Research

Big data represents a ground-breaking opportunity for healthcare industry to improve service quality. Clinical Decision Support will help the clinician answer clinical questions quickly when information is needed, especially if the learning method is built into the Electronic Health Record (EHR). This paper delineates how unstructured data can be mined from different point of care devices to improve the prognosis and diagnosis of Chronic Kidney Disease (CKD).

 

This proposed mining system will scrape and analyze data from ubiquitous point of care devices and make predictions.  This system will help predict potential events which may be symptoms of Kidney disease or a drug interaction that could lead to kidney problems. CKD is among the top ten leading causes of death in Canada according to Statistics Canada. This research will help clinicians in the areas of chronic disease management, prenatal patients monitoring, preventative medicine, and aggregation of medical information.

Interests

  • Agile Software Development
  • Clinical Decision Support
  • Community Engagement
  • Bio-medical Informatics
  • Digital Image Processing
  • computer-aided design

Colleagues

Oluyemi Badmus

Oluyemi Badmus

Research Colleague

Sergio Camoprlinga

Sergio Camoprlinga

Supervisor and Principal Investigator

Visit him here
Parth Bramhbhatt

Parth Bramhbhatt

Research Colleague

See him here
Orlando Simpson

Orlando Simpson

Research Colleague

Great Colleagues

I have the best supervisor in the world!. My colleagues are very easy to work with. I guess his awesomeness reflects on all his students

Research Projects

  • MATLAB IMPLEMENTATION OF FACE DETECTION USING GABOR WAVELET AND SUPPORT VECTOR MACHINE

    MATLAB IMPLEMENTATION OF FACE DETECTION USING GABOR WAVELET AND SUPPORT VECTOR MACHINE

    Face detection is the first step in face recognition algorithms. It is the act of localizing a face or groups of faces in an image or sequence of images. It is also used to localize faces in a video sequence. After this step various face recognition algorithm can now be applied to enable verification or identification. The detection process is a challenging topic in Biometrics because the lightning condition and other environmental factors in the detection process can affect the performance of the recognition algorithm.

    This could be a full decription about the project. You can put media, hyperlinks, etc here

    math-photography-hd-wallpaper-1920x1080-9897

     

    Heading

    some hyperlink

  • IMPLEMENTATION AND OPTIMIZATION OF PCA ALGORITHM FOR FACE RECOGNITION ON THE GPU

    IMPLEMENTATION AND OPTIMIZATION OF PCA ALGORITHM FOR FACE RECOGNITION ON THE GPU

    Parallel computing project

    Face recognition is the task of matching an already detected face as a known face or unknown face and in some cases, identifying whose face was detected. It is one of the oldest and most popular modalities used in biometrics systems. The face recognition system takes an input face and performs recognition against a database of faces. It involves face detection, tracking, and rendering. This paper delineates the implementation of face recognition using the PCA (Principal Component Analysis) algorithm.

     

    The PCA algorithm requires computational intensive matrix calculations and the performance declines as the template dataset grows. A simple solution is to implement the algorithm on a hardware that can provide significant latency and throughput. The PCA algorithm is used for face recognition by finding out the principal components in the data set [1]. The CUDA (Compute Unified Device Architecture) architecture, developed by NVIDIA, will be used to demonstrate the dramatic increase in performance of this algorithm by utilizing the power of the GPGPU.

     

    In this paper, a CUDA C++ implementation of PCA face recognition algorithm will be employed and the results will be compared to the same algorithm used on the CPU. The many core capabilities of the GPGPU will be used to split the PCA phase into parallel dimensions and each dimension will be processed in parallel. This will be a cheaper alternative to map-reduce approach which involves multi-agent systems and comes with a significantly higher cost.

     

  • IMPROVING THE PROGNOSIS AND DIAGNOSIS OF CHRONIC KIDNEY DISEASE USING KNOWLEDGE BASED MINING

    IMPROVING THE PROGNOSIS AND DIAGNOSIS OF CHRONIC KIDNEY DISEASE USING KNOWLEDGE BASED MINING

    Thesis topic

    Abstract

    Big data represents a ground-breaking opportunity for healthcare industry to improve service quality. Clinical Decision Support will help the clinician answer clinical questions quickly when information is needed, especially if the learning method is built into the Electronic Health Record (EHR). This paper delineates how unstructured data can be mined from different point of care devices to improve the prognosis and diagnosis of Chronic Kidney Disease (CKD).

     

    This proposed mining system will scrape and analyze data from ubiquitous point of care devices and make predictions.  This system will help predict potential events which may be symptoms of Kidney disease or a drug interaction that could lead to kidney problems. CKD is among the top ten leading causes of death in Canada according to Statistics Canada. This research will help clinicians in the areas of chronic disease management, prenatal patients monitoring, preventative medicine, and aggregation of medical information.

  • USING FUZZY TECHNIQUES FOR INTENSITY TRANSFORMATION

    USING FUZZY TECHNIQUES FOR INTENSITY TRANSFORMATION

    Digital Image processing project

    Introduction

    Fuzzy sets are used to blend knowledge into the solutions of problems, whose formulation is based on imprecise concepts.

    Let B be a set of elements (objects) with a generic element of B denoted as b; that is B = {b}. This set is called a universe or discourse. A fuzzy set A in B is described by a membership function  that matches a real number in [0 1] with each element of B. The value of  at z is a grade of membership of z in A: the closer it is to one, the higher the grade of membership is. In ordinary (crisp) sets, an element either belongs or does not belong to a set…