Education

Post-doctoral fellowship
Western University – IBM Research Collaboration

This video shows the classification of ultrasound anatomical regions at a rate of 100 classifications per second. The convolutional neural network ran in the IBM TrueNorth chip.

My post-doctoral fellowship (codename PERSEUS for PERceptive UltraSound) was all about using AI to identify anatomical regions in ultrasound images of the spine. This is a very hard task for any human (and most machines), but worth exploring as it could help minimize damage to the spine in procedures such as epidural injections and spinal taps (lumbar punctures).

ICCV PosterShort Presentation

Ph.D. in Biomedical Engineering
Western University, Canada

These maps show brain regions affected by the presence of temporal lobe epilepsy (MTS and non-MTS). The color code indicates the type of MRI where a change is detected. Improving epilepsy location/lateralization is critical for surgical planning to relieve seizures.

My doctoral research focused on diagnosing temporal lobe epilepsy using magnetic resonance imaging (MRI). You can find my thesis here. For this, I studied quantitative imaging (pure T1 and T2 maps) along with diffusion tensor imaging (DTI) and I performed machine learning analyses. My algorithms were implemented using python numpy and the scikit-learn library.  The results were published in the Computerized Medical Imaging and Graphics Journal

Master of Engineering in Computer and Systems Engineering
Universidad de Los Andes,  Colombia

Marcela Hernandez, Ph.D. shows one of the
3D printed models that we created to study carotid stenosis.

My master’s project consisted of the measurement of atherosclerotic plaque from CT images, in the context of cardiovascular disease. For this purpose, I implemented the fast marching algorithm in C++ to model and 3D print vascular models.  With my interest in software design, my code followed the OSGI specification. Since OSGI is a Java-native component-based architecture,  I wrote and tested the corresponding  C++ framework. My master thesis received the “Gonzalo Esguerra” national award conferred by the Colombian Radiology Association.

Electronic Design and Automation Engineering
Universidad de La Salle, Colombia

My final year project was a robot that was able to follow a person’s gaze and then use this information to position the cursor on a computer screen. This final project consisted of a pan-tilt mechanism, the control, and communications circuitry, and software. The relevant code was written using the Java Media Framework and LabView.