Research interests

• Healthcare computer vision clinical applications.

• Trustworthy medical imaging algorithms for safe use in healthcare.

• Label-efficient and easy to use medical imaging algorithms.

Technical Topics

• Quality control of deep learning algorithms

• Label-efficient segmentation

• Robustness of deep learning algorithms

• Machine learning and deep learning for medical image analysis

Clinical topics

This is the current list of clinical topics:

• Volumetric body composition assessment in CT

• Handheld Ultrasound applications

• Forensic identification based on x-ray dental images

• Fetal MRI

hVision lab mission

Medical images play a key role in healthcare, encapsulating the essence of “a picture is worth a thousand words.” They enable us to see inside the human body and examine microscopic cells, aiding in various health procedures such as diagnosis, screening, and patient management. Recent advancements in Machine Learning (ML), particularly in Deep Learning (DL), have introduced new and exciting capabilities to the field of medical imaging. hVision lab mission is to harness the power of these cutting-edge ML techniques to unlock the full potential of medical images and enhance healthcare outcomes.

Despite significant ML advancements in medical imaging, challenges remain before these algorithms become routine in clinical practice. They often struggle to generalize across different sites and scanners, requiring more robust models and evaluation methods. Current algorithms focus on specific clinical questions, making it costly to develop solutions for every need, highlighting the necessity for cost-effective algorithms with minimal annotated data. Additionally, workflow tools are essential for seamless integration into clinical practice. hVision lab aims to address these issues and develop clinically relevant healthcare computer vision algorithms.