Research Lines

Cardiovascular and pulmonary imaging

This research line deals with the development of techniques for early detection of cardiac pathologies from ultrasound and magnetic resonance images. Systems have been proposed for the characterization of myocardial movement based on the spatio-temporal registration of image sequences. Other studies tried to identify the myocardial substrate of arrhythmias, which allows us to assess the success of possible therapies. Using new segmentation algorithms, it has also been possible to study the cardiac chambers or the displacement of the mitral and aortic valves in tomography image sequences (4D-CTA).

Machine learning algorithms have also been proposed to identify biomarkers for the diagnosis and prognosis of pulmonary pathologies from tomography images.

Collaborations

Some publications

Research Lines

Optical and microscopic imaging techniques

From optical coherent tomography (OCT) images and using advanced image processing techniques, screening and monitoring systems for high prevalence pathologies, such as glaucoma and some skin cancers, have been proposed and validated. Low-cost and non-invasive systems have also been proposed for neutrophil counting in chemotherapy patients. Likewise, image processing methods have been proposed for the study of embryonic development using advanced microscopy images (harmonic images, two photons, SPIM, etc.).

Collaborations

Some publications

Research Lines

Image-guided surgery and radiotherapy

Multimodal image processing algorithms have been proposed for the planning of various surgical cases (epilepsy, liver tumors, heart diseases, etc.). A methodology for planning and dosimetry in intraoperative radiotherapy has also been proposed, including the registration of preoperative tomography images and intraoperative projective radiographs.

Collaborations

Some publications

Research Lines

Fetal and neonatal imaging

New techniques for magnetic resonance image acquisition, reconstruction and processing have been proposed, providing support for image-based analysis of neurodevelopment including clinical applications and exploratory studies. Machine learning and statistical models have been developed for diagnosis, prognosis and understanding of perinatal development based on anatomical, micro-structural and functional brain information as well as whole-body scans.

Collaborations

Some publications

Research Lines

Image processing

This activity has a direct impact on the development of the remaining research lines. Spatio-temporal image registration (or superposition) algorithms have been developed to characterize myocardial movement or to compensate for respiratory movement during thoracic imaging. 2D-3D registration and fusion algorithms have also been proposed to create high-quality three-dimensional volumes from projection images, useful in applications as diverse as microscopy imaging or coronary angiography. Work has also been done on the combination of anatomical and functional images in multimodality registration.

Some publications

Research Lines

Computation of image-based biomarkers

In this research line, work has been done on the development of algorithms for the automatic extraction of information from images and clinical data. The pharmacokinetic analysis of perfusion imaging (DCE-MRI), or the extraction of characteristics (radiomics) using learning algorithms (including deep learning) can be highlighted. As an example, quantitative biomarkers derived from clinical images have been proposed for the prediction of response to immunotherapy in oncology (non-small cell lung cancer and gliobastoma).

Collaborations

Some publications

Research Lines

New technologies for image acquisition

We have worked on low-cost systems for the detection of global diseases, such as microscopy image acquisition devices based on 3D printing and mobile phones, with the aim of to contribute to the diagnosis in developing countries of eye diseases, childhood tuberculosis, intestinal parasites, etc. Along these lines, work has also been done on applications such as Malaria Spot and Tuber Spot, where citizens can help researchers diagnose diseases through games.

On the other hand, new data acquisition systems have been designed, built and validated for PET (positron emission tomography) and OCT (optical coherent tomography) tomography) using integrated digital architectures (system-on-chip).

Work has also been done on tomographic reconstruction, especially on fast methods of 3D statistical reconstruction for high-resolution PET cameras, as well as in experimental proton therapy systems.

Collaborations

Some publications