ± 6.8 years, 53% female, 55% SMC) from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study underwent amyloid PET and APOE genotyping. Logistic regression analyses were performed to determine the association of APOEɛ4 genotype, age, SMC, and episodic memory with Aβ pathology. Odds of high Aβ burden were greater at an older age (OR = 3.21; 95% CI = 1.68-6.14), when SMC were present (OR = 1.90; 95% CI = 1.03-3.48), and for APOEɛ4 carriers (OR = 7.49; 95% CI = 3.96-14.15), while episodic memory was not associated with odds of high Aβ burden. Stratified analyses...
Research Interests:
Non-local means is a recently proposed denoising technique that better preserves image structures than other methods. However, the computational cost of non-local means is prohibitive, especially for large 3D images. Modifications have... more
Non-local means is a recently proposed denoising technique that better preserves image structures than other methods. However, the computational cost of non-local means is prohibitive, especially for large 3D images. Modifications have previously been proposed to reduce the cost, which result in image artefacts. This paper proposes a compact rotation invariant descriptor. Testing demonstrates improved denoising performance relative to optimized non-local means. Rotation invariant non-local means is an order of magnitude faster.
Research Interests:
Susceptibility-weighted imaging (SWI) is recognized as the preferred MRI technique for visualizing cerebral vasculature and related pathologies such as cerebral microbleeds (CMBs). Manual identification of CMBs is time-consuming, has... more
Susceptibility-weighted imaging (SWI) is recognized as the preferred MRI technique for visualizing cerebral vasculature and related pathologies such as cerebral microbleeds (CMBs). Manual identification of CMBs is time-consuming, has limited reliability and reproducibility, and is prone to misinterpretation. In this paper, a novel computer-aided microbleed detection technique based on machine learning is presented: First, spherical-like objects (potential CMB candidates) with their corresponding bounding boxes were detected using a novel multi-scale Laplacian of Gaussian technique. A set of robust 3-dimensional Radon- and Hessian-based shape descriptors within each bounding box were then extracted to train a cascade of binary random forests (RF). The cascade consists of consecutive independent RF classifiers with low to high posterior probability constraints to handle imbalanced training sets (CMBs and non-CMBs), and to progressively improve detection rates. The proposed method was validated on 66 subjects whose CMBs were manually stratified into "possible" and "definite" by two medical experts. The proposed technique achieved a sensitivity of 87% and an average false detection rate of 27.1 CMBs per subject on the "possible and definite" set. A sensitivity of 93% and false detection rate of 10 CMBs per subject was also achieved on the "definite" set. The proposed automated approach outperforms state of the art methods, and promises to enhance manual expert screening. Benefits include improved reliability, minimization of intra-rater variability and a reduction in assessment time.
Research Interests: Algorithms, Biomedical Engineering, Machine Learning, Humans, Computer Simulation, and 8 moreClinical Sciences, Image Enhancement, Observer Variation, Reproducibility of Results, Sensitivity and Specificity, Cerebral Hemorrhage, Magnetic resonance angiography, and Diffusion magnetic resonance imaging
Structural MRI is widely used for investigating brain atrophy in many neurodegenerative disorders, with several research groups developing and publishing techniques to provide quantitative assessments of this longitudinal change. Often... more
Structural MRI is widely used for investigating brain atrophy in many neurodegenerative disorders, with several research groups developing and publishing techniques to provide quantitative assessments of this longitudinal change. Often techniques are compared through computation of required sample size estimates for future clinical trials. However interpretation of such comparisons is rendered complex because, despite using the same publicly available cohorts, the various techniques have been assessed with different data exclusions and different statistical analysis models. We created the MIRIAD atrophy challenge in order to test various capabilities of atrophy measurement techniques. The data consisted of 69 subjects (46 Alzheimer's Disease, 23 control) who were scanned multiple (up to twelve) times at nine visits over a follow up period of one to two years, resulting in 708 total image sets. Nine participating groups from 6 countries completed the challenge by providing volumetric measurements of key structures (whole brain, lateral ventricle, left and right hippocampus) for each data set and atrophy measurements of these structures for each time point pair (both forward and backward) of a given subject. From these results, we formally compared techniques using exactly the same dataset. First, we assessed the repeatability of each technique using rates obtained from short intervals where no measurable atrophy is expected. For those measures that provided direct measures of atrophy between pairs of images, we also assessed symmetry and transitivity. Then, we performed a statistical analysis in a consistent manner using linear mixed effect models. The models, one for repeated measures of volume made at multiple time-points and a second for repeated…
Research Interests:
Research Interests:
Longer life expectancies lead to increases in the prevalence of age-associated illnesses. The number of Australians with dementia is predicted to rise, from 234,000 in 2009 to over 1 million by 2050, as a result of the increased... more
Longer life expectancies lead to increases in the prevalence of age-associated illnesses. The number of Australians with dementia is predicted to rise, from 234,000 in 2009 to over 1 million by 2050, as a result of the increased prevalence of Alzheimer's disease (AD), the leading cause of dementia in the elderly. Early diagnosis of AD will become more important as disease-modifying therapies emerge within the next decade. Advances in molecular neuroimaging with amyloid-β-specific radioligands for positron emission tomography, aided by magnetic resonance imaging techniques, allow detection of AD years before symptoms of dementia develop. Longitudinal prospective studies, such as the Australian Imaging Biomarkers and Lifestyle (AIBL) study of ageing, will determine the sensitivity and specificity of these analysis techniques for diagnosing AD and predicting cognitive decline.
