ORCID Profile
0000-0003-0109-6844
Current Organisations
Queensland University of Technology (QUT)
,
Queensland University of Technology
,
Australian Institute of Company Directors
Does something not look right? The information on this page has been harvested from data sources that may not be up to date. We continue to work with information providers to improve coverage and quality. To report an issue, use the Feedback Form.
Research Topics
In Research Link Australia (RLA), "Research Topics" refer to ANZSRC FOR and SEO codes. These topics are either sourced from ANZSRC FOR and SEO codes listed in researchers' related grants or generated by a large language model (LLM) based on their publications.
Top 5 Research Topics
ANZSRC Field of Research (FoR)
Artificial Intelligence and Image Processing | Image Processing | Pattern Recognition | Pattern Recognition and Data Mining | Computer Vision | Image Processing | Biomedical Engineering not elsewhere classified | Neurosciences Not Elsewhere Classified | Neurogenetics | Learning, Memory, Cognition And Language | Clinical Engineering | Information Systems | Computer-Human Interaction | Computer-Human Interaction | Communications Technologies | Psychology | Interdisciplinary Engineering Not Elsewhere Classified | Antennas and Propagation | Microwave and Millimetrewave Theory and Technology
ANZSRC Socio-Economic Objective (SEO)
Cancer and related disorders | Expanding Knowledge in Engineering | Expanding Knowledge in the Information and Computing Sciences | Women’s health | Medical instrumentation | Computer software and services not elsewhere classified | Education and Training Systems not elsewhere classified | Nervous system and disorders | Health related to ageing | Integrated systems | Ground Transport not elsewhere classified | Expanding Knowledge in the Medical and Health Sciences | Physical sciences | Expanding Knowledge in Technology | Transport Equipment not elsewhere classified | Behavioural and cognitive sciences | Cancer and Related Disorders |
Related Links
Publications
Automated Analysis of Unregistered Multi-View Mammograms With Deep Learning
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 11-2017
Training medical image analysis systems like radiologists
Publisher: Springer International Publishing
Date: 2018
Performance of tumour microenvironment deconvolution methods in breast cancer using single-cell simulated bulk mixtures
Publisher: Springer Science and Business Media LLC
Date: 16-09-2023
A wavelet visible difference predictor
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 05-1999
DOI: 10.1109/83.760338
Abstract: In this paper, we describe a model of the human visual system (HVS) based on the wavelet transform. This model is largely based on a previously proposed model, but has a number of modifications that make it more amenable to potential integration into a wavelet based image compression scheme. These modifications include the use of a separable wavelet transform instead of the cortex transform, the application of a wavelet contrast sensitivity function (CSF), and a simplified definition of subband contrast that allows one to predict the noise visibility directly from the wavelet coefficients. Initially, we outline the luminance, frequency, and masking sensitivities of the HVS and discuss how these can be incorporated into the wavelet transform. We then outline a number of limitations of the wavelet transform as a model of the HVS, namely the lack of translational invariance and poor orientation sensitivity. In order to investigate the efficacy of this wavelet based model, a wavelet visible difference predictor (WVDP) is described. The WVDP is then used to predict visible differences between an original and compressed (or noisy) image. Results are presented to emphasize the limitations of commonly used measures of image quality and to demonstrate the performance of the WVDP. The paper concludes with suggestions on how the WVDP can be used to determine a visually optimal quantization strategy for wavelet coefficients and produce a quantitative measure of image quality.
Combining deep learning and structured prediction for segmenting masses in mammograms
Publisher: Springer International Publishing
Date: 2017
Mass segmentation in mammograms: A cross-sensor comparison of deep and tailored features
Publisher: IEEE
Date: 09-2017
Segmentation of skin strata in reflectance confocal microscopy depth stacks
Publisher: SPIE
Date: 20-03-2015
DOI: 10.1117/12.2081737
How to Exploit Weaknesses in Biomedical Challenge Design and Organization
Publisher: Springer International Publishing
Date: 2018
Automated Segmentation of Skin Strata in Reflectance Confocal Microscopy Depth Stacks
Publisher: Public Library of Science (PLoS)
Date: 18-04-2016
Machine learning for medical diagnostics : techniques for feature extraction, classification, and evaluation
Publisher: University of Queensland Library
New Spatiotemporal Features for Improved Discrimination of Benign and Malignant Lesions in Dynamic Contrast-Enhance-Magnetic Resonance Imaging of the Breast
Publisher: Ovid Technologies (Wolters Kluwer Health)
Date: 09-2011
Extended Naïve Bayes for Group Based Classification
Publisher: Springer International Publishing
Date: 2014
ROC curves and the X2 test
Publisher: Elsevier BV
Date: 03-1996
Automated Nucleus and Cytoplasm Segmentation of Overlapping Cervical Cells
Publisher: Springer Berlin Heidelberg
Date: 2013
DOI: 10.1007/978-3-642-40811-3_57
Abstract: In this paper we describe an algorithm for accurately segmenting the in idual cytoplasm and nuclei from a clump of overlapping cervical cells. Current methods cannot undertake such a complete segmentation due to the challenges involved in delineating cells with severe overlap and poor contrast. Our approach initially performs a scene segmentation to highlight both free-lying cells, cell clumps and their nuclei. Then cell segmentation is performed using a joint level set optimization on all detected nuclei and cytoplasm pairs. This optimisation is constrained by the length and area of each cell, a prior on cell shape, the amount of cell overlap and the expected gray values within the overlapping regions. We present quantitative nuclei detection and cell segmentation results on a database of synthetically overlapped cell images constructed from real images of free-lying cervical cells. We also perform a qualitative assessment of complete fields of view containing multiple cells and cell clumps.
Multi-scale sifting for mammographic mass detection and segmentation
Publisher: IOP Publishing
Date: 22-01-2019
1st MICCAI workshop on deep learning in medical image analysis
Publisher: Informa UK Limited
Date: 05-04-2018
Performance Analysis of Three Microscope Slide Scanning Techniques
Publisher: IEEE
Date: 11-2013
Multi-scale mass segmentation for mammograms via cascaded random forests
Publisher: IEEE
Date: 04-2017
A comparison of DCT and DWT block based watermarking on medical image quality
Publisher: Springer Berlin Heidelberg
Date: 2008
Taking the Confusion Out of Multinomial Confusion Matrices and Imbalanced Classes
Publisher: Springer Singapore
Date: 2021
Automated Mass Detection in Mammograms Using Cascaded Deep Learning and Random Forests
Publisher: IEEE
Date: 11-2016
Use of pulse transit time to distinguish respiratory events from tidal breathing in sleeping children
Publisher: Elsevier BV
Date: 10-2005
Investigation of the inherent left‐right flow balancing of rotary total artificial hearts by means of a resistance box
Publisher: Wiley
Date: 05-02-2020
DOI: 10.1111/AOR.13631
Abstract: With the incidence of end-stage heart failure steadily increasing, the need for a practical total artificial heart (TAH) has never been greater. Continuous flow TAHs (CFTAH) are being developed using rotary blood pumps (RBPs), leveraging their small size, mechanical simplicity, and excellent durability. To completely replace the heart with currently available RBPs, two are required one for providing pulmonary flow and one for providing systemic flow. To prevent hazardous states, it is essential to maintain balance between the pulmonary and systemic circulation at a wide variety of physiologic states. In this study, we investigated factors determining a CFTAH's inherent ability to balance systemic and pulmonary flow passively, without active management of pump rotational speed. Four different RBPs (ReliantHeart HA5, Thoratec HMII, HeartWare HVAD, and Ventracor VentrAssist) were used in various combinations to construct CFTAHs. Each CFTAH's ability to autonomously maintain pressures and flows within defined ranges was evaluated in a hybrid mock loop as systemic and pulmonary vascular resistance (PVR) were changed. The resistance box, a method to quantify the range of vascular resistances that can be safely supported by a CFTAH, was used to compare different CFTAH configurations in an efficient and predictive way. To reduce the need for future in vitro tests and to aid in their analysis, a novel analytical evaluation to predict the resistance box of various CFTAH configurations was also performed. None of the investigated CFTAH configurations fully satisfied the predefined benchmarks for inherent flow balancing, with the VentrAssist (left) and HeartAssist 5 (right) offering the best combination. The extent to which each CFTAH was able to autonomously maintain balance was determined by the pressure sensitivity of each RPB: the sensitivity of outflow to changes in the pressure head. The analytical model showed that by matching left and right pressure sensitivity the inherent balancing performance can be improved. These findings may ultimately lead to a reduced need for manual speed changes or active control systems.
