Enhancing PSMA PET/CT imaging of prostate cancer: investigating the impact of multiple time point evaluation, diuretic administration, cribriform pattern, and intraductal carcinoma.

OBJECTIVE: Our objective was to correlate staging PSMA PET imaging parameters to final histopathology. Second objective was to assess the performance of standard versus delayed PSMA PET to detect primary prostate tumor. METHODS: Thirty-one patients (mean age, 61.4 ± 8.2) who underwent radical prostatectomy and preoperative staging PSMA PET scans were included in the study. After defining dominant lesion in pathology, correlations with PET images were performed. Additionally, two physicians blind to clinical and pathological information retrospectively reviewed staging Ga-68 PSMA PET scans with standard and delayed imaging. RESULTS: Dominant lesion SUV's increased with time 8.2(± 4.5), 10(± 7.1), and 10.2(± 7.8) at 1, 2, and 3 h (P = .03 T1-T3). WHO Grade group 3 had highest SUV (group 3 11.9 ± 5.6 vs. group 2 7.9 ± 1.5, p = .02). Addition of cribriform pattern on intraductal component was associated with higher SUV's (11 ± 2.9 vs. 6.5 ± 2.1, p = .01) and higher Gleason four ratios (64 ± 9% vs. 37 ± 17%, p = .01). Intraductal carcinoma was associated with larger tumors (6.3 ± 2.3 cm3 vs. 2.6 ± 1.7 cm3, p < .001). Physician sensitivities ranged from 61 to 81%. Excluding Gleason 3 + 3 lesions and small lesions (< 1 cm3), sensitivities increased to 80-100%. Differences of sensitivity between different time points were not significant. Combined evaluation of all time points did not increase sensitivity. CONCLUSIONS: Cribriform pattern correlates with higher Gleason 4 ratios and SUVs in PSMA PET. Intraductal carcinoma is associated with larger tumors but not higher Gleason 4 ratios and SUVs. Multiple late imaging times did not enhance tumor detection and may pose tolerability issues for some patients.

Physiological fluorodeoxyglucose uptake of spinal cord in adults.

OBJECTIVE: Physiological fluorodeoxyglucose (FDG) uptake of spinal cord needs to be correctly recognized during evaluation of whole-body PET scans, especially for oncological cases. Our aim was to analyze physiological cord FDG uptake and its relation to gender, age, body weight, environmental temperature and time to imaging. MATERIALS AND METHODS: PET scans of 254 patients in a single year, one patient for every working day were retrospectively selected. Temperature data were obtained from meteorology recordings. Maximum standard uptake value (SUVmax) of spinal cord at cervical and lower thoracic levels were noted. Spinal canal at L5 level, cerebellum and liver were used for normalization. Correlations with age, body weight, time to imaging and environmental temperature were analyzed. RESULTS: Cervical SUV was higher than thoracic SUV (2.5-2.3). Cervical and lower thoracic SUV's were strongly correlated, highest when corrected with L5 level vertebral canal and liver (corr coeff 0.84 and 0.75) and lowest with cerebellum (corr coeff 0.4). Cervical spinal cord FDG uptake was higher for females than males (2.6 to 2.4). Temperature and age did not change spinal cord uptake. There were weak positive correlations with body weight (corr coeff 0.16 and 0.28, cervical and thoracic). There was weak negative correlation of cervical uptake with time to imaging (corr coeff -0.17). CONCLUSION: Spinal cord FDG uptake at cervical and lower thoracic levels are strongly correlated. Females have slightly higher cervical SUV. Age and temperature does not change spinal cord FDG uptake in adults. Cord SUV's slightly increased with body weight.

Evaluating the Role of Routine Prone Acquisition on Visual Evaluation of SPECT Images.

