Asae Nozawa MD, Presenter: Nothing to Disclose

Santosh Kesari MD, Abstract Co-Author: Nothing to Disclose

Ali Hosseini Rivandi MD, Abstract Co-Author: Nothing to Disclose

Sebastian Obrzut MD, Abstract Co-Author: Nothing to Disclose

Carl Kalon Hoh MD, Abstract Co-Author: Nothing to Disclose

This study evaluates the utility of standarized uptake value (SUV) and glucose metabolic rate (GMR) parametric images of fluorine-18 fluorodeoxyglucose positron emission tomography (FDG PET) to differentiate tumor and benign CNS lesions.

In 18 patients with dynamic FDG PET scan, there were 33 CNS lesions: tumor (n=26), benign or post treatment changes (n=7). Final true diagnosis was based on brain biopsy (16) and clinical outcome (17). Parametric images of GMR were generated using the Patlak method applied to the 3-36 minute scan interval, a single plasma glucose level and assumed lumped constant of 1. SUV images were generated by summing the last 30 minutes of the dynamic scan. Both the GMR and SUV images were coregistered to a recent (< 1 month) brain contrast MRI using manual and automated software. Ratios of activity were calculated on regions of interested (ROIs) from the lesion to the ipsilateral cerebellar cortex (Rc) and contralateral white matter (centrum semiovali) (Rw).

The mean SUV value for tumor lesions was 7.20±4.24 versus 2.67±0.39 for benign lesions (p<0.01). The mean GMR for tumor lesions was 25.9±15.4 versus and 9.55±1.85 for benign lesions (p<0.01). The Rc was 2.02 ± 0.78 for tumor and 0.86 ± 0.37 for benign lesions using the SUV image (p<0.01) and 2.02 ± 0.78 for tumor and 0.86 ± 0.37 for benign lesions using the GMR image (p<0.01). The Rw was 2.88 ± 1.35 for tumor and 1.46 ± 0.39 for benign lesions using the SUV image (p < 0.01) and 3.54 ± 1.77 for tumor and 1.55 ± 0.51 for benign lesions using the GMR images (p<0.01). Using a threshold of >3 for SUV and 13 for GMR both provide an accuracy of 90%. Using a threshold of 1 for Rc, the accuracy was 80% using the SUV and 93% using the GMR. Using a threshold of 2 for Rw, the accuracy was 70% using the SUV and 79% for GMR.

These results suggest that the GMR and SUV value and ratios to the ipsilateral cerebellum or contralateral white matter may assist in the interpretation of FDG PET brain images in patient with CNS lesions. The simplicity of the SUV method avoids the complexity of calculating the GMR.

In the evaluation of patients with CNS lesions with FDG PET, quantiative and semiquantitative indices of the lesion FDG uptake may assist in image interpretation.

Nozawa, A, Kesari, S, Hosseini Rivandi, A, Obrzut, S, Hoh, C, Parametric Metabolic FDG PET Brain Scans for the Detection of Brain Tumors.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11013660.html