High Performance Computing Applied to Pathology Informatics

This search topic investigates run-time systems and optimizations for large-scale execution of pathology image analysis using modern parallel processors. Digital Pathology is an important application domain that investigates the disease morphology at the cellular and sub-cellular level and can reveal important clues about disease mechanisms that are not possible to capture by other imaging modalities. This information may be used, for instance, to perform patient survival analysis, to correlate disease progression with genomic data, in precision treatments, etc. Analysis of a whole slide tissue image is computation intensive because of the complexity of analysis operations and data sizes – a color image can have 120Kx120K pixels and about one billion nuclei may be found in an image. Processing these images efficiently and, consequently, enabling studies in large datasets is our main goal in this research direction.

Large-Scale Content Based Multimedia Retrieval

This topic addresses the problem of performing efficient similarity search in high-dimensional spaces for online multimedia retrieval applications, such as image search engines. With the popularity of these applications, they are required to handle very large and increasing datasets, while keeping the response time low. This problem is worsened in the context of online applications, mostly due to the fact that load on these systems vary during the execution according to the users demands. Therefore, we are investigating the parallelization of modern indexing algorithms in order to provide these systems with the ability of searching in extreme-large (web scale) datasets.

Pathology Image Analysis

In this area, we are performing integrative analysis of a variety of tumors to address scientific questions about disease onset and progression. We intensively made use of microscopy imaging data, MRI image features and omics data for large-scale correlation studies. The actual areas of research in this topic include the study and development of deep learning based approaches for image classification and segmentation, the proposition of new active learning strategies to mitigate the cost of creating training datasets, and the development of methodologies to quantify output uncertainty on the scientific results attained with this class of applications.