A5004046579- Retinal Imaging and Analysis
- Psoriasis: Treatment and Pathogenesis
- AI in cancer detection
- Optical Coherence Tomography Applications
- Human Mobility and Location-Based Analysis
- Complex Network Analysis Techniques
- Digital Imaging for Blood Diseases
- Radiomics and Machine Learning in Medical Imaging
- Cutaneous Melanoma Detection and Management
- Glaucoma and retinal disorders
- melanin and skin pigmentation
- Data Management and Algorithms
- Healthcare and Environmental Waste Management
- Renal cell carcinoma treatment
- Municipal Solid Waste Management
- Systemic Lupus Erythematosus Research
- Cell Image Analysis Techniques
- Data-Driven Disease Surveillance
- Advanced Neural Network Applications
- Retinal Diseases and Treatments
- Brain Tumor Detection and Classification
- Biometric Identification and Security
- Topic Modeling
- Skin Diseases and Diabetes
- Infrared Thermography in Medicine
Monash University
2023-2024
Australian Regenerative Medicine Institute
2023
Monash University Malaysia
2022
IBM Research - Australia
2019-2021
The University of Melbourne
2015-2021
Benha University
2012-2013
The purpose of this study is to develop an intelligent remote detection and diagnosis system for breast cancer based on cytological images. First, paper presents a fully automated method cell nuclei segmentation in locations the image were detected with circular Hough transform. elimination false-positive (FP) findings (noisy circles blood cells) was achieved using Otsu's thresholding fuzzy c-means clustering technique. boundaries accomplished application marker-controlled watershed Next,...
The direct analysis of 3D Optical Coherence Tomography (OCT) volumes enables deep learning models (DL) to learn spatial structural information and discover new bio-markers that are relevant glaucoma. Downsampling input is the state-of-art solution accommodate for limited number training as well available computing resources. However, this limits network's ability from small retinal structures in OCT volumes. In paper, our goal improve performance by providing guidance DL model during order...
Abstract A key challenge in mining social media data streams is to identify events which are actively discussed by a group of people specific local or global area. Such useful for early warning accident, protest, election breaking news. However, neither the list nor resolution both event time and space fixed known beforehand. In this work, we propose an online spatio-temporal detection system using that able detect at different resolutions. First, address related unknown spatial events,...
This paper presents the challenge report for 2021 Kidney and Tumor Segmentation Challenge (KiTS21) held in conjunction with international conference on Medical Image Computing Computer Assisted Interventions (MICCAI). KiTS21 is a sequel to its first edition 2019, it features variety of innovations how was designed, addition larger dataset. A novel annotation method used collect three separate annotations each region interest, these were performed fully transparent setting using web-based...
Psoriasis is a chronic skin condition. Its clinical assessment involves four measures: erythema, scales, induration, and area. In this paper, we introduce scale severity scoring framework for two-dimensional psoriasis images. Specifically, leverage the bag-of-visual words (BoVWs) model lesion feature extraction using superpixels as key points. BoVWs based on building vocabulary with specific number of (i.e., codebook size) by clustering algorithm some local features extracted from...
Psoriasis is a chronic skin disease that assessed visually by dermatologists. The Area and Severity Index (PASI) the current gold standard used to measure lesion severity evaluating four parameters, namely, area, erythema, scaliness, thickness. In this context, psoriasis segmentation required as basis for PASI scoring. An automatic method leveraging multiscale superpixels [Formula: see text]-means clustering outlined. Specifically, we apply superpixel strategy on CIE-[Formula: text] color...
Presence of hair in psoriasis skin images may adversely affect the extraction features required for computer aided analysis, thus compromise detection and diagnostic results. Therefore, diagnosis to be accurate, it is vitally important remove hair, if exists, from preprocessing stage. This paper presents, first time, a removal algorithm 2D images. The process starts with markers algorithm, where shape are extracted binary input image. outcome this step all objects that obscure image lesions...
In this paper, we present a detailed comparison study of skin segmentation methods for psoriasis images. Different techniques are modified and then applied to set images acquired from the Royal Melbourne Hospital, Melbourne, Australia, with aim finding best technique suited application We investigate effect different colour transformations on detection performance. respect, explicit thresholding is evaluated three decision boundaries (CbCr, HS rgHSV). Histogram-based Bayesian classifier...
Identifying events happening in a specific locality is important as an early warning for accidents, protests, elections or breaking news. However, this location-specific event detection challenging the locations and types of are not known beforehand. To address problem, we propose online spatio-temporal system using social media that able to detect at different time space resolutions. First, exploit quad-tree method split geographical into multiscale regions based on density data. Then,...
The significant increase in hazardous waste generation Australia has led to the discussion over incorporation of artificial intelligence into management system. Recent studies explored potential applications various processes managing waste. However, no study examined use text mining sector for purpose informing policymakers. This developed a living review framework which applied supervised classification and techniques extract knowledge using domain literature data between 2022 2023....
Skin segmentation, which involves detecting human skin areas in an image, is important process for disease analysis. The aim of this paper to identify the regions a newly collected set psoriasis images. For purpose, we present committee machine learning (ML) classifiers. A training first by using pixel values five different color spaces. Experiments are then performed investigate impact both size and number features per pixel, on performance each classifier. classifiers constructed combining...
Geotagged tweets serve many important applications, e.g., crisis management, but only a small proportion of are explicitly geotagged. We propose Convolutional Neural Network (CNN) architecture for geotagging to landmarks, based on the text in and other meta information, such as posting time source. Using dataset Melbourne tweets, experimental results show that our algorithm out-performed various state-of-the-art baselines.
The presence of nipples in human trunk images is considered a main problem psoriasis images. Existing segmentation methods fail to differentiate between lesions and due the high degree visual similarity. In this paper, we present an automated nipple detection method as important component for severity assessment psoriasis. First, edges are extracted using Canny edge detector where smoothing sigma parameter automatically customized every image based on level. Then, circular hough transform...
Low-dose positron emission tomography (PET) image reconstruction methods have potential to significantly improve PET as an imaging modality. Deep learning provides a promising means of incorporating prior information into the problem produce quantitatively accurate images from compromised signal. learning-based for low-dose are generally poorly conditioned and perform unreliably on with features not present in training distribution. We method which explicitly models deep latent space using...
Deep learning-based techniques have potential to optimize scan and post-processing times required for MRI-based fat quantification, but they are constrained by the lack of large training datasets. Generative models a promising tool perform data augmentation synthesizing realistic However no previous methods been specifically designed generate datasets quantitative MRI (q-MRI) tasks, where reference maps variability in scanning protocols usually required. We propose Physics-Informed Latent...
Pancreatic steatosis and metabolic-dysfunction-associated steatotic liver disease are characterised by fat accumulation in abdominal organs, but their correlation remains inconclusive. Recently proposed deep learning (DL) for proton density fraction (PDFF) estimation, which quantifies organ fat, has primarily been assessed quantifying fat. This study aims to validate DL models pancreatic PDFF quantification compare pancreas content. We evaluated three models—Non-Linear Variables Neural...
In optical coherence tomography (OCT) volumes of retina, the sequential acquisition individual slices makes this modality prone to motion artifacts, misalignments between adjacent being most noticeable. Any distortion in OCT can bias structural analysis and influence outcome longitudinal studies. The presence speckle noise characteristic imaging leads inaccuracies when traditional registration techniques are employed. Also, lack a well-defined ground truth supervised deep-learning ill-posed...