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

• DOI: 10.1007/s00259-021-05473-2
• Publication Date: 2021-07-07T09:19:55Z
• Authors: Nicolò Capobianco; Ludovic Sibille; Maythinee Chantad...; Andrei Gafita; Thomas Langbein; Guenther Platsch; Esteban Lucas Solari; Vijay Shah; Bruce Spottiswoode; Matthias Eiber; Wolfgang A. Weber; Nassir Navab; Stephan G. Nekolla
• Keywords: Transfer of learning

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