Medical Imaging @ IPLAB

Medical Imaging encompasses the science and technologies dedicated to acquiring images of the human body—whether as a whole or in detailed anatomical regions—for clinical interpretation, diagnosis, or intervention. It is a cornerstone of modern healthcare, providing insights into the progression of diseases and enabling personalized treatment strategies.

In this context, the development of robust and accurate computational methods is crucial. These methods aim to extract clinically relevant information from complex imaging data, offering tools to monitor disease progression, predict outcomes, and support medical decision-making processes.

At the Image Processing LABoratory(IPLAB), our mission is to contribute to this evolving field by designing innovative methodologies and systems tailored to specific challenges in medical imaging. Below, you can explore highlights from our recent research endeavors:

• Multiple Sclerosis Lesion Segmentation: Developing advanced segmentation techniques for multiple sclerosis lesions, focusing on integrating diffusion models, brain parcellation, and attention mechanisms. Key initiatives include the MSLesSeg competition, innovative pipeline integrations, and improvements to traditional architectures using state-of-the-art methods.
• Brain Age Estimation: Creation of models for estimating brain age capable of handling heterogeneous MRI sequences and resolutions. These models aim to minimize acquisition variability and establish correlations between brain predicted age difference (PAD) and Alzheimer’s cognitive measures.
• Atrophy Mapping: Design of U-net-based architectures for rapid and efficient brain atrophy mapping. This research simplifies multi-step processes into a single-step solution, significantly reducing computation time while maintaining state-of-the-art performance.
• Brain Progression Modeling: Development of temporally-aware diffusion models to predict neurodegenerative brain progression. The models generate accurate future MRI scans from baseline data, achieving significant improvements in similarity metrics.
• Medical Applications Beyond Neurology:
  • COVID-19 Detection: Implementation of convolutional neural networks with attention mechanisms for the accurate detection of COVID-19 from CT scans, contributing to faster and more reliable diagnostics.
  • Periprosthetic Joint Infection Detection: Development of ResNeSt-based convolutional neural networks for early and accurate detection of periprosthetic hip infections using CT scan data.

The IPLAB team is dedicated to pushing the boundaries of medical imaging research, collaborating with clinicians and industry partners to ensure our solutions address real-world clinical needs.

Whether you're a researcher, clinician, or industry expert, we invite you to explore our work and join us in advancing the science of medical imaging.