Automated 5-year mortality prediction using deep learning and radiomics features from chest computed tomography
Publisher: IEEE
Date: 04-2017
Physiological Control of Dual Rotary Pumps as a Biventricular Assist Device Using a Master/Slave Approach
Publisher: Wiley
Date: 21-04-2014
DOI: 10.1111/AOR.12303
Abstract: Dual rotary left ventricular assist devices (LVADs) can provide biventricular mechanical support during heart failure. Coordination of left and right pump speeds is critical not only to avoid ventricular suction and to match cardiac output with demand, but also to ensure balanced systemic and pulmonary circulatory volumes. Physiological control systems for dual LVADs must meet these objectives across a variety of clinical scenarios by automatically adjusting left and right pump speeds to avoid catastrophic physiological consequences. In this study we evaluate a novel master/slave physiological control system for dual LVADs. The master controller is a Starling-like controller, which sets flow rate as a function of end-diastolic ventricular pressure (EDP). The slave controller then maintains a linear relationship between right and left EDPs. Both left/right and right/left master/slave combinations were evaluated by subjecting them to four clinical scenarios (rest, postural change, Valsalva maneuver, and exercise) simulated in a mock circulation loop. The controller's performance was compared to constant-rotational-speed control and two other dual LVAD control systems: dual constant inlet pressure and dual Frank-Starling control. The results showed that the master/slave physiological control system produced fewer suction events than constant-speed control (6 vs. 62 over a 7-min period). Left/right master/slave control had lower risk of pulmonary congestion than the other control systems, as indicated by lower maximum EDPs (15.1 vs. 25.2-28.4 mm Hg). During exercise, master/slave control increased total flow from 5.2 to 10.1 L/min, primarily due to an increase of left and right pump speed. Use of the left pump as the master resulted in fewer suction events and lower EDPs than when the right pump was master. Based on these results, master/slave control using the left pump as the master automatically adjusts pump speed to avoid suction and increases pump flow during exercise without causing pulmonary venous congestion.
Quantifiable brain atrophy synthesis for benchmarking of cortical thickness estimation methods
Publisher: Elsevier BV
Date: 11-2022
DOI: 10.1016/J.MEDIA.2022.102576
Abstract: Cortical thickness (CTh) is routinely used to quantify grey matter atrophy as it is a significant biomarker in studying neurodegenerative and neurological conditions. Clinical studies commonly employ one of several available CTh estimation software tools to estimate CTh from brain MRI scans. In recent years, machine learning-based methods emerged as a faster alternative to the main-stream CTh estimation methods (e.g. FreeSurfer). Evaluation and comparison of CTh estimation methods often include various metrics and downstream tasks, but none fully covers the sensitivity to sub-voxel atrophy characteristic of neurodegeneration. In addition, current evaluation methods do not provide a framework for the intra-method region-wise evaluation of CTh estimation methods. Therefore, we propose a method for brain MRI synthesis capable of generating a range of sub-voxel atrophy levels (global and local) with quantifiable changes from the baseline scan. We further create a synthetic test set and evaluate four different CTh estimation methods: FreeSurfer (cross-sectional), FreeSurfer (longitudinal), DL+DiReCT and HerstonNet. DL+DiReCT showed superior sensitivity to sub-voxel atrophy over other methods in our testing framework. The obtained results indicate that our synthetic test set is suitable for benchmarking CTh estimation methods on both global and local scales as well as regional inter-and intra-method performance comparison.
Anatomical Skin Segmentation in Reflectance Confocal Microscopy with Weak Labels
Publisher: IEEE
Date: 11-2015
GLOBALLY OPTIMAL BREAST MASS SEGMENTATION FROM DCE-MRI USING DEEP SEMANTIC SEGMENTATION AS SHAPE PRIOR
Publisher: IEEE
Date: 04-2017
Correction to: Taking the Confusion Out of Multinomial Confusion Matrices and Imbalanced Classes
Publisher: Springer Singapore
Date: 09-12-2021
Illumination effects in quantitative virtual microscopy
Publisher: Springer Berlin Heidelberg
Date: 2013
The use of the area under the ROC curve in the evaluation of machine learning algorithms
Publisher: Elsevier BV
Date: 07-1997
A Hybrid Mock Circulation Loop for a Total Artificial Heart
Publisher: Wiley
Date: 09-2014
DOI: 10.1111/AOR.12380
Abstract: Rotary blood pumps are emerging as a viable technology for total artificial hearts, and the development of physiological control algorithms is accelerated with new evaluation environments. In this article, we present a novel hybrid mock circulation loop (HMCL) designed specifically for evaluation of rotary total artificial hearts (rTAH). The rTAH is operated in the physical domain while all vasculature elements are embedded in the numerical domain, thus combining the strengths of both approaches: fast and easy exchange of the vasculature model together with improved controllability of the pump. Parameters, such as vascular resistance, compliance, and blood volume, can be varied dynamically in silico during operation. A hydraulic-numeric interface creates a real-time feedback loop between the physical and numerical domains. The HMCL uses computer-controlled resistance valves as actuators, thereby reducing the size and number of hydraulic elements. Experimental results demonstrate a stable interaction over a wide operational range and a high degree of flexibility. Therefore, we demonstrate that the newly created design environment can play an integral part in the hydraulic design, control development, and durability testing of rTAHs.
Deep Reinforcement Learning for Detecting Breast Lesions from DCE-MRI
Publisher: Springer International Publishing
Date: 2019
FULLY AUTOMATED CLASSIFICATION OF MAMMOGRAMS USING DEEP RESIDUAL NEURAL NETWORKS
Publisher: IEEE
Date: 04-2017
Multiple Instance Learning for Breast Cancer Magnetic Resonance Imaging
Publisher: IEEE
Date: 11-2014
Modeling of a Rotary Blood Pump
Publisher: Wiley
Date: 08-2013
DOI: 10.1111/AOR.12142
Abstract: The accurate representation of rotary blood pumps in a numerical environment is important for meaningful investigation of pump-cardiovascular system interactions. Although numerous models for ventricular assist devices (VADs) have been developed, modeling methods for rotary total artificial hearts (rTAHs) are still required. Therefore, an rTAH prototype was characterized in a steady flow, hydraulic test bench over a wide operational range for pump and hydraulic parameters. In order to develop a generic modeling method, a data-driven modeling approach was chosen. k-Nearest-neighbors, artificial neural networks, and support vector machines (SVMs) were the machine learning approaches evaluated. The best performing parameters for each algorithm were determined via optimization. The resulting multiple-input-multiple-output models were subsequently assessed under identical conditions, and a SVM with a radial basis function kernel was identified as the best performing. The achieved root mean squared errors were 0.03 L/min, 0.06 L/min, and 0.18 W for left and right flow and motor power consumption, respectively. In comparison with existing models for VADs, the flow errors are more than 70% lower. Further advantages of the SVM model are the robustness to measurement noise and the capability to operate outside of the trained parameter range. This proposed modeling method will accelerate further device refinements by providing a more appropriate numerical environment in which to evaluate the pump-cardiovascular system interaction.
Closed-form equation to estimate the dielectric properties of biological tissues as a function of age
Publisher: Wiley
Date: 21-04-2017
DOI: 10.1002/BEM.22054
Abstract: Developing microwave systems for biomedical applications requires accurate dielectric properties of biological tissues for reliable modeling before prototyping and subject testing. Dielectric properties of tissues decrease with age due to the change in their water content, but there are no detailed age-dependent data, especially for young tissue-like newborns, in the literature. In this article, an age-dependent formula to predict the dielectric properties of biological tissues was derived. In the proposed method, the variation of water concentration in each type of tissue as a function of age was used to calculate its relative permittivity and conductivity. The derived formula shows that the concentration of water in each tissue type can be modeled as a negative exponential function of age. The dielectric properties of each tissue type can then be calculated as a function of the dielectric properties of water and dielectric properties of the organ forming the tissue and its water concentration. The derived formula was used to generate the dielectric properties of several types of human tissues at different ages using the dielectric properties of a human adult. Moreover, the formula was validated on pig tissues of different ages. A close agreement was achieved between the calculated and measured data with a maximum difference of only 2%. Bioelectromagnetics. 38:474-481, 2017. © 2017 Wiley Periodicals, Inc.
Virtual Microscopy with Extended Depth of Field
Publisher: IEEE
Date: 2005
Multiple instance learning for breast MRI based on generic spatio-temporal features
Publisher: IEEE
Date: 04-2015
Dielectric properties of dog brain tissue measured in vitro across the 0.3–3 GHz band
Publisher: Wiley
Date: 22-09-2016
DOI: 10.1002/BEM.22007
Abstract: Dielectric properties of dead Greyhound female dogs' brain tissues at different ages were measured at room temperature across the frequency range of 0.3-3 GHz. Measurements were made on excised tissues, in vitro in the laboratory, to carry out dielectric tests on s le tissues. Each dataset for a brain tissue was parametrized using the Cole-Cole expression, and the relevant Cole-Cole parameters for four tissue types are provided. A comparison was made with the database available in literature for other animals and human brain tissue. Results of two types of tissues (white matter and skull) showed systematic variation in dielectric properties as a function of animal age, whereas no significant change related to age was noticed for other tissues. Results provide critical information regarding dielectric properties of animal tissues for a realistic animal head model that can be used to verify the validity and reliability of a microwave head scanner for animals prior to testing on live animals. Bioelectromagnetics. 37:549-556, 2016. © 2016 Wiley Periodicals, Inc.
