DAUIN - GR-24 - SMILIES - reSilient coMputer archItectures and LIfE Sci
Sonification of high-dimensional biological data
External reference persons ROBERTA BARDINI
Thesis type RESEARCH
Description Visualization is the main approach to provide accessibile representations of complex information in different contexts, from science communication exhibitions to data exploration and analysis. Yet, the graphical representation of complex and multidimensional data often fails to provide the clear and immediate insights that we expect from graphical representations of simpler data. This is the case for complex biological data. For example, multi-omic analysis links single cells to multiple omic profiles, thus each cell from a dataset corresponds to two or more sets of high-dimensional, heterogeneous data, whose hidden structures and relations are very difficult to highlight. Similarly, medical images contain rich information, but often with details that are hidden to the human eye, such as vessels, organs or tumors that are not possible to directly spot. Such hidden details would prove crucial to support better clinical decisions. These, among other examples, are applications where high-dimensional datasets must find a more intuitive, comprehensive and rich representation. Responding to this challenge, the discipline of sonification explores and develops methods for mapping complex data into an auditory representation that can be readily comprehended by human listeners. Sonification can benefit data exploration and inspection providing alternative, enhanced or simply different access to information. Auditory displays support high temporal and frequency resolution, multiple parallel audio streams, perception of complex sound patterns as a whole and prioritization of specific sounds while maintaining awareness of the background. This allows to understand both the overall shape and the fine granular structure of the complex data from biological systems, where structure exists at many different scales, leading to better hypothesis generation in data analysis. The proposed thesis work aims at the design and development of a sonification approach to support the inspection, exploration and analysis of high-dimensional data from systems biology and biomedical applications.
Deadline 15/11/2023 PROPONI LA TUA CANDIDATURA