UNLABELLED: Attenuation artifacts reduce our ability to evaluate perfusion of affected myocardial segments. The aim of this study was to evaluate the impact of routine prone-position image evaluation within a stepwise visual interpretation of myocardial perfusion studies. METHODS: We have included 279 consecutive patients who were referred for evaluation of myocardial ischemia. All patients underwent routine electrocardiogram-gated supine SPECT imaging and non-electrocardiogram-gated prone-position SPECT imaging. Three nuclear medicine physicians interpreted the images in the following order: polar maps, supine images, raw images, motion-frozen gated images, and prone images, using a scale of 0-4. Segments with perfusion abnormalities were noted. RESULTS: All physicians reported lower proportions of equivocal evaluations after evaluating prone images (18.3% vs. 4.7%, 19% vs. 11.1%, and 12.2% vs. 6.1%, P < 0.0001, P = 0.0077, and P = 0.0125, respectively). At the prone stage, normalcy rates were 89%, 87%, and 91%. Two physicians had significantly increased normalcy rates at the prone stage (72%-89%, P = 0.039, and 66%-87%, P = 0.006). At the prone stage, a decision reversal to normal or probably normal was observed in 40% (29/72), 33% (17/51), and 43% (21/48). In men, apical, mid, and basal inferior walls and in women apical and mid parts of anterior walls were more likely to be attributed to attenuation. The 2 steps that increased normalcy rates for interpreters were the review of raw images and of prone images. CONCLUSION: Routine prone imaging increases interpretive certainty and interobserver agreement and changes the final evaluation in a substantial number of patients and significantly decreases the number of equivocal evaluations.

Comparison of diagnostic performances of three different software packages in detecting coronary artery disease.

PURPOSE: Quantification of myocardial perfusion scintigraphy is frequently performed to assist physicians in detecting coronary artery disease (CAD). Software packages provide automated quantification of perfusion data. We aimed to compare the three commonly used software packages, Emory Cardiac Toolbox (ECT v2 and ECT v3), 4D-MSPECT (4DM v2 and 4DM v4) and Quantitative Perfusion SPECT (QPS v3 and QPS v4). METHODS: We selected 283 patients who had a myocardial perfusion scintigraphy with (201)Tl followed by coronary angiography within 3 months. Summed stress score (SSS), summed difference score (SDS), total stress defect extent (TDE) and regional stress defect extent values were obtained from programs. A ≥70% stenosis in coronary arteries and their major branches was considered positive for CAD. A subgroup of patients was used to form an institutional normal database for QPS and 4DM. Receiver-operating characteristic (ROC) analysis to detect CAD was performed. RESULTS: Mean SSS ± SD (vendor) for ECT v3, QPS v4 and 4DM v4 were 9.2 ± 7.1, 10.1 ± 6.8 and 5.5 ± 6.1, respectively. Area under the curve (AUC) values of SSS ROC analysis were 0.738 ± 0.031 for QPS v3, 0.755 ± 0.030 for QPS v4, 0.758 ± 0.030 for ECT v2, 0.778 ± 0.029 for ECT v3 and 0.771 ± 0.030 for 4DM v4. The AUC values for TDE were 0.755 ± 0.030 for QPS v4, 0.769 ± 0.030 for ECT v3 and 0.775 ± 0.029 for 4DM v4. The differences were not significant for both SSS and TDE. Differences of AUC between regional stress defect extent values of programs and AUC of SSS between institutional and vendor normal databases were not significant. CONCLUSION: The diagnostic performances of programs to detect CAD are similar. However, there are differences in the magnitudes of the quantitative values produced by the programs.

An open-source framework of neural networks for diagnosis of coronary artery disease from myocardial perfusion SPECT.

BACKGROUND: The purpose of this study is to develop and analyze an open-source artificial intelligence program built on artificial neural networks that can participate in and support the decision making of nuclear medicine physicians in detecting coronary artery disease from myocardial perfusion SPECT (MPS). METHODS AND RESULTS: Two hundred and forty-three patients, who had MPS and coronary angiography within three months, were selected to train neural networks. Six nuclear medicine residents, one experienced nuclear medicine physician, and neural networks evaluated images of 65 patients for presence of coronary artery stenosis. Area under the curve (AUC) of receiver operating characteristics analysis for networks and expert was .74 and .84, respectively. The AUC of the other physicians ranged from .67 to .80. There were no significant differences between expert, neural networks, and standard quantitative values, summed stress score and total stress defect extent. CONCLUSIONS: The open-source neural networks developed in this study may provide a framework for further testing, development, and integration of artificial intelligence into nuclear cardiology environment.