SUV phantom measurements in integrated PET/MR system without using CT for attenuation correction

• Published in: The Journal of nuclear medicine (1978) Vol. 58; p. 1347
• Main Authors: Soret, Marine, Maisonobe, Jacques-Antoine, Khalife, Maya, Lebon, Vincent, Jenny, Catherine, Kas, Aurelie
• Format: Journal Article
• Language: English
• Published: New York Society of Nuclear Medicine 01.05.2017
• Subjects: Alternating current; Attenuation coefficients; Classification; Computation; Computed tomography; Contrast agents; Cylinders; Data acquisition; Health care; Human tissues; Image acquisition; Image processing; Image segmentation; Magnetic resonance imaging; Mathematical analysis; Measurement; Moisture content; NMR; Nuclear magnetic resonance; Nuclear medicine; Positron emission; Positron emission tomography; Quality control; Systems integration; Tomography; Water content
• ISSN: 0161-5505; 1535-5667

Objectives: Attenuation correction (AC) is recognized as one of the main issue in PET/MR. For patient imaging, the currently implemented AC method is based on MR data acquisition with tissue segmentation and classification. This method is not adapted for phantoms as they are composed of plastic and filled with water, which have different density from human tissues. Furthermore, in 3T MRI, the amount of water may induce artifacts when scanning large phantoms, and plastic walls are not visible. AC implemented on the PET/MR for Standardized Uptake Value (SUV) phantom uses previously acquired CT derived attenuation map. Therefore, the PET/MR acquisitions as performed clinically cannot be fully controlled when performing the quality control of the SUV with the manufacturer's method. The objective of this study was to assess the accuracy of a new method to validate SUV in phantoms without using CT for AC. Methods: The SIGNA PET/MR (GE Healthcare) was used. SUV phantom is a fluid fillable cylindrical phantom of 5.64 L. First, SUV phantom was filled with water and attenuation map was derived from a predefined CT as implemented by the manufacturer to measure SUVCT. Secondly, the clinical protocol was used: DIXON MR sequence was performed simultaneously with PET acquisition. Attenuation was calculated automatically from MR images and SUVMR-clinical was computed. Thirdly, phantom was filled with a mixture of CT contrast agent, salt and water, to obtain an attenuation coefficient of the liquid of 0.1cm-1 at 511keV. Vaseline was positioned on the phantom surface to allow better tissue classification of fat and water by the system. Attenuation was calculated from measured MRI as previously and SUVMR was computed. In all cases, 18F-FDG was added in the phantom to obtain an activity of 30 MBq at scan time and a 10 min single-bed-acquisition was reconstructed with TOF including all corrections. SUVCT, SUVMR-clinical and SUVMR were measured with a circular region of interest positioned on 10 transaxial slices. Results: The attenuation maps obtained were visually correct except those derived from MR clinical protocol that led to a signal void in the phantom. The phantom walls were invisible apart from the cylinder bases with manufacturer's protocol. When using the manufacturer's protocol for SUV validation, attenuation coefficient value was correct as expected. With the clinical protocol, an incorrect attenuation coefficient was assigned to the liquid (0.086cm-1 as opposed to the 0.096cm-1 for water). Using our new protocol, the system classified the liquid as soft tissue and consequently assigned a correct attenuation coefficient to it (0.1cm-1). Average SUVCT was 1.037±0.005, SUVMR-clinical was 0.860±0.004 and SUVMR was 1.054±0.007. The corresponding averaged percent errors in SUV estimates were 3.7%, -14.0% and 5.4%, respectively. Conclusion: The standard clinical protocol cannot be used in PET/MR for SUV phantom measurement. CT-derived AC method and new MR-derived AC method provide similar SUV measurements in phantom. By modifying the attenuation of the liquid and increasing the MR contrast, our new protocol for SUV phantom enables the use of MR-based AC. Our method increases the robustness of the SUV measurement as AC method is similar to the one used in clinical routine, which is particularly important in the context of PET/MR.