PET-CT scans using 18F-FDG are increasingly used to detect cancer, but interpretation can be challengingdue to non-specific uptake and complex anatomical structures nearby. To aide this process, we investigatethe potential of automated detection of lesions in 18F-FDG scans using deep learning tools. A 5-layerconvolutional neural network (CNN) with 2×2 kernels, rectified linear unit (ReLU) activations and two denselayers was trained to detect cancerous lesions in 2D axial image segments from PET scans. Pre-contouredscans from a retrospective cohort study of 486 oesophageal cancer patients were split 80:10:10 into training,validation and test sets. These were then used to generate a total of ~14000 25×25×25 voxel image segments,where tumor present segments were centred on the marked lesion, and tumor absent segments were randomlylocated outside the marked lesion. ROC curves generated from the test dataset produced an average AUC of~99%. Ten-fold cross validation on unseen test data was performed which resulted in a sensitivity of99.5±0.4% and a specificity of 99.4±0.3%. A representative model was used to successfully generatevolumetric tumor probability maps for the test dataset.
展开▼