Increasing the confidence of 18F-Florbetaben PET interpretations: Machine learning quantitative approximation.

AIM: To assess the added value of semiquantitative parameters on the visual assessment and to study the patterns of 18F-Florbetaben brain deposition. MATERIALS AND METHODS: Retrospective analysis of multicenter study performed in patients with mild cognitive impairment or dementia of uncertain origin. 18F-Florbetaben PET scans were visually interpreted by two experienced observers, analyzing target regions in order to calculate the interobserver agreement. Semiquantification of all cortical regions with respect to three reference regions was performed to obtain standardized uptake value ratios (SUVRs). The ability of SUVRs to predict the visual evaluation, the possibility of preferential radiotracer deposition in some target regions and interhemisphere differences were analyzed. RESULTS: 135 patients were evaluated. In the visual assessment, 72 were classified as positive. Interobserver agreement was excellent. All SUVRs were significantly higher in positive PET scans than in negative ones. Prefrontal area and posterior cingulate were the cortical regions with the best correlations with the visual evaluation, followed by the composite region. Using ROC analysis, the SUVRs obtained in same target locations showed the best diagnostic performance. CONCLUSIONS: The derived information from target regions seems to help the visual classification, based on a preferential amyloid ß deposit, allowing machine learning. The amyloid ß deposit, although diffuse in all cortical regions, seems not to be uniform and symmetric.

Intratumoral heterogeneity in 18F-FDG PET/CT by textural analysis in breast cancer as a predictive and prognostic subrogate.

AIM: To assess the predictive and prognostic value of textural parameters in locally advanced breast cancer (LABC) obtained by 18F-FDG PET/CT. METHODS: Prospective study including 68 patients with LABC, neoadjuvant chemotherapy (NC) indication and a baseline 18F-FDG PET/CT. Breast specimens were grouped into molecular phenotypes and classified as responders or non-responders after completion of NC. Patients underwent standard follow-up to obtain the disease-free survival (DFS) and overall survival (OS). After breast tumor segmentation, three-dimensional (3D) textural measures were computed based on run-length matrices (RLM) and co-occurrence matrices (CM). Relations between textural features with risk categories attending to molecular phenotypes were explored. Kaplan-Meier analysis and univariate and multivariate Cox proportional hazard analysis were used to study the potential of textural variables, molecular phenotypes and histologic response to predict DFS and OS. Receiver operating characteristic (ROC) analysis was used to obtain the best cut-off value, the area under the curve (AUC) and sensitivity and specificity considering OS and DFS. RESULTS: Eighteen patients were classified as responders. Mean ± SD of DFS and OS was 70.87 ± 21.85 and 76.77 ± 18.80 months, respectively. Long run emphasis (LRE) and long run high gray-level emphasis (LRHGE) showed a relation with risk categories. Low gray-level run emphasis (LGRE), LRHGE and run-length non-uniformity (RLNU) showed association with the NC response. Textural variables were significantly associated with OS and DFS in univariate analysis. Regarding the multivariate Cox regression analysis, PET stage with short run high gray-level emphasis (SRHGE) was significantly associated with OS, and PET stage and high gray-level run emphasis (HGRE) with DFS. CONCLUSION: Textural variables obtained with 18F-FDG PET/CT were predictors of neoadjuvant chemotherapy response and prognosis, being as relevant as PET stage at diagnosis for OS and DFS prediction.

Do clinical and laboratory variables have any impact on the diagnostic performance of 18F-FDG PET/CT in patients with fever of unknown origin?

OBJECTIVE: To assess the influence of clinical features and laboratory test results on the determination of fever of unknown origin (FUO) by means of 18F-FDG PET/CT. METHODS: Retrospective and longitudinal analysis, including all the PET/CT studies requested for FUO. Reference standard was established by serology, cultures or biopsy with other laboratory tests or clinical follow-up when necessary. Clinical variables, inflammation markers, protein analysis, serology and culture results close to the PET scan were obtained. The final diagnosis was classified into three groups attending to the etiology; group 1: infection or neoplasm, group 2: vasculitis, autoimmune disease or non-infectious inflammatory disease and group 3: auto-limited fever or persistent fever without diagnosis. PET/CT scans were classified as positive or negative and helpful or not in the diagnosis of the fever origin. The effect of clinical features and laboratory variables on the PET/CT results was analyzed. RESULTS: Sixty-seven patients were evaluated. The final diagnosis was: Group 1 (25), Group 2 (20) and Group 3 (22). 89.6% of patients had a positive inflammation marker, 28.4% proteinogram alterations and 20.9% positive cultures. PET/CT was positive in 52/67 patients. PET/CT helped in the establishment of the fever origin in 35 cases and was especially helpful in groups 1 and 2. Sensitivity, specificity and accuracy of PET/CT were: 84, 31 and 61%. PET results shown significant relations with the final diagnosis (p = 0.035) and culture results (p = 0.037). No significant relations were observed with the rest of clinical or laboratory variables. CONCLUSIONS: 18F-FDG PET/CT had a high sensitivity but a low specificity in the diagnosis of the fever origin, probably due to the high rate of diffuse and auto-limited aetiologies. Patients who are most likely to benefit from the PET/CT study would be those with several positive inflammation markers, reflecting a higher pre-test probability of active disease.

Textural features and SUV-based variables assessed by dual time point 18F-FDG PET/CT in locally advanced breast cancer.

AIM: To study the influence of dual time point 18F-FDG PET/CT in textural features and SUV-based variables and their relation among them. METHODS: Fifty-six patients with locally advanced breast cancer (LABC) were prospectively included. All of them underwent a standard 18F-FDG PET/CT (PET-1) and a delayed acquisition (PET-2). After segmentation, SUV variables (SUVmax, SUVmean, and SUVpeak), metabolic tumor volume (MTV), and total lesion glycolysis (TLG) were obtained. Eighteen three-dimensional (3D) textural measures were computed including: run-length matrices (RLM) features, co-occurrence matrices (CM) features, and energies. Differences between all PET-derived variables obtained in PET-1 and PET-2 were studied. RESULTS: Significant differences were found between the SUV-based parameters and MTV obtained in the dual time point PET/CT, with higher values of SUV-based variables and lower MTV in the PET-2 with respect to the PET-1. In relation with the textural parameters obtained in dual time point acquisition, significant differences were found for the short run emphasis, low gray-level run emphasis, short run high gray-level emphasis, run percentage, long run emphasis, gray-level non-uniformity, homogeneity, and dissimilarity. Textural variables showed relations with MTV and TLG. CONCLUSION: Significant differences of textural features were found in dual time point 18F-FDG PET/CT. Thus, a dynamic behavior of metabolic characteristics should be expected, with higher heterogeneity in delayed PET acquisition compared with the standard PET. A greater heterogeneity was found in bigger tumors.