BioMedIA
Events Search and Views Navigation
May 2016
BioMedIA Talk: Enzo Ferrante
Title: Graph-based deformable registration: slice-to-volume mapping and context specific methods Abstract: Image registration methods, which aim at aligning two or more images into one coordinate system, are among the oldest and most widely used algorithms in computer vision. A particular type of registration algorithm, known as graph-based deformable registration, has become popular during the last decade given its robustness, scalability, efficiency and theoretical simplicity. In this thesis, we extend this flexible framework to new scenarios, and propose novel methodological contributions.…
Find out more »BioMedIA Talk: Ioannis Katramados (COSMONiO)
Title: The challenges of developing an active deep-learning platform Abstract: Deep learning is usually linked to Big Data. However, there are several scientific, engineering and medical imaging problems with limited data available (e.g. rare medical conditions). Can Deep Learning prove a useful tool in such cases? COSMONiO is designing NOUS, an active deep learning platform that aims to make high-accuracy predictions using significantly smaller training datasets. NOUS aims to allow experts from any field to train neural networks without any…
Find out more »BioMedIA Talk: Leo Grady (HeartFlow)
Personalized Blood Flow Simulation from an Image-Derived Model: Changing the Paradigm for Cardiovascular Diagnostics Abstract: Coronary heart disease is the leading cause of mortality worldwide, accounting for 1/3 of all global deaths. Treatment of stable coronary heart disease is typically performed by medication/lifestyle for a lower disease burden or PCI (stenting) for a greater disease burden. The choice between these treatments is best determined by an invasive diagnostic test that measures blood flow through a diseased area. Unfortunately, this invasive…
Find out more »BioMedIA Talk: Archontis Giannakidis (Royal Brompton)
Improving clinical application of cardiac diffusion tensor MRI Abstract Magnetic resonance diffusion tensor imaging (MRDTI), also known as DTI, has emerged as a powerful non-invasive tool for mapping the orientation-dependent microanatomical organization of fibrous organs such as brain white matter and myocardium. To do so, it elegantly relates the self-diffusion of water molecules that undergo Brownian motion to proton spin relaxation MR signals. In this talk it will be shown how DTI can shed some light on the left-ventricular micro-structural remodeling following…
Find out more »July 2016
BioMedIA Talk: Purang Abolmaesumi (University of British Columbia)
Quantitative Ultrasound Imaging for Diagnosis and Intervention: A Machine Learning Approach Abstract: In recent years, quantitative ultrasound has emerged as a promising technology to improve diagnosis, increase the precision of interventions, and automate the acquisition of data specially in point-of- care settings. Advancements in machine learning and computation power have contributed significantly to these developments in quantitative imaging. In this talk, I present three main developments in our laboratory based on quantitative ultrasound: For prostate cancer diagnosis, I present our work based on the analysis…
Find out more »September 2016
BioMedIA Talk: Enrico Grisan (University of Padova)
Segmenting 2D+T ultrasound imaging data in fetus: improving the estimation of biomarkers of risky adaptations Abstract: Approximately 10% of the pregnancies are complicated by growth restriction and 7% of pregnancies are complicated by gestational diabetes mellitus (GDM). In addition, the increasing rate of obesity will likely rise the proportion of pregnancies complicated by both Type 2 diabetes mellitus and GDM. Exposure to either dysglycaemia and poor nutrition may have a long-term impact on the developing child's physiology, potentially programming for…
Find out more »January 2017
BioMedIA Talk: Olaf Ronneberger (Google DeepMind / Albert-Ludwigs-Universität Freiburg)
U-net: Convolutional Networks for Biomedical Image Segmentation Abstract: In this talk I will present our u-net for biomedical image segmentation. The architecture consists of an analysis path and a synthesis path with additional shortcut-connections. The network is trained end-to-end from scratch with a very low number of annotated images per application. The surprisingly simple strategy to train a network with such a low number of samples is a data augmentation with elastic deformations. Furthermore the u-net can segment arbitrarily large images…
Find out more »February 2017
BioMedIA Talk: Matthan Caan (Academic Medical Centre & Spinoza Centre for Neuroimaging, Amsterdam)
Quantifying MRI: from reconstruction to application in neuroimaging Abstract: To understand the origin and development of neurological disorders, Magnetic Resonance Imaging (MRI) has proven to provide valuable quantitative measures. In this overview presentation, I will touch upon several topics. Volumetric measures obtained via segmentation find their application in prenatal famine exposure, HIV and ischemic stroke. We developed a scattering transform that has no learnable parameters for computing convolutional neural networks in small patient studies. Diffusion MRI is a sensitive method…
Find out more »August 2017
BioMedIA Talk: Loic Le Folgoc (Microsoft Research)
Title: Learning structure in complex data: machine learning for medical image segmentation, registration and shape analysis
Find out more »September 2017
BioMedIA Talk: Polina Golland (MIT CSAIL)
Title: MRI of placenta: from pixels to function
Find out more »