Whole-body uptake classification and prostate cancer staging in 68Ga-PSMA-11 PET/CT using dual-tracer learning.

PURPOSE: In PSMA-ligand PET/CT imaging, standardized evaluation frameworks and image-derived parameters are increasingly used to support prostate cancer staging. Clinical applicability remains challenging wherever manual measurements of numerous suspected lesions are required. Deep learning methods are promising for automated image analysis, typically requiring extensive expert-annotated image datasets to reach sufficient accuracy. We developed a deep learning method to support image-based staging, investigating the use of training information from two radiotracers. METHODS: In 173 subjects imaged with 68Ga-PSMA-11 PET/CT, divided into development (121) and test (52) sets, we trained and evaluated a convolutional neural network to both classify sites of elevated tracer uptake as nonsuspicious or suspicious for cancer and assign them an anatomical location. We evaluated training strategies to leverage information from a larger dataset of 18F-FDG PET/CT images and expert annotations, including transfer learning and combined training encoding the tracer type as input to the network. We assessed the agreement between the N and M stage assigned based on the network annotations and expert annotations, according to the PROMISE miTNM framework. RESULTS: In the development set, including 18F-FDG training data improved classification performance in four-fold cross validation. In the test set, compared to expert assessment, training with 18F-FDG data and the development set yielded 80.4% average precision [confidence interval (CI): 71.1-87.8] for identification of suspicious uptake sites, 77% (CI: 70.0-83.4) accuracy for anatomical location classification of suspicious findings, 81% agreement for identification of regional lymph node involvement, and 77% agreement for identification of metastatic stage. CONCLUSION: The evaluated algorithm showed good agreement with expert assessment for identification and anatomical location classification of suspicious uptake sites in whole-body 68Ga-PSMA-11 PET/CT. With restricted PSMA-ligand data available, the use of training examples from a different radiotracer improved performance. The investigated methods are promising for enabling efficient assessment of cancer stage and tumor burden.

Evaluation of MLACF based calculated attenuation brain PET imaging for FDG patient studies.

Calculating attenuation correction for brain PET imaging rather than using CT presents opportunities for low radiation dose applications such as pediatric imaging and serial scans to monitor disease progression. Our goal is to evaluate the iterative time-of-flight based maximum-likelihood activity and attenuation correction factors estimation (MLACF) method for clinical FDG brain PET imaging. FDG PET/CT brain studies were performed in 57 patients using the Biograph mCT (Siemens) four-ring scanner. The time-of-flight PET sinograms were acquired using the standard clinical protocol consisting of a CT scan followed by 10 min of single-bed PET acquisition. Images were reconstructed using CT-based attenuation correction (CTAC) and used as a gold standard for comparison. Two methods were compared with respect to CTAC: a calculated brain attenuation correction (CBAC) and MLACF based PET reconstruction. Plane-by-plane scaling was performed for MLACF images in order to fix the variable axial scaling observed. The noise structure of the MLACF images was different compared to those obtained using CTAC and the reconstruction required a higher number of iterations to obtain comparable image quality. To analyze the pooled data, each dataset was registered to a standard template and standard regions of interest were extracted. An SUVr analysis of the brain regions of interest showed that CBAC and MLACF were each well correlated with CTAC SUVrs. A plane-by-plane error analysis indicated that there were local differences for both CBAC and MLACF images with respect to CTAC. Mean relative error in the standard regions of interest was less than 5% for both methods and the mean absolute relative errors for both methods were similar (3.4% ± 3.1% for CBAC and 3.5% ± 3.1% for MLACF). However, the MLACF method recovered activity adjoining the frontal sinus regions more accurately than CBAC method. The use of plane-by-plane scaling of MLACF images was found to be a crucial step in order to obtain improved activity estimates. Presence of local errors in both MLACF and CBAC based reconstructions would require the use of a normal database for clinical assessment. However, further work is required in order to assess the clinical advantage of MLACF over CBAC based method.