Virtual Microscopy with Extended Depth of Field
Publisher: IEEE
Date: 12-2005
Effect of competing stimuli on SSVEP-based BCI
Publisher: IEEE
Date: 08-2011
Screening of obstructive and central apnoea/hypopnoea in children using variability: A preliminary study
Publisher: Wiley
Date: 05-2006
DOI: 10.1080/08035250500477552
Abstract: Polysomnography (PSG) is the current standard protocol for sleep disordered breathing (SDB) investigation in children. Presently, there are limited reliable screening tests for both central (CE) and obstructive (OE) respiratory events. This study compared three indices, derived from pulse oximetry and electrocardiogram (ECG), with the PSG gold standard. These indices were heart rate (HR) variability, arterial blood oxygen de-saturation (SaO2) and pulse transit time (PTT). 15 children (12 male) from routine PSG studies were recruited (aged 3-14 years). The characteristics of the three indices were based on known criteria for respiratory events (RPE). Their estimation singly and in combination was evaluated with simultaneous scored PSG recordings. 215 RPE and 215 tidal breathing events were analysed. For OE, the obtained sensitivity was HR (0.703), SaO2 (0.047), PTT (0.750), considering all three indices (0) and either of the indices (0.828) while specificity was (0.891), (0.938), (0.922), (0.953) and (0.859) respectively. For CE, the sensitivity was HR (0.715), SaO2 (0.278), PTT (0.662), considering all indices (0.040) and either of the indices (0.868) while specificity was (0.815), (0.954), (0.901), (0.960) and (0.762) accordingly. Preliminary findings herein suggest that the later combination of these non-invasive indices to be a promising screening method of SDB in children.
Visual attention for region of interest coding in JPEG 2000
Publisher: Elsevier BV
Date: 09-2003
A preliminary investigation into the use of an auditory brainstem response (ABR) simulator for training audiology students in waveform analysis
Publisher: Informa UK Limited
Date: 04-04-2014
DOI: 10.3109/14992027.2014.897763
Abstract: To determine if a computer simulation can be used to improve the ability of audiology students to analyse ABR waveforms. A pretest-posttest, quasi-experimental design was used. All participants completed a pretraining examination of their ability to analyse ABR waveforms, eight hours of ABR analysis training over eight weeks using one of three training modes-manual, simulator or combined manual and simulator training, and a posttraining examination of their ability to analyse ABR waveforms. Fourteen third-year audiology students (13 female, one male, aged 21 to 22 years) participated in this study. Participants who completed the manual or the combined manual and simulator training achieved significantly higher normalized gain scores on their ABR waveform analysis examinations compared to those who completed the simulator training (p < 0.05). Our findings suggest that the improvements seen in the ability of these audiology students to analyse ABR waveforms were driven primarily by the manual training. The minimal improvements seen in the students who received the simulator training suggest that face-to-face instruction could be required to enhance the ability of audiology students to analyse ABR waveforms.
Low-frequency spectroscopic analysis of monomeric and fibrillar lysozyme
Publisher: SAGE Publications
Date: 03-2011
DOI: 10.1366/10-06162
Abstract: Terahertz time-domain spectroscopy (THz-TDS) and Fourier transform infrared (FT-IR) spectroscopy were used to generate far-infrared and low-frequency spectral measurements of monomeric lysozyme and lysozyme fibrils. The formation of lysozyme fibrils was verified by the Thioflavin T assay and transmission electron microscopy (TEM). It was evident in the FT-IR spectra that between 150 and 350 cm −1 the two spectra erge, with the lysozyme fibrils showing higher absorbance intensity than the monomeric form. The broad absorption phenomenon is likely due to light scattered from the fibrillar architecture of lysozyme fibrils as supported by simulation of Rayleigh light scattering. The lack of discrete phonon-like peaks suggest that far-infrared spectroscopy cannot detect vibrational modes between the highly ordered hydrogen-bonded beta-pleated sheets of the lysozyme subunit.
Half-AUC for the evaluation of sensitive or specific classifiers
Publisher: Elsevier BV
Date: 03-2014
On the methodology for comparing learning algorithms: a case
Publisher: IEEE
Date: 1994
Lesser of Two Evils Improves Learning in the Context of Cortical Thickness Estimation Models - Choose Wisely
Publisher: Springer Nature Switzerland
Date: 2022
Automated analysis of the auditory brainstem response using derivative estimation wavelets
Publisher: S. Karger AG
Date: 22-12-2005
DOI: 10.1159/000081544
Abstract: In this paper, we describe an algorithm that automatically detects and labels peaks I–VII of the normal, suprathreshold auditory brainstem response (ABR). The algorithm proceeds in three stages, with the option of a fourth: (1) all candidate peaks and troughs in the ABR waveform are identified using zero crossings of the first derivative, (2) peaks I–VII are identified from these candidate peaks based on their latency and morphology, (3) if required, peaks II and IV are identified as points of inflection using zero crossings of the second derivative and (4) interpeak troughs are identified before peak latencies and litudes are measured. The performance of the algorithm was estimated on a set of 240 normal ABR waveforms recorded using a stimulus intensity of 90 dBnHL. When compared to an expert audiologist, the algorithm correctly identified the major ABR peaks (I, III and V) in 96–98% of the waveforms and the minor ABR peaks (II, IV, VI and VII) in 45–83% of waveforms. Whilst peak II was correctly identified in only 83% and peak IV in 77% of waveforms, it was shown that 5% of the peak II identifications and 31% of the peak IV identifications came as a direct result of allowing these peaks to be found as points of inflection.
On wavelet analysis of auditory evoked potentials
Publisher: No publisher found
Date: 2004
Automated analysis of the auditory brainstem response using derivative estimation wavelets
Publisher: No publisher found
Date: 2005
DOI: 10.1159%2F000081544
Mixed content image compression by gradient field integration
Publisher: IEEE
Date: 2008
Rule Extraction from Support Vector Machines: A Sequential Covering Approach
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 06-2007
Use of pulse transit time to distinguish respiratory events from tidal breathing in sleeping children
Publisher: No publisher found
Date: 2005
Automatic lesion detection, segmentation and characterization via 3D multiscale morphological sifting in breast MRI
Publisher: IOP Publishing
Date: 11-2020
Abstract: Previous studies on computer aided detection/diagnosis (CAD) in 4D breast magnetic resonance imaging (MRI) usually regard lesion detection, segmentation and characterization as separate tasks, and typically require users to manually select 2D MRI slices or regions of interest as the input. In this work, we present a breast MRI CAD system that can handle 4D multimodal breast MRI data, and integrate lesion detection, segmentation and characterization with no user intervention. The proposed CAD system consists of three major stages: region candidate generation, feature extraction and region candidate classification. Breast lesions are firstly extracted as region candidates using the novel 3D multiscale morphological sifting (MMS). The 3D MMS, which uses linear structuring elements to extract lesion-like patterns, can segment lesions from breast images accurately and efficiently. Analytical features are then extracted from all available 4D multimodal breast MRI sequences, including T1-, T2-weighted and DCE sequences, to represent the signal intensity, texture, morphological and enhancement kinetic characteristics of the region candidates. The region candidates are lastly classified as lesion or normal tissue by the random under-s ling boost (RUSboost), and as malignant or benign lesion by the random forest. Evaluated on a breast MRI dataset which contains a total of 117 cases with 141 biopsy-proven lesions (95 malignant and 46 benign lesions), the proposed system achieves a true positive rate (TPR) of 0.90 at 3.19 false positives per patient (FPP) for lesion detection and a TPR of 0.91 at a FPP of 2.95 for identifying malignant lesions without any user intervention. The average dice similarity index (DSI) is 0.72 ± 0.15 for lesion segmentation. Compared with previously proposed lesion detection, detection-segmentation and detection-characterization systems evaluated on the same breast MRI dataset, the proposed CAD system achieves a favourable performance in breast lesion detection and characterization.
