Automatic liver detection and standardised uptake value evaluation in whole-body Positron Emission Tomography/Computed Tomography scans. (March 2018)

• Record Type: Journal Article
• Authors: Chauvie, Stephane; Bertone, Elisa; Bergesio, Fabrizio; Terulla, Alessandra; Botto, Davide; Cerello, Piergiorgio
• Is Part Of: Computer methods and programs in biomedicine. Volume 156(2018)
• Page Start: 47
• Page End: 52
• Publication Date: 2018-03
• Subjects: Positron Emission Tomography -- SUV -- Liver -- Quality control; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2017.12.026
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: The liver SUV is often used as a reference standard both in visual both in semi-quantitative evaluation of PET/CT images. We designed, implemented and validated an algorithm for the liver identification and SUV evaluation in PET/CT, to be used as part of the quality assurance process. The algorithm was tested on a large dataset (630 PET/CT scans from multi/centre trials). The true finding rate in the liver identification was 98.8% and a strong correlation between the automated and manual SUV evaluation was demonstrated. The algorithm could be used as an automatic image quality evaluation in clinical trials. Abstract: Background and objective: Standardised Uptake Value (SUV), in clinical research and practice, is a marker of tumour avidity in Positron Emission Tomography/Computed Tomography (PET/CT). Since many technical, physical and physiological factors affect the SUV absolute measurement, the liver uptake is often used as reference value both in quantitative and semi-quantitative evaluation. The purpose of this investigation was to automatically detect the liver position in whole-body PET/CT scans and extract its average SUV value. Methods: We developed an algorithm, called LIver DEtection Algorithm (LIDEA), that analyses PET/CT scans, and under the assumption that the liver is a large homogeneous volume near the centre of mass of the patient, finds its position and automatically places a region of interest (ROI) in the liver, which is used to calculate the average SUV. The algorithm was validated on a population of 630 PET/CT scans coming from more than 60 different scanners. The SUV was also calculated by manually placing a large ROI in the liver. Results: LIDEA identified the liver with a 97.3% sensitivity with PET/CT images only and reached a 98.9% correct detection rate when using the co-registered CT scan to avoid liver misidentification in the right lung. The average liver SUV obtained with LIDEA was successfully validated against its manual assessment, with no systematic difference (0.11 ± 0.36 SUV units) and a R 2 = 0.89 correlation coefficient. Conclusions: LIDEA proved to be a reliable tool to automatically identify and extract the average SUV of the liver in oncological whole-body PET/CT scans.