Variaciones en los parámetros cuantitativos y en la puntuación de la escala de Deauville aplicando diferentes algoritmos de reconstrucción en las imágenes PET/TC con [18F]FDG de pacientes con linfoma / The variation of quantitative parameters and Deauville scores with different reconstruction algorithms in FDG PET/CT imaging of lymphoma patients

Objetivos La escala de Deauville (DS) en la tomografía de emisión de positrones (PET) con [18F]fludeoxiglucosa ([18F]FDG) es un método semicuantitativo único para la evaluación del linfoma. Sin embargo, el tipo de algoritmo de reconstrucción empleado para el cálculo de los valores de captación estándar (max, medio y pico) podría afectar a la DS. Comparamos el algoritmo de reconstrucción de probabilidad penalizada bayesiano (BPL) con el de maximización de expectativas de subconjuntos ordenados (OSEM) respecto a los parámetros cuantitativos y en la DS en el linfoma. Investigamos el efecto del tamaño del ganglio linfático sobre la variación cuantitativa. Métodos Se reconstruyeron por separado los resultados de la PET sin procesar de 255 pacientes con linfoma utilizando la aplicación Q.Clear (General Electric Healthcare, Milwaukee, WI, EE. UU.), un algoritmo BPL, y la aplicación SharpIR (General Electric Healthcare, Milwaukee, WI, EE. UU.), un algoritmo OSEM. En ambas imágenes, para cada paciente, se valoró hígado, el pool sanguíneo mediastínico y los valores de captación estándar (SUV) (SUVmáx, SUVmedio y SUVpico) de un total de 487 lesiones seleccionadas. Se compararon DSmáx, DSmedio y DSpico. Resultados En nuestro estudio hubo un aumento significativo de la DS con el BPL (p<0,001) que pasó a una puntuación de 4 a 5 en 30 pacientes inicialmente catalogados como 1-2-3 mediante el algoritmo OSEM. Se observó que los valores cuantitativos de los ganglios linfáticos aumentaban de forma estadísticamente significativa con el BPL (p<0,001), mientras que la disminución de los valores de hígado fue notable respecto a las regiones de referencia (p<0,001). Además, la diferencia en los ganglios linfáticos se asoció de forma independiente con el tamaño de la lesión y fue considerablemente más pronunciada en las lesiones de pequeño tamaño (p<0,001) (AU)

Introduction and Objectives 18F-FDG PET with the Deauville score (DS) is a unique semiquantitative method for lymphoma. However, type of standard uptake values (max, mean, and peak) reconstruction algorithms could affect DS. We compared the Bayesian Penalized Likelihood reconstruction algorithm (BPL) with Ordered Subsets Expectation Maximization (OSEM) on quantitative parameters and DS in lymphoma. We investigated the effect of the size of the lymph node on quantitative variation. Patients and Methods Raw PET data of 255 lymphoma patients were reconstructed separately with Q.Clear (GE Healthcare), a BPL, and SharpIR (GE Healthcare), an OSEM algorithm. In both images, each patient's liver, mediastinal blood pool, and SUVs (SUVmax, SUVmean, and SUVpeak) of a total of 487 lesions selected from the patients were performed. DSmax, DSmean, and DSpeak were compared. Results In our study, DS increased significantly with BPL (p<0.001), and the DS increased to 4-5 in 30 patients evaluated as 1-2-3 with OSEM. It was found that the quantitative values of the lymph nodes increased statistically with BPL (p<0.001), and the liver from the reference regions were significantly decreased (p<0.001). In addition, difference in lymph node was independently associated with size of lesion and was significantly more pronounced in small lesions (p<0.001). The effects of BPL algorithm were more pronounced in SUVmax than in SUVmean and SUVpeak. DS-mean and DS-peak scores were less changed by BPL than DS-max. Conclusion Different reconstruction algorithms in FDG PET/CT affect the quantitative evaluation. That variation may affect the change in DS in lymphoma patients, thus affecting patient management (AU)

The variation of quantitative parameters and Deauville scores with different reconstruction algorithms in FDG PET/CT imaging of lymphoma patients.

INTRODUCTION AND OBJECTIVES: 18F-FDG PET with the Deauville score (DS) is a unique semiquantitative method for lymphoma. However, type of standard uptake values (max, mean, and peak) reconstruction algorithms could affect DS. We compared the Bayesian Penalized Likelihood reconstruction algorithm (BPL) with Ordered Subsets Expectation Maximization (OSEM) on quantitative parameters and DS in lymphoma. We investigated the effect of the size of the lymph node on quantitative variation. PATIENTS AND METHODS: Raw PET data of 255 lymphoma patients were reconstructed separately with Q.Clear (GE Healthcare), a BPL, and SharpIR (GE Healthcare), an OSEM algorithm. In both images, each patient's liver, mediastinal blood pool, and SUVs (SUVmax, SUVmean, and SUVpeak) of a total of 487 lesions selected from the patients were performed. DSmax, DSmean, and DSpeak were compared. RESULTS: In our study, DS increased significantly with BPL (p < 0.001), and the DS increased to 4-5 in thirty patients evaluated as 1-2-3 with OSEM. It was found that the quantitative values of the lymph nodes increased statistically with BPL (p < 0.001), and the liver from the reference regions were significantly decreased (p < 0.001). In addition, difference in lymph node was independently associated with size of lesion and was significantly more pronounced in small lesions (p < 0.001). The effects of BPL algorithm were more pronounced in SUVmax than in SUVmean and SUVpeak. DS-mean and DS-peak scores were less changed by BPL than DS-max. CONCLUSION: Different reconstruction algorithms in FDG PET/CT affect the quantitative evaluation. That variation may affect the change in DS in lymphoma patients, thus affecting patient management.