Validation and algorithmic audit of a deep learning system for the detection of proximal femoral fractures in patients in the emergency department: a diagnostic accuracy study
Publisher: Elsevier BV
Date: 05-2022
A Novel Fitness function of metaheuristic algorithms for test data generation for simulink models based on mutation analysis
Publisher: Elsevier BV
Date: 10-2016
Detail Matters: High-Frequency Content for Realistic Synthetic MRI Generation
Publisher: Springer International Publishing
Date: 2021
Enhancing the classification accuracy of steady-state visual evoked potential-based Brain-Computer interface using component synchrony measure
Publisher: IEEE
Date: 06-2012
On the dual structure of the auditory brainstem response in dogs
Publisher: No publisher found
Date: 2006
Nearest neighbour group-based classification
Publisher: Elsevier BV
Date: 10-2010
Visual quality assessment of watermarked medical images - art. no. 65151L
Publisher: SPIE
Date: 08-03-2007
DOI: 10.1117/12.707571
3d brain mri gan-based synthesis conditioned on partial volume maps
Publisher: Springer International Publishing
Date: 2020
ROC curve equivalence using the Kolmogorov–Smirnov test
Publisher: Elsevier BV
Date: 04-2013
Alterations in regional shape on ipsilateral and contralateral cortex contrast in children with unilateral cerebral palsy and are predictive of multiple outcomes
Publisher: Wiley
Date: 03-06-2016
DOI: 10.1002/HBM.23262
Progress towards universal neonatal hearing screening: A world review
Publisher: Portico
Date: 05-2009
DOI: 10.1375/AUDI.31.1.3
Intelligible support vector machines for diagnosis of diabetes mellitus
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 07-2010
Towards Optimal Image Stitching for Virtual Microscopy
Publisher: IEEE
Date: 2005
Microscopic specimen delineation using auto-phase correlation index
Publisher: IEEE
Date: 04-2014
The effect of domain knowledge on rule extraction from support vector machines
Publisher: Springer Berlin Heidelberg
Date: 2009
A system to generate patient-specific stimuli for use with the auditory brainstem response test
Publisher: IEEE
Date: 08-2007
Author Correction: Why rankings of biomedical image analysis competitions should be interpreted with care
Publisher: Springer Science and Business Media LLC
Date: 30-01-2019
DOI: 10.1038/S41467-019-08563-W
Abstract: In the original version of this Article the values in the rightmost column of Table 1 were inadvertently shifted relative to the other columns. This has now been corrected in the PDF and HTML versions of the Article.
Fully Automatic Computer-aided Mass Detection and Segmentation via Pseudo-color Mammograms and Mask R-CNN
Publisher: IEEE
Date: 04-2020
The filtered words test and the influence of lexicality
Publisher: American Speech Language Hearing Association
Date: 10-2014
DOI: 10.1044/2014_JSLHR-L-13-0210
Abstract: In the present study, the authors aimed to investigate the language confounds of filtered words tests by examining the repetition of real words versus nonsense words as a function of level of filtering. Fifty-five young, native-English-speaking women with normal hearing were required to repeat 80 real-word and 80 nonsense-word monosyllables that were matched for phonemic content and low-pass filtered. Thirty participants were tested using a harsher filter range of 2000 to 500 Hz, and 25 participants were tested using a milder filter range of 3000 to 1500 Hz. Paired-s le t tests compared accuracy (percentage of phonemes correct) for word and nonsense-word stimuli at each filter level. At filter levels between 3000 and 1750 Hz, performance for word stimuli was significantly better than for nonsense-word stimuli. Conversely, at filter levels between 500 and 1250 Hz, performance was significantly better for nonsense words. The linguistic content of real-word stimuli benefits performance on low-pass filtered speech tests at filter levels above 1500 Hz. Caution must be taken when using real-word stimuli in low-pass filtered speech tests as part of an auditory processing diagnostic test battery, because language ability will impact on performance.
Erratum: A Neuroethics Framework for the Australian Brain Initiative (Neuron (2019) 101(3) (365–369), (S0896627319300054), (10.1016/j.neuron.2019.01.004))
Publisher: Elsevier BV
Date: 2020
Estimation of neuronal firing rates with the three-state biological point process model
Publisher: Elsevier BV
Date: 09-2008
DOI: 10.1016/J.JNEUMETH.2008.05.026
Abstract: In the subcortex of the human brain, neuronal firing events are stochastic and the inter-arrival times of action potentials (APs) are highly irregular. It has been shown that stimulation of the subthalamic nucleus (STN), a small subcortical structure located within the basal ganglia, can help ameliorate the motor symptoms associated with Parkinson's disease (PD). However, success of image guided stereotactic surgery is reliant upon the refinement of the anatomic target (in this case the STN) based on micro-electrode recordings (MERs) of background activity and firing rate. In practice MERs must be analysed on-line and in real-time. Currently, the most common method of performing on-line MER analysis is a manual thresholding procedure. However, this is subjective in nature and often complicated by the presence of variable amounts of background noise. Therefore, in this paper, we present an automated adaptive thresholding technique, based on a modified 'top-hat' operator, which detects APs exceeding the local background activity. We then go on to model these inter-arrival times using a coupled Poisson process that provides improved estimates of both inter-burst and intra-burst neuronal firing activity in the STN.
An Evaluation of Multi-resolution Microscope Slide Scanning Algorithms
Publisher: IEEE
Date: 12-2011
Deep Reinforcement Learning for Active Breast Lesion Detection from DCE-MRI
Publisher: Springer International Publishing
Date: 2017
A two-stage method to correct aberrations induced by slide slant in bright-field microscopy
Publisher: Elsevier BV
Date: 08-2016
DOI: 10.1016/J.MICRON.2016.05.001
Abstract: To achieve optimal image quality in bright field microscopy, the slide surface should be perpendicular to the optical axis of the microscope. However, in the recently proposed "slanted scan" slide acquisition technique, scan speed is increased by purposely slanting the slide by a small angle (of 3-5°) so that multiple focal depths can be imaged simultaneously. In this case, the slanted slide introduces a bend in the point spread function (PSF), resulting in a coma and other aberrations that degrade image quality. In this paper, we propose a two-stage deconvolution method specifically designed to correct the aberrations induced by a slanted scan, but with general applicability to high-resolution bright-field microscopy. Specifically, we initially apply phase deconvolution to correct the dominating coma aberration, before applying a conventional semi-blind deconvolution method to further improve image resolution and contrast. We also propose a novel method to estimate the degree of coma aberration and the PSF of the optics utilising actual cytology specimens. The efficacy of the proposed algorithm is demonstrated quantitatively on simulated data, against a ground-truth (object) image, and qualitatively on cervical cytology specimens. Results demonstrate both improved convergence speed of the two-stage approach, especially when correcting the bend in the PSF, and a resultant image quality that is comparable to a conventionally (flat) scanned specimen.
Tree RE-weighted belief propagation using deep learning potentials for mass segmentation from mammograms
Publisher: IEEE
Date: 04-2015
Deep Learning Models for Classifying Mammogram Exams Containing Unregistered Multi-View Images and Segmentation Maps of Lesions11This work is an extension of the paper published by the same authors at
Publisher: Elsevier
Date: 2017
Expectation-Maximization with Image-Weighted Markov Random Fields to Handle Severe Pathology
Publisher: IEEE
Date: 11-2015
A Starling‐like total work controller for rotary blood pumps: An in vitro evaluation
Publisher: Wiley
Date: 20-10-2019
DOI: 10.1111/AOR.13570
Abstract: Due to improved durability and survival rates, rotary blood pumps (RBPs) are the preferred left ventricular assist device when compared to volume displacement pumps. However, when operated at constant speed, RBPs lack a volume balancing mechanism which may result in left ventricular suction and suboptimal ventricular unloading. Starling-like controllers have previously been developed to balance circulatory volumes however, they do not consider ventricular workload as a feedback and may have limited sensitivity to adjust RBP workload when ventricular function deteriorates or improves. To address this, we aimed to develop a Starling-like total work controller (SL-TWC) that matched the energy output of a healthy heart by adjusting RBP hydraulic work based on measured left ventricular stroke work and ventricular preload. In a mock circulatory loop, the SL-TWC was evaluated using a HeartWare HVAD in a range of simulated patient conditions. These conditions included changes in systemic hypertension and hypotension, pulmonary hypertension, blood circulatory volume, exercise, and improvement and deterioration of ventricular function by increasing and decreasing ventricular contractility. The SL-TWC was compared to constant speed control where RBP speed was set to restore cardiac output to 5.0 L/min at rest. Left ventricular suction occurred with constant speed control during pulmonary hypertension but was prevented with the SL-TWC. During simulated exercise, the SL-TWC demonstrated reduced LVSW (0.51 J) and greater RBP flow (9.2 L/min) compared to constant speed control (LVSW: 0.74 J and RBP flow: 6.4 L/min). In instances of increased ventricular contractility, the SL-TWC reduced RBP hydraulic work while maintaining cardiac output similar to the rest condition. In comparison, constant speed overworked and increased cardiac output. The SL-TWC balanced circulatory volumes by mimicking the Starling mechanism, while also considering changes in ventricular workload. Compared to constant speed control, the SL-TWC may reduce complications associated with volume imbalances, adapt to changes in ventricular function and improve patient quality of life.
Over-sampling for accurate masking thershold calculation wavelet packet audio coders
Publisher: IEEE
Date: 2004
The effects of electrode montage on the amplitude of wave V in the auditory brainstem response to maximum length sequence stimuli
Publisher: S. Karger AG
Date: 22-08-2007
DOI: 10.1159/000107432
Abstract: The use of maximum length sequence (MLS) stimuli to elicit an auditory brainstem response (ABR) has been limited, in part, by the observation that these stimuli reduce ABR wave litudes. This study recorded ABR waveforms from 14 normally hearing adults using MLS click stimuli (maximum stimulus rate = 250 clicks per second) at stimulus levels of 70, 60, 50, 40, 30 and 20 dB nHL, with a vertical and then an ipsilateral electrode montage. The vertical electrode montage produced significantly larger (p 0.05) wave V litudes, with no change in wave V latencies (p 0.05), at all stimulus levels. This result suggests a vertical electrode montage could be used to counter some of the loss in wave V litude observed when using MLS stimuli.
