Neuroscience and Biomedical Imaging

Biomedical image analysis

Thanks to the continuous development of high-throughput biomedical instrumentation, modern experimental protocols in biomedical research often deal with large amounts of high-dimensional data. In principle, this condition would help to deepen the understanding of biological phenomena at the cost of massive data acquisitions. However, it is still very often difficult to reach a clear understanding of the complex structures hidden within these large datasets. This is caused by the lack of appropriate statistical tools allowing to explain the multi-faceted nature of the phenomena under investigation.

Hence, the development of advanced computational methods specifically designed for the analysis of biomedical data is becoming a common practice as they proved to be extremely helpful and in some cases fundamental for both diagnostic and research purposes, allowing to decipher complex and hidden relations in big datasets.

Addressing this computational issue, however, is not straightforward and requires investment in terms of computer vision and machine learning research. For this reason, there is a growing interest among these communities for biomedical problems that quite easily produce big amount of imaging data, and more in general structured data.

Our group exploits the strong expertise achieved in Pattern Recognition, Machine Learning, Computer Vision and Image Analysis in order to harness the various data modalities arising from biomedical problems, such as signals and time-series, 2D images and videos, 3D images and sequences, etc.

In particular, PAVIS aims at investigating the different facets of behavioral phenotyping in animals, hence providing a fertile substrate for exploring many aspects of neuroscience and biology domains: from the phenotyping of social behavior in mice to the investigation of corresponding brain correlates, from the characterization of cells and neuronal network connectivity to the analysis of the manifestations of certain mental diseases (such as schizophrenia, autism, etc.), through electrophysiology, structural and functional brain imaging (as well as genetics, possibly), including the assessment of interactions with drugs.

In this perspective, the research on biomedical data analysis currently carried out by PAVIS can be seen as organized around the three main projects below indicated.