Projects
Automated Glioma Segmentation
Gliomas are one of the most common brain tumors. Magnetic resonance imaging (MRI) is used to diagnose Gliomas, especially multiparametric MRI using diffusion-weighted imaging or perfusion-weighted imaging. We are developing an automated model that can measure the size of the Glioma and predict the specific genetic features in the Glioma based on the mp MRI images.
Prediction Model for Ambulatory Blood Pressure Monitoring
With four machine learning methods – support vector machine, decision tree, random forest, and extreme gradient boosting (XGBoost) – BP patterns were accurately predicted. Missing data were imputed using K nearest neighbours. Models were developed in a record time and high quality to aid in clinical delivery and research.
AI-based system to detect morphological alterations of the condyle using CBCT scans
DenseNet network architecture provided in a Keras framework was used to detect morphological condylar alterations using large volumes of CBCT scans within a shorter time period. This enables researchers and governmental databases to gain epidemiological data on condylar abnormalities relatively faster.
Medical Practitioner Encounter Transcriber
An AI-powered medical transcriber application that creates and manages the transcription of conversation between medical practitioner and the patients. The application listens to the conversation and automatically transform the voice into transcripts for the medical practitioner’s review and approval using state-of-art technologies supported by Natural Language Processing and Machine Learning.
Extracting insights from medical literature for Corona Virus Disease-2019 (COVID-19) using a biomedical natural language processing application
In collaboration with multi-disciplinary team of data scientists, clinicians, medical researchers and software engineers, developed an innovative NLP platform combining an advanced search engine with a biomedical named entity recognition extraction package to extract and analyse information relating to COVID-19 clinical risk factors from COVID-19 open research dataset (CORD-19) comprising over 45,000 scholarly articles.
COVID-19 Algorithm Registry
Completion of a pilot to screen literature (non-systematic collection) for COVID-19 related algorithms using a customised checklist and then detailing select studies in a spreadsheet (registry). Based on this process, we reviewed over 80 manuscripts and selected 23 studies for inclusion in the registry.