Producing radiologist-quality reports for interpretable deep learning.
Publisher: IEEE
Date: 04-2019
Conducting shorter VEP tests to estimate visual acuity via assessment of SNR
Publisher: Springer Science and Business Media LLC
Date: 26-10-2013
DOI: 10.1007/S10633-012-9355-9
Abstract: The estimation of visual acuity (VA) via visual evoked potentials (VEP) is a valuable measure for all preverbal and non-verbal subjects whether adults or children. The aim of this study is to introduce a novel technique of VEP acquisition based on estimates of signal-to-noise ratio (SNR) and comparison to a predefined detection threshold. We aim to demonstrate the reduction in total study time without compromising the accuracy of the VEP-determined acuity estimate. The VEP-determined acuity of twelve normal subjects was assessed via a spatial frequency (SF) sweep. A pattern reversal checkerboard stimulus utilised SFs ranging from 0.1 to 28 cycles per degree (cpd). Using linear extrapolation and Bland-Altman analysis, VEP-acquired acuity was compared to a conventional Snellen Acuity measurement. An SNR test, Fsp, assessed signal quality to determine the minimum amount of sweep data required for VEP-based VA estimation. VEP acuity estimates correlated strongly (r2=0.91, SD=0.06), leading to a VA limit via extrapolation. Bland-Altman analysis revealed agreement between tests is statistically valid (95% CI -0.11 to 0.42 logMAR). The Fsp statistic indicated SFs 1.3-3.6 cpd yielded Fsp>3.1 within 15 s of acquisition with frequencies>3.6 cpd being sub-threshold. The Kruskal-Wallis statistic compared final Fsp values for SFs as groups, where F=208.82 ranking each frequency, with frequencies>7.2 cpd ranking lowest. The Fsp as an SNR measurement shows that rapid, quality-driven clinical tests for VEP-based acuity estimates can be conducted without compromising accuracy.
A Comparison of Multiple Instance and Group Based Learning
Publisher: IEEE
Date: 12-2012
ROC curves and the x2 test
Publisher: No publisher found
Date: 1996
A deep learning approach for the analysis of masses in mammograms with minimal user intervention
Publisher: Elsevier BV
Date: 04-2017
DOI: 10.1016/J.MEDIA.2017.01.009
Abstract: We present an integrated methodology for detecting, segmenting and classifying breast masses from mammograms with minimal user intervention. This is a long standing problem due to low signal-to-noise ratio in the visualisation of breast masses, combined with their large variability in terms of shape, size, appearance and location. We break the problem down into three stages: mass detection, mass segmentation, and mass classification. For the detection, we propose a cascade of deep learning methods to select hypotheses that are refined based on Bayesian optimisation. For the segmentation, we propose the use of deep structured output learning that is subsequently refined by a level set method. Finally, for the classification, we propose the use of a deep learning classifier, which is pre-trained with a regression to hand-crafted feature values and fine-tuned based on the annotations of the breast mass classification dataset. We test our proposed system on the publicly available INbreast dataset and compare the results with the current state-of-the-art methodologies. This evaluation shows that our system detects 90% of masses at 1 false positive per image, has a segmentation accuracy of around 0.85 (Dice index) on the correctly detected masses, and overall classifies masses as malignant or benign with sensitivity (Se) of 0.98 and specificity (Sp) of 0.7.
An improved joint optimization of multiple level set functions for the segmentation of overlapping cervical cells
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 04-2015
Fast assessment of canine hearing using high click-rate BAER
Publisher: Elsevier BV
Date: 2011
DOI: 10.1016/J.TVJL.2009.10.009
Abstract: The aim of this study was to determine if high stimulus repetition rates could reduce the time taken to obtain brainstem auditory evoked response (BAER) waveforms of equivalent quality in dogs. Click-evoked BAER waveforms were obtained from nine healthy, adult, mixed-breed dogs at stimulus intensities of 70, 60, 50 and 40 decibels (normal hearing level) (dBnHL) and stimulus repetition rates of 11, 33 and 91 clicks-per-second (cps). The quality of the BAER waveforms was kept constant by ensuring all waveforms achieved the same signal-to-noise (SNR), as shown by their F(sp) value of 3.1. Increasing the stimulus repetition rate from 11 to 91Hz significantly (P<0.01) reduced the median time to obtain BAER waveforms of equivalent quality by 3.29-14.07s per waveform, or alternatively, increased the recording speed by 4.6-13.7 times per waveform (depending on the stimulus intensity). The use of high stimulus repetition rate BAER shows significant promise for the rapid assessment of auditory function in dogs.
Pre and post-hoc diagnosis and interpretation of malignancy from breast DCE-MRI
Publisher: Elsevier BV
Date: 12-2019
DOI: 10.1016/J.MEDIA.2019.101562
Abstract: We propose a new method for breast cancer screening from DCE-MRI based on a post-hoc approach that is trained using weakly annotated data (i.e., labels are available only at the image level without any lesion delineation). Our proposed post-hoc method automatically diagnosis the whole volume and, for positive cases, it localizes the malignant lesions that led to such diagnosis. Conversely, traditional approaches follow a pre-hoc approach that initially localises suspicious areas that are subsequently classified to establish the breast malignancy - this approach is trained using strongly annotated data (i.e., it needs a delineation and classification of all lesions in an image). We also aim to establish the advantages and disadvantages of both approaches when applied to breast screening from DCE-MRI. Relying on experiments on a breast DCE-MRI dataset that contains scans of 117 patients, our results show that the post-hoc method is more accurate for diagnosing the whole volume per patient, achieving an AUC of 0.91, while the pre-hoc method achieves an AUC of 0.81. However, the performance for localising the malignant lesions remains challenging for the post-hoc method due to the weakly labelled dataset employed during training.
The multiscale classifier
Publisher: No publisher found
Date: 1996
DOI: 10.1109%2F34.481538
On the dual structure of the auditory brainstem response in dogs
Publisher: Elsevier BV
Date: 10-2006
DOI: 10.1016/J.CLINPH.2006.06.711
Abstract: To use the over-complete discrete wavelet transform (OCDWT) to further examine the dual structure of auditory brainstem response (ABR) in the dog. ABR waveforms recorded from 20 adult dogs at supra-threshold (90 and 70dBnHL) and threshold (0-15dBSL) levels were decomposed using a six level OCDWT and reconstructed at in idual scales (frequency ranges) A6 (0-391Hz), D6 (391-781Hz), and D5 (781-1563Hz). At supra-threshold stimulus levels, the A6 scale (0-391Hz) showed a large litude waveform with its prominent wave corresponding in latency with ABR waves II/III the D6 scale (391-781Hz) showed a small litude waveform with its first four waves corresponding in latency to ABR waves I, II/III, V, and VI and the D5 scale (781-1563Hz) showed a large litude, multiple peaked waveform with its first six waves corresponding in latency to ABR waves I, II, III, IV, V, and VI. At threshold stimulus levels (0-15dBSL), the A6 scale (0-391Hz) continued to show a relatively large litude waveform, but both the D6 and D5 scales (391-781 and 781-1563Hz, respectively) now showed relatively small litude waveforms. A dual structure exists within the ABR of the dog, but its relative structure changes with stimulus level. The ABR in the dog differs from that in the human both in the relative contributions made by its different frequency components, and the way these components change with stimulus level.
Over-complete discrete wavelet transformation of the normal auditory brainstem response improves prediction of outcome following severe acute closed head injury
Publisher: S. Karger AG
Date: 2006
DOI: 10.1159/000093111
Abstract: Previous research has shown that complex statistical analysis (discriminant function analysis) of a ‘normal’ auditory brainstem response (ABR) result can improve this measure’s ability to predict subject outcome following severe acute closed head injury (ACHI). We hypothesized that adding the ABR’s time-frequency information to such an analysis would improve this predictive value even further. ‘Normal’ ABR results were s led from 69 severe ACHI subjects (22 of whom died and 47 of whom lived) and their time-frequency information extracted using an over-complete discrete wavelet transformation (OCDWT). A series of logistic regression analyses then showed correct predictions of death and survival as follows: ABR measures only 72 and 89% (respectively), ABR OCDWT measures only 82 and 89% (respectively), and ABR and ABR OCDWT measures combined 86 and 93% (respectively). These results showed that the addition of time-frequency information can improve the ability of the ‘normal’ ABR result to predict outcome following severe ACHI.
