Exploration of whole-body PET/MRI and clinical variables in type 2 diabetes for data-driven hypothesis generation

• Aim To explore the feasibility of using the automated holistic image analysis approach Imiomics in voxel-level association screening with clinical variables for hypothesis-generation in whole-body [18F]fluorodeoxyglucose (FDG) PET/MR images. Material and methods Three experimental groups consisting of healthy individuals (n=12), individuals with prediabetes (n=16) and individuals with type 2 diabetes (n=13) were examined with simultaneous whole-body PET/MRI during hyperinsulinemic euglycemic clamp. The Imiomics-framework was utilised to create correlation maps between the clinical biomarkers and PET/MRI data. Results Multiple significant moderate-strong associations were detected, the inflammatory biomarkers (P-CRP, B-Leukocytes and B-Neutrophils) were positively associated with visceral adipose tissue (VAT) volume and inversely associated with skeletal muscle Ki. B-monocytes were positively associated with VAT volume, and negatively associated with gluteofemoral SAT volume. Furthermore, insulin sensitivity (M-value) was shown to be negatively associated with brain Ki. Of the plasma lipids, HDL was positively associated with Ki in the liver, VAT and skeletal muscle. Several additional confirmatory and distinct findings are reported. Conclusions An Imiomics-based data-driven exploratory approach allows for rapid and holistic analysis of the massive image datasets generated. 

Ämnesord

MEDICIN OCH HÄLSOVETENSKAP  -- Klinisk medicin -- Radiologi och bildbehandling (hsv//swe)

MEDICAL AND HEALTH SCIENCES  -- Clinical Medicine -- Radiology, Nuclear Medicine and Medical Imaging (hsv//eng)

Nyckelord

Imiomics

PET/MRI

biomarkers

hypothesis generation

statistics

T2D

Publikations- och innehållstyp

vet (ämneskategori)

ovr (ämneskategori)