Discrimination between Benign and Malignant Lung Lesions Using a Tri-Compartmental Restriction Spectrum Imaging Model

Motivation: An accurate non-invasive detection method is required to identify benign and malignant lung lesions. Goal(s): To assess the predictive value of the tri-compartmental RSI model in distinguishing between benign and malignant lung lesions. Approach: In a prospective study, 137 suspected lung tumor patients who underwent PET/MRI scans were included. Results: Elevated RSI-C1, SUVmax, and lower ADC, RSI-C3 were observed in the malignant group compared with benign group (all P < 0.001). The AUCs, ranked from highest to lowest, were 0.819 (SUVmax) > 0.811 (RSI-C3) > 0.770 (RSI-C1) > 0.745 (ADC). The combined model of RSI-C1, RSI-C3, and SUVmax showed an AUC of 0.900. Impact: Our study provides preliminary evidence of the potential of RSI-MRI technology to enhance non-invasive imaging of lung lesions. PET/MRI combined with metabolic parameter is helpful to optimize the clinical diagnosis of lung lesions.