Rule Extraction from Support Vector Machines: Measuring the Explanation Capability Using the Area under the ROC Curve
Publisher: IEEE
Date: 2006
Spikes from compound action potentials in simulated microelectrode recordings
Publisher: IEEE
Date: 04-2015
Feature and classifier selection for automatic classification of lesions in dynamic contrast-enhanced MRI of the breast
Publisher: IEEE
Date: 2009
Spectral properties of neuronal pulse interval modulation
Publisher: IEEE
Date: 04-2015
Ethics and data mining in biomedical engineering
Publisher: Springer New York
Date: 2013
Teaching Histology to First-Year Veterinary Science Students Using Virtual Microscopy and Traditional Microscopy: A Comparison of Student Responses
Publisher: University of Toronto Press Inc. (UTPress)
Date: 04-2007
Abstract: Virtual microscopy (VM) is a comparatively recent innovation that is revolutionizing both the teaching of microscopic structure in human medicine and the concept of online diagnosis and telemedicine. The interactivity of the various commercially available browsers attempts to simulate the experience of looking down a microscope while offering advantages over traditional microscopy that include clarity of image, reduced infrastructure, and high flexibility, as the images are accessible online. We developed our own VM system, including customized software and a browser that was simple and intuitive to use, with the added advantage of further modifications possible to assist student learning. In this article, we report on a preliminary study wherein VM was introduced to veterinary science students in one course and directly compared to traditional microscopy to determine whether students would readily accept this new technology and which aspects of VM were advantageous. Responses from a survey form showed that students rated VM significantly higher than traditional microscopy as a tool to learn histology because it offers clearer images, the ability to learn collaboratively, more effective use of time, and the flexibility of online learning. Students also indicated a strong preference for the use of VM in future courses. These results suggest that VM is a flexible and enjoyable resource that could be useful to enhance the learning of microscopic structure in veterinary science courses.
When to Use Augmentation - Variability Insufficient for Cortical Thickness Estimation Improvement
Publisher: IEEE
Date: 18-04-2023
Smocam: Smooth Conditional Attention Mask For 3d-Regression Models
Publisher: IEEE
Date: 13-04-2021
Evaluation of a morphological filter in mean cardiac output determination: Application to left ventricular assist devices
Publisher: Springer Science and Business Media LLC
Date: 23-03-2013
DOI: 10.1007/S11517-013-1061-6
Abstract: A morphological filter (MF) is presented for the determination of beat-to-beat mean rotary left ventricular assist device (LVAD) flow rate, measured using an implanted flow probe. The performance of this non-linear filter was assessed using LVAD flow rate (QLVAD) data sets obtained from in silico and in vivo sources. The MF was compared with a third-order Butterworth filter (BWF) and a 10-s moving average filter (MAF). Performance was assessed by calculating the response time and steady state error across a range of heart rates and levels of noise. The response time of the MF was 3.5 times faster than the MAF, 0.5 s slower than the BWF, and had a steady state error of 2.61 %. It completely removed pulsatile signal components caused by residual ventricular function, and tracked sharp transient changes in QLVAD better than the BWF. The use of a two-stage MF improved the noise immunity compared to the single-stage MF. This study showed that the good performance characteristics of the non-linear MF make it a more suitable candidate for embedded real-time processing of QLVAD than linear filters.
Unregistered Multiview Mammogram Analysis with Pre-trained Deep Learning Models
Publisher: Springer International Publishing
Date: 2015
A Comparison of Saliency Methods for Deep Learning Explainability
Publisher: IEEE
Date: 11-2021
Model agnostic saliency for weakly supervised lesion detection from breast DCE-MRI
Publisher: IEEE
Date: 04-2019
Multisite accelerometry for sleep and wake classification in children
Publisher: IOP Publishing
Date: 16-12-2014
DOI: 10.1088/0967-3334/36/1/133
Abstract: Actigraphy is a useful alternative to the gold standard polysomnogram for non-invasively measuring sleep and wakefulness. However, it is unable to accurately assess sleep fragmentation due to its inability to differentiate restless sleep from wakefulness and quiet wake from sleep. This presents significant limitations in the assessment of sleep-related breathing disorders where sleep fragmentation is a common symptom. We propose that this limitation may be caused by hardware constraints and movement representation techniques. Our objective was to determine if multisite tri-axial accelerometry improves sleep and wake classification. Twenty-four patients aged 6-15 years (median: 8 years, 16 male) underwent a diagnostic polysomnogram while simultaneously recording motion from the left wrist and index fingertip, upper thorax and left ankle and great toe using a custom accelerometry system. Movement was quantified using several features and two feature selection techniques were employed to select optimal features for restricted feature set sizes. A heuristic was also applied to identify movements during restless sleep. The sleep and wake classification performance was then assessed and validated against the manually scored polysomnogram using discriminant analysis. Tri-axial accelerometry measured at the wrist significantly improved the wake detection when compared to uni-axial accelerometry (specificity at 85% sensitivity: 71.3(14.2)% versus 55.2(24.7)%, p < 0.01). Multisite accelerometry significantly improved the performance when compared to the single wrist placement (specificity at 85% sensitivity: 82.1(12.5)% versus 71.3(14.2)%, p < 0.05). Our results indicate that multisite accelerometry offers a significant performance benefit which could be further improved by analysing movement in raw multisite accelerometry data.
Automated, quantitative measures of grey and white matter lesion burden correlates with motor and cognitive function in children with unilateral cerebral palsy
Publisher: Elsevier BV
Date: 2016
Spectral and synchrony differences in auditory brainstem responses evoked by chirps of varying durations.
Publisher: Acoustical Society of America (ASA)
Date: 10-2010
DOI: 10.1121/1.3483738
Abstract: The chirp-evoked ABR has been termed a more synchronous response, referring to the fact that rising-frequency chirp stimuli theoretically compensate for temporal dispersions down the basilar membrane. This compensation is made possible by delaying the higher frequency content of the stimulus until the lower frequency traveling waves are closer to the cochlea apex. However, it is not yet clear how sensitive this temporal compensation is to variation in the delay interval. This study analyzed chirp- and click-evoked ABRs at low intensity, using a variety of tools in the time, frequency, and phase domains, to measure synchrony in the response. Additionally, this study also examined the relationship between chirp sweep rate and response synchrony by varying the delay between high- and low-frequency portions of chirp stimuli. The results suggest that the chirp-evoked ABRs in this study exhibited more synchrony than the click-evoked ABRs and that slight gender-based differences exist in the synchrony of chirp-evoked ABRs. The study concludes that a tailoring of chirp parameters to gender may be beneficial in pathologies that severely affect neural synchrony, but that such a customization may not be necessary in routine clinical applications.
Effect of posterized naturalistic stimuli on SSVEP-based BCI
Publisher: IEEE
Date: 07-2013
The automated learning of deep features for breast mass classification from mammograms
Publisher: Springer International Publishing
Date: 2016
Ethics for biomedical engineers
Publisher: Springer New York
Date: 2013
Deep learning and structured prediction for the segmentation of mass in mammograms
Publisher: Springer International Publishing
Date: 2015
Sample size estimation using the receiver operating characteristic curve
Publisher: IEEE
Date: 2004
Far-infrared spectroscopy of protein higher-order structures
Publisher: SAGE Publications
Date: 11-2010
DOI: 10.1366/000370210793335025
Abstract: Far-infrared (FIR) spectroscopy in the spectral region of 50–450 cm −1 was used to study a series of protein higher-order structures constructed using β-lactoglobulin and polyomavirus capsid protein VP1. There were marked differences in the spectra for β-lactoglobulin monomer and dimer and between untreated β-lactoglobulin and heat-induced gels formed at neutral pH. Untreated β-lactoglobulin and heat-induced gels formed at acidic pH exhibited little difference in their spectra. Assembly of the quaternary structure of polyomavirus virus-like particles also caused large changes in the FIR spectra. These findings suggest that FIR spectroscopy may prove useful in studying some protein quaternary and higher-order structures. There was evidence of detection of β-lactoglobulin dimerization, intermolecular disulfide bonding in heat-induced neutral gels, and polyomavirus virus-like particle assembly but no evidence that FIR could detect β-lactoglobulin fibrils with their polymeric structure and hydrogen-bonded intermolecular β-pleated sheeting.
Evaluation of Three Algorithms for the Segmentation of Overlapping Cervical Cells
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 03-2017
Stimulus specificity of a steady-state visual-evoked potential-based brain-computer interface
Publisher: IOP Publishing
Date: 15-05-2012
DOI: 10.1088/1741-2560/9/3/036008
Abstract: The mechanisms of neural excitation and inhibition when given a visual stimulus are well studied. It has been established that changing stimulus specificity such as luminance contrast or spatial frequency can alter the neuronal activity and thus modulate the visual-evoked response. In this paper, we study the effect that stimulus specificity has on the classification performance of a steady-state visual-evoked potential-based brain-computer interface (SSVEP-BCI). For ex le, we investigate how closely two visual stimuli can be placed before they compete for neural representation in the cortex and thus influence BCI classification accuracy. We characterize stimulus specificity using the four stimulus parameters commonly encountered in SSVEP-BCI design: temporal frequency, spatial size, number of simultaneously displayed stimuli and their spatial proximity. By varying these quantities and measuring the SSVEP-BCI classification accuracy, we are able to determine the parameters that provide optimal performance. Our results show that superior SSVEP-BCI accuracy is attained when stimuli are placed spatially more than 5° apart, with size that subtends at least 2° of visual angle, when using a tagging frequency of between high alpha and beta band. These findings may assist in deciding the stimulus parameters for optimal SSVEP-BCI design.
