Comparison of planar and 3D lung lobe quantification

Ziel/Aim This study presents a comparison of the to date clinical standard planar quantification to the 3D lung lobe quantification. The hit rate of planar analysis (2D) vs semiautomated 3D lung lobe quantification was investigated using the numerical ventilation/perfusion ratio (VQR). The receiver operating curve (ROC) analysis was used and prognostic validity between planar/3D software and nuclear medicine experts was evaluated.

Methodik/Methods A study group of n=50 patients (sex: male=22, female=28, age: 66.5±13.7 years) underwent static examination and ventilation/perfusion single photon emission computed tomography (V/P-SPECT/CT) to exclude acute PE. Two groups of clinicians made the final diagnosis and interrater reliability was tested using Cohens Kappa. Data was collected with OPTIMA NM/CT 640 SPECT/CT GE Healthcare. 3D lung lobe quantification analysis was performed using: Q. Lung, Xeleris 4.0, GE Healthcare; Hermes Hybrid 3D Lung Lobar Quantification, Hermes Medical Solutions and planar quantification was performed. A linear regression analysis compared both methods. The total perfusion mismatch was calculated for all patients and compared to 7 patients with normal lung structure. ROC analysis for numerical PE detection was performed with both methods.

Ergebnisse/Results Interrater reliability resulted in k=0.74 with a 95% confidence interval (CI) of [0.74; 0.75]. Linear regression of the mean values of the 3D software vs planar showed: r=0.36116 (p<0.0001). An overall lower perfusion mismatch was observed for all three methods. The 3D software tools showed a high correlation of r=0.87173 (p<0.0001) and RIOC values for 2D: [-2.69; 0.413], YOUDEN: [-0.67; 0.315] and AUC: [0.04; 0.65] were presented.

Schlussfolgerungen/Conclusions ROC evaluation revealed that accurate quantification of lung values in clinical practice using only a 2D software tool is limited and insufficient not only for PE detection but also for general diagnostic support. If a 3D approach is used in clinical practice, better clinical predictions are possible.

Publication History

Article published online:
14 April 2022