Development of AI algorithms for segmentation and measurement in intestinal ultrasound - BC-953

Project type: Research
Desired discipline(s): Engineering - biomedical, Engineering, Engineering - computer / electrical, Computer science, Mathematical Sciences
Company: Dova Health Intelligence
Project Length: Flexible
Preferred start date: As soon as possible.
Language requirement: English
Location(s): BC, Canada
No. of positions: 1
Desired education level: PhDPostdoctoral fellow
Open to applicants registered at an institution outside of Canada: No

About the company: 

Dova Health Intelligence is a company that develops AI products to diagnose disease activity in the gastrointestinal tract. We specialize in computer vision solutions in multiple modalities such as endoscopy and ultrasound. We are focused on enhancing clinical practice, and we also support pharmaceutical research (clinical trials).

Describe the project.: 

This project aims to refine and clinically validate an innovative AI-powered algorithm, DovaSound, for the precise detection and quantification of inflammatory bowel disease (IBD) activity, specifically Crohn’s disease and ulcerative colitis, using ultrasound imaging. Building upon existing proof-of-concept, this phase focuses on enhancing the algorithm's performance and robustness through advanced computer vision techniques.

Our approach leverages deep learning to analyze ultrasound-based imaging metrics, enabling objective and quantitative assessment of IBD activity. This will be achieved by expanding the training dataset with diverse patient populations and refining the algorithm's feature extraction and classification capabilities. The target population encompasses patients with and without IBD across various clinical settings, including general practice, community clinics, hospitals, and research institutions.

The project's clinical implementation strategy involves seamlessly integrating the AI software into existing intestinal ultrasound (IUS) workflows, empowering clinicians with a rapid, non-invasive diagnostic tool. This will improve diagnostic accuracy, facilitate timely treatment adjustments, and enhance patient management.

Key deliverables include a validated AI algorithm with improved accuracy and generalizability, a user-friendly software interface for IUS integration, and comprehensive clinical validation data. Our team, comprised of experts in medical imaging, artificial intelligence, and gastroenterology, is uniquely positioned to achieve these objectives.

Grant funding will primarily support the expansion of the training dataset, algorithm refinement, software development, and clinical validation studies. The successful completion of this project will lead to a significant advancement in IBD management, offering a cost-effective and accessible diagnostic solution for improved patient outcomes.

Required expertise/skills: 

Minimum Qualifications:
• Ph.D candidate in Computer Science, Engineering, Mathematics, Physics, or a related field.
• 3+ years of experience developing machine learning/computer vision models
• Experience working with cutting edge computer vision models like DINOv2, V-JEPA, ViT, etc.
• Proficiency in Python and experience with data processing frameworks (e.g., Spark, PySpark).
• Expertise with common data and machine learning libraries such as NumPy, Pandas, and PyTorch.
• Expertise with deploying ML models onto embedded devices (C++, CUDA)
• Familiarity with containerization (Docker) and orchestration technologies (Kubernetes)
• Hands-on experience with Databricks and AWS infrastructure.
• Excellent problem-solving skills and the ability to thrive in a fast-paced startup environment.
• Strong communication skills to effectively convey complex technical concepts to non-technical stakeholders.