Mitacs Accelerate International- Abroad Internship in AI validation
Institution: UBC- Okanagan/BC Cancer – Kelowna
Contact name and title: Dr. Rasika Rajapakshe, Senior Medical Physicist
Length of work term: 8 months full time
Start date: January 2021
Funding: $20,000
What we’re looking for: Undergraduate or Graduate student in Computer Science/ Data Science.
Project: Validating Artificial Intelligence Algorithms for Breast Cancer Detection
Recent advances in deep learning have kick-started a new generation of medical imaging research and development. Research teams have applied these techniques to screening mammographic images and predicting breast cancer risk based on breast density. In particular, there is current interest in using AI algorithms as an independent, unbiased “second reader” in conjunction with radiologists. Using these algorithms as independent peer reviewers has the potential to outperform radiologists alone while reducing radiologist workload. AI algorithms used in this way could also help make risk predictions about the likelihood of breast cancer
In this study, we are planning to evaluate the performance of a commercial Artificial Intelligence (AI) system for breast cancer detection using the digital mammograms from BC Cancer Breast Screening Program. We plan to compare the predictions of the AI systems to those made by breast screening radiologists in routine clinical practice. Digital screening mammograms and associated outcomes, demographics, and risk factors will be extracted from approximately 150,000 participants who underwent screening mammograms in BC with a minimum of one year of follow up. The de-identified images will be fed to the AI algorithm running on a computer with GeForce RTX 2080 Ti Graphical Processing Unit (GPU). We will also use Amazon Web Services (AWS) S3 bucket technology encrypt and to store ~30 TB of image data and will also use EC2 instances to validate the AI algorithm in parallel. AWS Snowball device will be used to physically transfer the data to the Canadian instance of AWS cloud. The AI model performance will be evaluated using Area Under the Curve (AUC) of Receiver Operating Characteristic (ROC) methodology.
Ideal candidate would be a masters or undergraduate student in Computer Science or Data Science with Python programming experience. Experience in handling medical image data in DICOM format and/or designing/operating deep learning algorithms would be an asset. Most of the work (data curation and AI algorithm operation) for this project would be conducted remotely, but occasional visits to the BC Cancer – Kelowna site for managing hardware resources would be required. The ideal candidate will be physically present and available in Kelowna. As required by Mitacs Accelerate International – Abroad conditions, only Canadian citizens or permanent residents are eligible for this position.
To apply, please email a cover letter and resume to amanda.wanner@bccancer.bc.ca. Hiring will be on a rolling basis.