Physiologic parameters that affect pulse transit time difference between the upper and lower limbs in children.
Publisher: Springer Science and Business Media LLC
Date: 05-01-2006
Towards data‐driven quantification of skin ageing using reflectance confocal microscopy
Publisher: Wiley
Date: 26-07-2021
DOI: 10.1111/ICS.12720
Abstract: Evaluation of skin ageing is a non‐standardized, subjective process, with typical measures relying coarse, qualitatively defined features. Reflectance confocal microscopy depth stacks contain indicators of both chrono‐ageing and photo‐ageing. We hypothesize that an ageing scale could be constructed using machine learning and image analysis, creating a data‐driven quantification of skin ageing without human assessment. En‐face sections of reflectance confocal microscopy depth stacks from the dorsal and volar forearm of 74 participants (36/18/20 training/testing/validation) were represented using a histogram of visual features learned using unsupervised clustering of small image patches. A logistic regression classifier was trained on these histograms to differentiate between stacks from 20‐ to 30‐year‐old and 50‐ to 70‐year‐old volunteers. The probabilistic output of the logistic regression was used as the fine‐grained ageing score for that stack in the testing set ranging from 0 to 1. Evaluation was performed in two ways: on the test set, the AUC was collected for the binary classification problem as well as by statistical comparison of the scores for age and body site groups. Final validation was performed by assessing the accuracy of the ageing score measurement on 20 depth stacks not used for training or evaluating the classifier. The classifier effectively differentiated stacks from age groups with a test set AUC of 0.908. Mean scores were significantly different when comparing age groups (mean 0.70 vs. 0.44 t = −6.62, p = 0.0000) and also when comparing stacks from dorsal and volar body sites (mean 0.64 vs. 0.53 t = 3.12, p = 0.0062). On the final validation set, 17 out of 20 depth stacks were correctly labelled. Despite being limited to only coarse training information in the form of ex le stacks from two age groups, the trained classifier was still able to effectively discriminate between younger skin and older skin. Curiously, despite being only trained with chronological age, there was still evidence for measurable differences in age scores due to sun exposure—with marked differences in scores on sun‐exposed dorsal sites of some volunteers compared with less sun‐exposed volar sites. These results suggest that fine‐grained data‐driven quantification of skin ageing is achievable.
The multiscale classifier
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 1996
DOI: 10.1109/34.481538
Perceptual quality metrics applied to still image compression
Publisher: Elsevier BV
Date: 11-1998
On chirp stimuli and neural synchrony in the suprathreshold auditory brainstem response
Publisher: Acoustical Society of America (ASA)
Date: 07-2010
DOI: 10.1121/1.3436527
Abstract: The chirp-evoked ABR has been regarded as a more synchronous response than the click-evoked ABR, referring to the belief that the chirp stimulates lower-, mid-, and higher-frequency regions of the cochlea simultaneously. In this study a variety of tools were used to analyze the synchronicity of ABRs evoked by chirp- and click-stimuli at 40 dB HL in 32 normal hearing subjects aged 18 to 55 years (mean=24.8 years, SD=7.1 years). Compared to the click-evoked ABRs, the chirp-evoked ABRs showed larger wave V litudes, but an absence of earlier waves in the grand averages, larger wave V latency variance, smaller FFT magnitudes at the higher component frequencies, and larger phase variance at the higher component frequencies. These results strongly suggest that the chirp-evoked ABRs exhibited less synchrony than the click-evoked ABRs in this study. It is proposed that the temporal compensation offered by chirp stimuli is sufficient to increase neural recruitment (as measured by wave V litude), but that destructive phase interactions still exist along the cochlea partition, particularly in the low frequency portions of the cochlea where more latency jitter is expected. The clinical implications of these findings are discussed.
Characterization of movements during restless sleep in children: A pilot study
Publisher: IEEE
Date: 08-2014
Temporal associations between arousal and body/limb movement in children with suspected obstructed sleep apnoea
Publisher: IOP Publishing
Date: 07-12-2015
DOI: 10.1088/0967-3334/37/1/115
Abstract: The inter-relationship between arousal events and body and/or limb movements during sleep may significantly impact the performance and clinical interpretation of actigraphy. As such, the objective of this study was to quantify the temporal association between arousals and body/limb movement. From this, we aim to determine whether actigraphy can predict arousal events in children, and identify the impact of arousal-related movements on estimates of sleep/wake periods. Thirty otherwise healthy children (5-16 years, median 9 years, 21 male) with suspected sleep apnoea were studied using full polysomnography and customised raw tri-axial accelerometry measured at the left fingertip, left wrist, upper thorax, left ankle and left great toe. Raw data were synchronised to within 0.1 s of the polysomnogram. Movements were then identified using a custom algorithm. On average 67.5% of arousals were associated with wrist movement. Arousals associated with movement were longer than those without movement (mean duration: 12.2 s versus 7.9 s respectively, p < 0.01) movements during wake and arousal were longer than other sleep movements (wrist duration: 6.26 s and 9.89 s versus 2.35 s respectively, p < 0.01) and the movement index (movements/h) did not predict apnoea-hypopnoea index (ρ = -0.11). Movements associated with arousals are likely to unavoidably contribute to actigraphy's poor sensitivity for wake. However, as sleep-related movements tend to be shorter than those during wake or arousal, incorporating movement duration into the actigraphy scoring algorithm may improve sleep staging performance. Although actigraphy-based measurements cannot reliably predict all arousal events, actigraphy can likely identify longer events that may have the greatest impact on sleep quality.
Rule extraction from support vector machines: A review
Publisher: Elsevier BV
Date: 12-2010
Self-Supervised Learning With Adaptive Distillation for Hyperspectral Image Classification
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 2022
Automatic segmentation of enhancing breast tissue in dynamic contrast-enhanced MR images
Publisher: IEEE
Date: 12-2007
Using ventricular modeling to robustly probe significant deep gray matter pathologies: Application to cerebral palsy
Publisher: Wiley
Date: 06-10-2016
DOI: 10.1002/HBM.23276
Estimation Of Epipolar Geometry Via The Radon Transform
Publisher: IEEE
Date: 2006
Screening of obstructive and central apnoea/hypopnoea in children using variability: A preliminary study
Publisher: No publisher found
Date: 2006
3D Motion Estimation from Three Orthographic Views without Matching Constraints or Brightness Gradients
Publisher: IEEE
Date: 2006
Correspondence-Free Determination of the Affine Fundamental Matrix
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 2007
DOI: 10.1109/TPAMI.2007.250601
Abstract: Fundamental matrix estimation is a central problem in computer vision and forms the basis of tasks such as stereo imaging and structure from motion. Existing algorithms typically analyze the relative geometries of matched feature points identified in both projected views. Automated feature matching is itself a challenging problem. Results typically have a large number of false matches. Traditional fundamental matrix estimation methods are very sensitive to matching errors, which led naturally to the application of robust statistical estimation techniques to the problem. In this work, an entirely novel approach is proposed to the fundamental matrix estimation problem. Instead of analyzing the geometry of matched feature points, the problem is recast in the frequency domain through the use of Integral Projection, showing how this is a reasonable model for orthographic cameras. The problem now reduces to one of identifying matching lines in the frequency domain which, most importantly, requires no feature matching or correspondence information. Experimental results on both real and synthetic data are presented that demonstrate the algorithm is a practical technique for fundamental matrix estimation. The behavior of the proposed algorithm is additionally characterized with respect to input noise, feature counts, and other parameters of interest.
Group-based meta-classification
Publisher: IEEE
Date: 12-2008
A new denoising method for dynamic contrast-enhanced MRI
Publisher: IEEE
Date: 08-2008
A Parametric Simulation of Neuronal Noise From Microelectrode Recordings
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 2017
Precision Radiology: Predicting longevity using feature engineering and deep learning methods in a radiomics framework
Publisher: Springer Science and Business Media LLC
Date: 10-05-2017
DOI: 10.1038/S41598-017-01931-W
Abstract: Precision medicine approaches rely on obtaining precise knowledge of the true state of health of an in idual patient, which results from a combination of their genetic risks and environmental exposures. This approach is currently limited by the lack of effective and efficient non-invasive medical tests to define the full range of phenotypic variation associated with in idual health. Such knowledge is critical for improved early intervention, for better treatment decisions, and for ameliorating the steadily worsening epidemic of chronic disease. We present proof-of-concept experiments to demonstrate how routinely acquired cross-sectional CT imaging may be used to predict patient longevity as a proxy for overall in idual health and disease status using computer image analysis techniques. Despite the limitations of a modest dataset and the use of off-the-shelf machine learning methods, our results are comparable to previous ‘manual’ clinical methods for longevity prediction. This work demonstrates that radiomics techniques can be used to extract biomarkers relevant to one of the most widely used outcomes in epidemiological and clinical research – mortality, and that deep learning with convolutional neural networks can be usefully applied to radiomics research. Computer image analysis applied to routinely collected medical images offers substantial potential to enhance precision medicine initiatives.
