Comparing various AI approaches to traditional quantitative assessment of the myocardial perfusion in [82Rb] PET for MACE prediction.

Assessing the individual risk of Major Adverse Cardiac Events (MACE) is of major importance as cardiovascular diseases remain the leading cause of death worldwide. Quantitative Myocardial Perfusion Imaging (MPI) parameters such as stress Myocardial Blood Flow (sMBF) or Myocardial Flow Reserve (MFR) constitutes the gold standard for prognosis assessment. We propose a systematic investigation of the value of Artificial Intelligence (AI) to leverage [ 82 Rb] Silicon PhotoMultiplier (SiPM) PET MPI for MACE prediction. We establish a general pipeline for AI model validation to assess and compare the performance of global (i.e. average of the entire MPI signal), regional (17 segments), radiomics and Convolutional Neural Network (CNN) models leveraging various MPI signals on a dataset of 234 patients. Results showed that all regional AI models significantly outperformed the global model ( p < 0.001 ), where the best AUC of 73.9% (CI 72.5-75.3) was obtained with a CNN model. A regional AI model based on MBF averages from 17 segments fed to a Logistic Regression (LR) constituted an excellent trade-off between model simplicity and performance, achieving an AUC of 73.4% (CI 72.3-74.7). A radiomics model based on intensity features revealed that the global average was the least important feature when compared to other aggregations of the MPI signal over the myocardium. We conclude that AI models can allow better personalized prognosis assessment for MACE.

Lung CT stabilization with high-frequency non-invasive ventilation (HF-NIV) and breath-hold (BH) in lung nodule assessment by PET/CT.

PURPOSE: To evaluate the effect of lung stabilization using high-frequency non-invasive ventilation (HF-NIV) and breath-hold (BH) techniques on lung nodule detection and texture assessment in PET/CT compared to a free-breathing (FB) standard lung CT acquisition in PET/CT. MATERIALS AND METHODS: Six patients aged 65 ± 7 years, addressed for initial assessment of at least one suspicious lung nodule with 18F-FDG PET/CT, underwent three consecutive lung PET/CT acquisitions with FB, HF-NIV and BH. Lung nodules were assessed on all three CT acquisitions of the PET/CT and characterized for any size, volume and solid/sub-solid nature. RESULTS: BH detected a significantly higher number of nodules (n = 422) compared to HF-NIV (n = 368) and FB (n = 191) (p < 0.001). The mean nodule size (mm) was 2.4 ± 2.1, 2.6 ± 1.9 and 3.2 ± 2.4 in BH, HF-NIV and FB, respectively, for long axis and 1.5 ± 1.3, 1.6 ± 1.2 and 2.1 ± 1.7 in BH, HF-NIV and FB, respectively, for short axis. Long- and short-axis diameters were significantly different between BH and FB (p < 0.001) and between HF-NIV and FB (p < 0.001 and p = 0.008), but not between BH and HF-NIV. A trend for higher volume was shown in FB compared to BH (p = 0.055) and HF-NIV (p = 0.068) without significant difference between BH and HF-NIV (p = 1). We found a significant difference in detectability of sub-solid nodules between the three acquisitions, with BH showing a higher number of sub-solid nodules (n = 128) compared to HF-NIV (n = 72) and FB (n = 44) (p = 0.002). CONCLUSION: We observed a higher detection rate of pulmonary nodules on CT under BH or HF-NIV conditions applied to PET/CT than with FB. BH and HF-NIV demonstrated comparable texture assessment and performed better than FB in assessing size and volume. BH showed a better performance for detecting sub-solid nodules compared to HF-NIV and FB. The addition of BH or HF-NIV to PET/CT can help improve the detection and texture characterization of lung nodules by CT, therefore improving the accuracy of oncological lung disease assessment. The ease of use of BH and its added value should prompt its use in routine practice.

A novel approach for fibrous dysplasia assessment using combined planar and quantitative SPECT/CT analysis of Tc-99m-diphosphonate bone scan in correlation with biological bone turnover markers of disease activity.

Purpose: To investigate the emerging role of Tc-99m-labeled diphosphonate (Tc-99m-DPD) uptake quantification by SPECT/CT in fibrous dysplasia (FD) bone lesions and its correlation with biological bone turnover markers (BTMs) of disease activity. Materials and methods: Seven patients (49 ± 16 years) with a confirmed diagnosis of FD were included in this retrospective study. Bone scans with Tc-99m-DPD and quantitative SPECT/CT (xSPECT/CT) were performed. SUVmax (maximum standard unit value) and SUVmean (mean standard unit value) were measured in all FD bone lesions. The skeletal burden score (SBS) was assessed on planar scintigraphy and multiplied by mean SUV max and SUVmean to generate two new parameters, SBS_SUVmax and SBS_SUVmean, respectively. Planar and xSPECT/CT quantitative measures were correlated with biological BTMs of disease activity, including fibroblast growth factor 23 (FGF-23), alkaline phosphatase (ALP), procollagen 1 intact N-terminal propeptide (P1NP) and C-terminal telopeptide (CTX), as well as scoliosis angle measured on radiographs. Statistical significance was evaluated with Spearman's correlations. Results: A total of 76 FD bone lesions were analyzed, showing an average SUVmax and SUVmean (g/mL) of 13 ± 7.3 and 8 ± 4.5, respectively. SBS, SBS_SUVmax and SBS_SUVmean values were 30.8 ± 25.6, 358 ± 267 and 220.1 ± 164.5, respectively. Mean measured values of FGF-23 (pg/mL), ALP (U/L), P1NP (µg/L) and CTX (pg/mL) were 98.4 (22-175), 283.5 (46-735), 283.1 (31-1,161) and 494 (360-609), respectively. Mean scoliosis angle was 15.7 (7-22) degrees. We found a very strong positive correlation between planar-derived SBS and CTX (r = 0.96, p = 0.010), but no significant correlation between SUVmax or SUVmean and biological BTMs. SBS_SUVmax showed a strong to very strong positive correlation with CTX (ρ = 0.99, p = 0.002), FGF-23 (ρ = 0.91, p = 0.010), ALP (ρ = 0.82, p = 0.020), and P1NP (ρ = 0.78, p = 0.039), respectively. Conclusion: This study showed that biological BTMs are significantly correlated with diphosphonate uptake on bone scan, quantified by a new parameter combining information from both planar and quantitative SPECT/CT. Further analysis of bone scan quantitative SPECT/CT data in a larger patient population might help better characterize the skeletal disease burden in FD, and guide treatment and follow-up.