Artificial Intelligence and Translational Theranostics

Deep learning for radionuclide therapy optimization

W-NET for the characterizing tumor load for the optimization of radionuclide therapy [Xu et al. Contrast Media Mol Imaging 2018]

Deep learning for computer-aided diagnosis

Deep projection neural network on 18F-FDG imaging for differential diagnosis of parkinsonian syndromes [Kumar et al. MICCAI-DLMIA 2018].
Densely-connected convolutional neural network for differential diagnosis of pancreatic cystic lesions [[Li et al. arXiv 1806.01023v1].



Machine learning for instrumentation optimization

Multi-resolution regional learning for individual refinement of attenuation correction maps for hybrid PET/MR [Shi et al. CMIG 2017].
Support vector machine for super-resolution of pixelated semiconductor detector (Timepix) by identifying the principal incidence of position [Wang et al. Phys Med Biol 2015].

Development of ex vivo molecular imaging system

A continuously infused microfluidic radioassay system for real time monitoring of cellular uptake and characterization of cellular pharmacokinetics [Liu et al. J Nucl Med 2016].

Quantitative analysis of molecular imaging

Adjustment of age-associated metabolism for the analysis of PET imaging in dementia [Zhang et al. Neuroimage 2017].
Pharmacokinetic modeling on preclinical and clinical PET imaging [Shi et al J Nucl Med 2010, Shi et al. Mol Imaging Bol 2017].
Parametric image reconstruction for the improvement of PET quantification as well rapid multi-tracer imaging [Cheng et al. MICCAI 2013, Cheng et al Phys Med Biol 2014, Cheng et al. IEEE Trans Med Imaging 2015].

Researcher Portrait (Center for Artificial Intelligence in Medicine)