Research Interests:
We are involved in a comprehensive program to characterize atherosclerotic disease using multiple MR images having different contrast mechanisms (T1W, T2W, PDW, magnetization transfer, etc.) of human carotid and animal model arteries. We... more
We are involved in a comprehensive program to characterize atherosclerotic disease using multiple MR images having different contrast mechanisms (T1W, T2W, PDW, magnetization transfer, etc.) of human carotid and animal model arteries. We use specially designed intravascular and surface array coils that give high signal-to-noise but suffer from sensitivity inhomogeneity and significant noise. We present here a new non-parametric method for correcting the images without assumption of the number of different tissues. Intensity inhomogeneity is modeled with cubic spline and is locally optimized using an entropy criterion. Validation has been performed on a specially design neck phantom as well as actual MR scans on patient neck. The steep bias is corrected sufficiently to aid human interpretation of gray scales. It should also make possible computerized tissue classification.
Research Interests:
ABSTRACT Automated detection of small, low level shapes such as circular/spherical objects in images is a challenging computer vision problem. For many applications, especially microbleed detection in Alzheimer's disease, an... more
ABSTRACT Automated detection of small, low level shapes such as circular/spherical objects in images is a challenging computer vision problem. For many applications, especially microbleed detection in Alzheimer's disease, an automatic pre-screening scheme is required to identify potential seeds with high sensitivity and reasonable specificity. A new method is proposed to detect spherical objects in 3D medical images within the multi-scale Laplacian of Gaussian framework. The major contributions are(1)breaking down 3D sphere detection into 1D line profile detection along each coordinate dimension, (2) identifying center of structures bynormalizing the line response profile and (3) employing eigenvalues of the Hessian matrix at optimum scale for the center points to determine spherical objects. The method is validated both on simulated data and susceptibility weighted MRI images with ground truth provided by a medical expert. Validation results demonstrate that the current approach has higher performance in terms of sensitivity and specificity and is effective in detecting adjacent microbleeds, with invariance to intensity, orientation, translation and object scale.
Research Interests:
Abstract We are characterizing atherosclerotic disease in patients and animal models using multiple MR images having different contrasts. We use intravascular and surface array coils giving high signal-to-noise but significant sensitivity... more
Abstract We are characterizing atherosclerotic disease in patients and animal models using multiple MR images having different contrasts. We use intravascular and surface array coils giving high signal-to-noise but significant sensitivity inhomogeneity. In human carotid ...