On the estimation of extrinsic and intrinsic parameters for optical microscope calibration
Publisher: IEEE
Date: 12-2010
Denoising of dynamic contrast-enhanced MR images using dynamic non-local means
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 02-2010
Profiling the change in refractive index using the self-mixing effect in lasers
Publisher: IEEE
Date: 12-2012
On the benefits of using chained stimuli for frequency-specific ABR acquisition
Publisher: Portico
Date: 11-2009
DOI: 10.1375/AUDI.31.2.80
Special issue: 4th MICCAI workshop on deep learning in medical image analysis
Publisher: Informa UK Limited
Date: 02-09-2020
Identifying relevant biomarkers of brain injury from structural MRI: Validation using automated approaches in children with unilateral cerebral palsy
Publisher: Public Library of Science (PLoS)
Date: 08-2017
An algorithm for microscopic specimen delineation and focus candidate selection
Publisher: Elsevier BV
Date: 11-2014
DOI: 10.1016/J.MICRON.2014.05.006
Abstract: In this paper, we compare four field-of-view (FOV) metrics that, when applied to a low-resolution image of a microscope slide, are capable of both accurately delineating the specimen and selecting a subset of focus candidate FOVs required for construction of high-resolution focus map. The metrics evaluated are: threshold index (TI) that measures image intensity normalised auto-correlation index (NACI) that measures spatial image similarity auto-phase correlation index (APCI) that measures image phase ersity and entropy index (EI) that measures the predictability of image intensities. Experiments are undertaken on a data set of forty slides including PAP stained Thin-prep cervical cytology and breast fine-needle aspiration slides and haematoxylin and eosin (HE) stained histology slides. These slides were scanned on an automated bright-field microscope and chosen to be indicative of a variety pathology specimens, containing artefacts such as excess coverslip glue and ink markers. Results are presented on the performance of each metric for correct ranking/segmentation of foreground (specimen) from background, and subsequently selecting focus candidate FOVs characteristic of the specimen's focal plane(s). The experimental results demonstrate that while NACI, APCI and EI are all effective at specimen delineation, only APCI is capable of effectively selecting superior focus candidates and ignoring artefacts.
A method for quantitative analysis of clump thickness in cervical cytology slides
Publisher: Elsevier BV
Date: 2016
DOI: 10.1016/J.MICRON.2015.09.002
Abstract: Knowledge of the spatial distribution and thickness of cytology specimens is critical to the development of digital slide acquisition techniques that minimise both scan times and image file size. In this paper, we evaluate a novel method to achieve this goal utilising an exhaustive high-resolution scan, an over-complete wavelet transform across multi-focal planes and a clump segmentation of all cellular materials on the slide. The method is demonstrated with a quantitative analysis of ten normal, but difficult to scan Pap stained, Thin-prep, cervical cytology slides. We show that with this method the top and bottom of the specimen can be estimated to an accuracy of 1 μm in 88% and 97% of the fields of view respectively. Overall, cellular material can be over 30 μm thick and the distribution of cells is skewed towards the cover-slip (top of the slide). However, the median clump thickness is 10 μm and only 31% of clumps contain more than three nuclei. Therefore, by finding a focal map of the specimen the number of 1 μm spaced focal planes that are required to be scanned to acquire 95% of the in-focus material can be reduced from 25.4 to 21.4 on average. In addition, we show that by considering the thickness of the specimen, an improved focal map can be produced which further reduces the required number of 1 μm spaced focal planes to 18.6. This has the potential to reduce scan times and raw image data by over 25%.
Deep structured learning for mass segmentation from mammograms
Publisher: IEEE
Date: 09-2015
Message from the general chair DICTA 2011
Publisher: IEEE
Date: 12-2011
DOI: 10.1109/DICTA.2011.5
On wavelet analysis of auditory evoked potentials
Publisher: Elsevier BV
Date: 05-2004
An evaluation of four parametric models of contrast enhancement for dynamic magnetic resonance imaging of the breast
Publisher: IEEE
Date: 08-2007
Analysis of the non-Markov parameter in continuous-time signal processing
Publisher: American Physical Society (APS)
Date: 10-02-2014
Why rankings of biomedical image analysis competitions should be interpreted with care
Publisher: Springer Science and Business Media LLC
Date: 06-12-2018
DOI: 10.1038/S41467-018-07619-7
Abstract: International challenges have become the standard for validation of biomedical image analysis methods. Given their scientific impact, it is surprising that a critical analysis of common practices related to the organization of challenges has not yet been performed. In this paper, we present a comprehensive analysis of biomedical image analysis challenges conducted up to now. We demonstrate the importance of challenges and show that the lack of quality control has critical consequences. First, reproducibility and interpretation of the results is often h ered as only a fraction of relevant information is typically provided. Second, the rank of an algorithm is generally not robust to a number of variables such as the test data used for validation, the ranking scheme applied and the observers that make the reference annotations. To overcome these problems, we recommend best practice guidelines and define open research questions to be addressed in the future.
Precision-recall operating characteristic (P-ROC) curves in imprecise environments
Publisher: IEEE
Date: 2006
Related Organisations
Cooperative Research Centre For Sensor Signal And Information Processing (CSSIP)
Location: Australia
Canon Information Systems Research Australia Pty Ltd
Location: Australia
Canon Information Systems Research Australia (CiSRA)
Location: Australia
University Of Plymouth
Location: United Kingdom of Great Britain and Northern Ireland
Related Funding Activities
ARC Network In Imaging Science And Technology
Start Date: 2003
End Date: 2003
Funder: Australian Research Council
View Funded ActivityMulti-modal, Multi-dimensional Virtual Microscopy For Diagnostic Quantitative Pathology
Start Date: 2007
End Date: 2009
Funder: Australian Research Council
View Funded ActivityInfrastructure For An Integrated Cognitive Neurophysiological Research Facility: Mapping The Neurobiology Of Memory And Language
Start Date: 2008
End Date: 2008
Funder: Australian Research Council
View Funded ActivityDeep Reinforcement Learning For Discovering And Visualising Biomarkers
Start Date: 2018
End Date: 2020
Funder: Australian Research Council
View Funded ActivityARC Future Fellow
Start Date: 2011
End Date: 2015
Funder: Australian Research Council
View Funded ActivityIn-road Microwave System For Traffic Monitoring And Vehicle Identification
Start Date: 2016
End Date: 2018
Funder: Australian Research Council
View Funded ActivityImproved Detection And Characterisation Of Breast Cancer Using Magnetic Resonance Imaging, And Novel Image Analysis And Pattern Recognition Techniques
Start Date: 2008
End Date: 2010
Funder: Australian Research Council
View Funded ActivityImproved Early Detection Of Breast Cancer Enabled By Grid-computing And Advanced Modelling And Visualisation Of MR Images
Start Date: 2005
End Date: 2006
Funder: Australian Research Council
View Funded ActivityMulti-modal Virtual Microscopy For Quantitative Diagnostic Pathology
Start Date: 2011
End Date: 2015
Funder: Australian Research Council
View Funded ActivityAutomated Analysis Of Multi-modal Medical Data Using Deep Belief Networks
Start Date: 2014
End Date: 2016
Funder: Australian Research Council
View Funded ActivityARC Future Fellowships - Grant ID: FT110100623
Start Date: 02-2012
End Date: 02-2016
Amount: $771,208.00
Funder: Australian Research Council
View Funded ActivityLinkage Projects - Grant ID: LP0774994
Start Date: 04-2008
End Date: 01-2009
Amount: $501,000.00
Funder: Australian Research Council
View Funded ActivityDiscovery Projects - Grant ID: DP0879250
Start Date: 2008
End Date: 12-2011
Amount: $275,000.00
Funder: Australian Research Council
View Funded ActivityLinkage Projects - Grant ID: LP160100916
Start Date: 10-2016
End Date: 03-2020
Amount: $410,000.00
Funder: Australian Research Council
View Funded ActivityLinkage Infrastructure, Equipment And Facilities - Grant ID: LE0882219
Start Date: 07-2008
End Date: 07-2009
Amount: $100,000.00
Funder: Australian Research Council
View Funded ActivityDiscovery Projects - Grant ID: DP200103582
Start Date: 12-2020
End Date: 12-2024
Amount: $506,000.00
Funder: Australian Research Council
View Funded ActivitySpecial Research Initiatives - Grant ID: SR0567196
Start Date: 09-2005
End Date: 09-2007
Amount: $55,000.00
Funder: Australian Research Council
View Funded ActivitySpecial Research Initiatives - Grant ID: SR0354604
Start Date: 2004
End Date: 12-2004
Amount: $10,000.00
Funder: Australian Research Council
View Funded ActivityDiscovery Projects - Grant ID: DP140102794
Start Date: 12-2014
End Date: 12-2018
Amount: $295,000.00
Funder: Australian Research Council
View Funded ActivityDiscovery Projects - Grant ID: DP180103232
Start Date: 10-2018
End Date: 12-2022
Amount: $387,884.00
Funder: Australian Research Council
View Funded Activity