Searching for Linear Dependencies between Heart Magnetic Resonance Images and Lipid Profiles

Marko Sysi-Aho, Juha Koikkalainen, Jyrki Lötjönen, Tuulikki Seppänen-Laakso, Hans Söderlund, Tiina Heliö, Matej Orešič

    Research output: Chapter in Book/Report/Conference proceedingChapter or book articleScientificpeer-review

    Abstract

    Information derived from "omics" data in life science research are frequently limited by specific spatial or temporal scales these data describe. As a case study of integrating physiological and molecular data in human, here we study associations between the heart magnetic resonance images and serum lipidomic profiles. In the best case, such associations could help infer the physiologic state of the heart from a blood serum sample without need to use expensive imaging techniques. Strong marginal and partial correlations are found between the lipid profiles and parameters derived from the heart images. Regression analyses are applied to study these dependencies in more detail. This study demonstrates the feasibility of mapping lipid profiles to heart images, and thus combining information from two very different scales, small molecules and macroscopic physiologic features. Such mappings could be generalized to other "omics" data as well to complete our picture of the holistic function of a living organism.
    Original languageEnglish
    Title of host publicationAlgorithms and Applications
    Subtitle of host publicationEssays Dedicated to Esko Ukkonen on the Occasion of His 60th Birthday
    EditorsTapio Elomaa, Heikki Mannila, Pekka Orponen
    Place of PublicationBerlin
    PublisherSpringer
    Pages232-243
    ISBN (Electronic)978-3-642-12476-1
    ISBN (Print)978-3-642-12475-4
    DOIs
    Publication statusPublished - 2010
    MoE publication typeA3 Part of a book or another research book

    Publication series

    SeriesLecture Notes in Computer Science
    Volume6060
    ISSN0302-9743

    Fingerprint

    Dive into the research topics of 'Searching for Linear Dependencies between Heart Magnetic Resonance Images and Lipid Profiles'. Together they form a unique fingerprint.

    Cite this