EMATA: a toolbox for the automatic extraction and modeling of arterial inputs for tracer kinetic analysis in [18F]FDG brain studies

• Published in: EJNMMI physics Vol. 11; no. 1; pp. 105 - 21
• Main Authors: De Francisci, Mattia, Silvestri, Erica, Bettinelli, Andrea, Volpi, Tommaso, Goyal, Manu S., Vlassenko, Andrei G., Cecchin, Diego, Bertoldo, Alessandra
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 24.12.2024; Springer Nature B.V; SpringerOpen
• Subjects: [18F]FDG; Applied and Technical Physics; Biomarkers; Blood; Brain; Computational Mathematics and Numerical Analysis; Datasets; Engineering; Fluorine isotopes; IDIF; Imaging; Kinetic modeling; Medical imaging; Medicine; Medicine & Public Health; Modelling; Nuclear Medicine; Original Research; Positron emission; PVE; Quantitative analysis; Radioactive tracers; Radiology; IDIF; [; Kinetic modeling; PVE; F]FDG; [18F]FDG
• ISSN: 2197-7364; 2197-7364
• DOI: 10.1186/s40658-024-00707-2

Purpose PET imaging is a pivotal tool for biomarker research aimed at personalized medicine. Leveraging the quantitative nature of PET requires knowledge of plasma radiotracer concentration. Typically, the arterial input function (AIF) is obtained through arterial cannulation, an invasive and technically demanding procedure. A less invasive alternative, especially for [ 18 F]FDG, is the image-derived input function (IDIF), which, however, often requires correction for partial volume effect (PVE), usually performed via venous blood samples. The aim of this paper is to present EMATA: Extraction and Modeling of Arterial inputs for Tracer kinetic Analysis, an open-source MATLAB toolbox. EMATA automates IDIF extraction from [ 18 F]FDG brain PET images and additionally includes a PVE correction procedure that does not require any blood sampling. Methods To assess the toolbox generalizability and present example outputs, EMATA was applied to brain [ 18 F]FDG dynamic data of 80 subjects, extracted from two distinct datasets (40 healthy controls, 40 glioma patients). Additionally, to compare with the reference standard, quantification using both IDIF and AIF was carried out on a third open-access dataset of 18 healthy individuals. Results EMATA consistently performs IDIF extraction across all datasets, despite differences in scanners and acquisition protocols. Remarkably high agreement is observed when comparing Patlak’s K i between IDIF and AIF (R 2 : 0.98 ± 0.02). Conclusion EMATA proved adaptability to different datasets characteristics and the ability to provide arterial input functions that can be used for reliable PET quantitative analysis.