Research Interests:
ABSTRACT We are involved in a comprehensive program to characterize atherosclerotic disease using multiple MR images having different contrast mechanisms (T1W, T2W, PDW, magnetization transfer, etc.) of human carotid and animal model... more
ABSTRACT We are involved in a comprehensive program to characterize atherosclerotic disease using multiple MR images having different contrast mechanisms (T1W, T2W, PDW, magnetization transfer, etc.) of human carotid and animal model arteries. We use specially designed intravascular and surface array coils that give high signal-to-noise but suffer from sensitivity inhomogeneity. With carotid surface coils, challenges include: (1) a steep bias field with an 80% change; (2) presence of nearby muscular structures lacking high frequency information to distinguish bias from anatomical features; (3) many confounding zero-valued voxels subject to fat suppression, blood flow cancellation, or air, which are not subject to coil sensitivity; and (4) substantial noise. Bias was corrected using a modification of the adaptive fuzzy c-mean method reported by Pham et al. (IEEE TMI, 18:738-752), whereby a bias field modeled as a mechanical membrane was iteratively improved until cluster means no longer changed. Because our images were noisy, we added a noise reduction filtering step between iterations and used about 5 classes. In a digital phantom having a bias field measured from our MR system, variations across an area comparable to a carotid artery were reduced from 50% to
Research Interests:
Research Interests:
Autofluorescence (AF) techniques improve the diagnostic yield of white light inspection for preneoplastic lesions in the bronchus and head and neck region. Although highly sensitive, AF has poor specificity, particularly in situations... more
Autofluorescence (AF) techniques improve the diagnostic yield of white light inspection for preneoplastic lesions in the bronchus and head and neck region. Although highly sensitive, AF has poor specificity, particularly in situations where there have been earlier biopsies or treatments such as radiotherapy. Narrow band imaging (NBI) is a newer imaging technique that enhances the early abnormal angiogenesis seen in preneoplastic lesions. NBI has higher specificity when compared with AF. We aimed to combine these imaging modalities, using AF as an effective screening tool and NBI to confirm AF findings. We also used computer-assisted image analysis techniques to give objective confirmation to our visual inspection. Three patients were selected for image analysis of their NBI images using the L*a*b* color scale in manually drawn regions of interest of biopsy-confirmed areas. Each case compared pathology with a different benign condition: normal tissue, postbiopsy effect, and postradiation therapy change. Patients had white light followed by AF inspection. Abnormal areas of AF were cross-examined with NBI. NBI clearly showed dysplasia and carcinoma in situ. It also confirmed abnormal fluorescence because of earlier biopsies and radiation therapy. Analysis of the L*a*b* color space scale in each case showed segmentation between pathology and the benign tissue. There may be additive and discriminatory benefits of NBI after AF inspection. Further study with computer-assisted color segmentation techniques and image analysis is required before optical diagnosis can become a reality in bronchoscopic techniques in the future.
Research Interests:
ABSTRACT
Research Interests:
Research Interests:
ABSTRACT Many medical images suffer from the partial volume effect where a boundary between two structures of interest falls in the middle of a voxel giving a signal value that is a mixture of the two. We propose a method to restore the... more
ABSTRACT Many medical images suffer from the partial volume effect where a boundary between two structures of interest falls in the middle of a voxel giving a signal value that is a mixture of the two. We propose a method to restore the ideal boundary by splitting a voxel into sub-voxels and reapportioning the signal into the sub-voxels. We designed this method to correct MRI 2D slice images where partial volume can be a considerable limitation. Each voxel is divided into four (or more) sub-voxels by nearest neighbor interpolation. The gray level of each sub-voxel is considered as "materials" able to move between sub-voxels but not between voxels. A partial differential equation is written to allow the material to flow towards the highest gradient direction, creating a "reverse" diffusion process. Flow is subject to constraints that tend to create step edges. Material is conserved in the process thereby conserving MR signal. The method proceeds until the flow decreases to a low value. To test the method, synthetic images were down-sampled to simulate the partial volume artifact and restored. Corrected images were remarkably closer both visually and quantitatively to the original images than those obtained from common interpolation methods: on simulated data mean square errors were 0.35, 1.09, and 1.24 for the proposed method, bicubic, and bilinear interpolation respectively. The method was relatively insensitive to noise. On MRI physical phantom and brain images, restored images processed with the new method were visually much closer to high-resolution counter-parts than those obtained with common interpolation methods.
Research Interests:
ABSTRACT Non-local means is a recently proposed denoising technique that better preserves image structures than other methods. However, the computational cost of non-local means is prohibitive, especially for large 3D images.... more
ABSTRACT Non-local means is a recently proposed denoising technique that better preserves image structures than other methods. However, the computational cost of non-local means is prohibitive, especially for large 3D images. Modifications have previously been proposed to reduce the cost, which result in image artefacts. This paper proposes a compact rotation invariant descriptor. Testing demonstrates improved denoising performance relative to optimized non-local means. Rotation invariant non-local means is an order of magnitude faster.
Research Interests:
ABSTRACT
Research Interests:
We are involved in a comprehensive program to characterize atherosclerotic disease using multiple MR images having different contrast mechanisms (T1W, T2W, PDW, magnetization transfer, etc.) of human carotid and animal model arteries. We... more
We are involved in a comprehensive program to characterize atherosclerotic disease using multiple MR images having different contrast mechanisms (T1W, T2W, PDW, magnetization transfer, etc.) of human carotid and animal model arteries. We use specially designed intravascular and surface array coils that give high signal-to-noise but suffer from sensitivity inhomogeneity and significant noise. We present here a new non-parametric method for correcting the images without assumption of the number of different tissues. Intensity inhomogeneity is modeled with cubic spline and is locally optimized using an entropy criterion. Validation has been performed on a specially design neck phantom as well as actual MR scans on patient neck. This same algorithm has been successfully applied to the correction of very high resolution, intravascular coil images. The steep bias is corrected sufficiently to aid human interpretation of gray scales. It should also make possible computerized tissue classification.
