Improving accuracy of simultaneously reconstructed activity and attenuation maps using deep learning.

Simultaneous reconstruction of activity and attenuation using the maximum likelihood reconstruction of activity and attenuation (MLAA) augmented by time-of-flight (TOF) information is a promising method for positron emission tomography (PET) attenuation correction. However, it still suffers from several problems, including crosstalk artifacts, slow convergence speed, and noisy attenuation maps (μ-maps). In this work, we developed deep convolutional neural networks (CNNs) to overcome these MLAA limitations, and we verified their feasibility using a clinical brain PET data set. Methods: We applied the proposed method to one of the most challenging PET cases for simultaneous image reconstruction (18 F-FP-CIT PET scans with highly specific binding to striatum of the brain). Three different CNN architectures (convolutional autoencoder (CAE), U-net, hybrid of CAE and U-net) were designed and trained to learn x-ray computed tomography (CT) derived μ-map (μ-CT) from the MLAA-generated activity distribution and μ-map (μ-MLAA). PET/CT data of 40 patients with suspected Parkinson's disease were employed for five-fold cross-validation. For the training of CNNs, 800,000 transverse PET slices and CTs augmented from 32 patient data sets were used. The similarity to μ-CT of the CNN-generated μ-maps (μ-CAE, μ-Unet, and μ-Hybrid) and μ-MLAA was compared using Dice similarity coefficients. In addition, we compared the activity concentration of specific (striatum) and non-specific binding regions (cerebellum and occipital cortex) and the binding ratios in the striatum in the PET activity images reconstructed using those μ-maps. Results: The CNNs generated less noisy and more uniform μ-maps than original μ-MLAA. Moreover, the air cavities and bones were better resolved in the proposed CNN outputs. In addition, the proposed deep learning approach was useful for mitigating the crosstalk problem in the MLAA reconstruction. The hybrid network of CAE and U-net yielded the most similar μ-maps to μ-CT (Dice similarity coefficient in the whole head = 0.79 in the bone and 0.72 in air cavities), resulting in only approximately 5% errors in activity and biding ratio quantification. Conclusion: The proposed deep learning approach is promising for accurate attenuation correction of activity distribution in TOF PET systems.