Research Interests:
We are investigating methods to reduce partial volume effect (PVE) in medical images acquired as a series of thick slices, in particular magnetic resonance imaging (MRl) of the human neck for atherosclerosis characterization. We extend... more
We are investigating methods to reduce partial volume effect (PVE) in medical images acquired as a series of thick slices, in particular magnetic resonance imaging (MRl) of the human neck for atherosclerosis characterization. We extend the reverse diffusion algorithm to 3D and include a smoothing term to increase the robustness to noise. Evaluation on a synthetic data set showed a
Research Interests:
Intensity based registration (e.g., mutual information) suffers from a scalloping artifact giving rise to local maxima and sometimes a biased global maximum in a similarity objective function. Here, we demonstrate that scalloping is... more
Intensity based registration (e.g., mutual information) suffers from a scalloping artifact giving rise to local maxima and sometimes a biased global maximum in a similarity objective function. Here, we demonstrate that scalloping is principally due to the noise reduction filtering that occurs when image samples are interpolated. Typically at a much smaller scale (100 times less in our test cases), there are also fluctuations in the similarity objective function due to interpolation of the signal and to sampling of a continuous, band-limited image signal. Focusing on the larger problem from noise, we show that this phenomenon can even bias global maxima, giving inaccurate registrations. This phenomenon is readily seen when one registers an image onto itself with different noise realizations but is absent when the same noise realization is present in both images. For linear interpolation, local maxima and global bias are removed if one filters the interpolated image using a new constant variance filter for linear interpolation (cv-lin filter), which equalizes the variance across the interpolated image. We use 2D synthetic and MR images and characterize the effect of cv-lin on similarity objective functions. With a reduction of local and biased maxima, image registration becomes more robust and accurate. An efficient implementation adds insignificant computation time per iteration, and because optimization proceeds more smoothly, sometimes fewer iterations are needed.
Research Interests:
Research Interests:
Research Interests: Information Systems, Algorithms, Artificial Intelligence, Data Compression, 3-D Imaging, and 11 moreImage Filtering, Three Dimensional Imaging, Computational Efficiency, Artifacts, Image Enhancement, Three Dimensional, Reproducibility of Results, Sensitivity and Specificity, Electrical And Electronic Engineering, Hash Function, and Linear Interpolation
Research Interests: Engineering, Algorithms, Artificial Intelligence, Brain Imaging, Fuzzy Logic, and 15 moreFuzzy set theory, Image segmentation, Entropy, Anisotropy, Image Classification, Humans, Artifacts, Information Storage and Retrieval, Image Enhancement, Coronary Artery Disease, Linear Filtering, Carotid Artery, Field of View, Blood Vessel, and Magnetic resonance image
The aim of the present study was to compare the maximal isometric torque and cardio-respiratory parameters in well-trained young and master triathletes prior to and following an Olympic distance triathlon. One day before and 24 h... more
The aim of the present study was to compare the maximal isometric torque and cardio-respiratory parameters in well-trained young and master triathletes prior to and following an Olympic distance triathlon. One day before and 24 h following the event, participants performed three maximum voluntary isometric knee extensions and flexions and an incremental running test on a treadmill to determine the maximal isometric torque, maximal oxygen uptake VO(2max), speed at VO(2max) (vVO(2)max), speed at ventilatory thresholds (VT1 and VT2) and submaximal running economy. Prior to the event VO(2max), vVO(2)max, speed at ventilatory thresholds and running economy were significantly lower in master athletes, but maximal voluntary torque was similar between the groups. 24 h following the race, a similar significant decrease in VO(2max) (-3.1% in masters, and -6.2% in young, p < 0.05), and vVO(2)max (-9.5% in masters, and -5.6% in young, p < 0.05) was observed in both the groups. The speed at VT2 significantly decreased only in master athletes (-8.3%, p < 0.05), while no change was recorded in maximal voluntary torque or submaximal running economy following the event. The results indicate that for well-trained subjects, the overall relative exercise intensity during an Olympic distance triathlon and the fatigue 24 h following the event seem to be independent of age.
Research Interests: Aging, France, Humans, Male, Heart rate, and 15 moreExercise, European, Aged, Middle Aged, Adult, Analysis of Variance, Age Factors, Exercise Test, Oxygen Consumption, Maximal Oxygen Uptake, Knee Extension, Maximal Oxygen Consumption, Biomechanical Phenomena, Isometric Contraction, and Physical Endurance
Research Interests:
Research Interests:
ABSTRACT