Development of Radiotherapeutic Pharmaceuticals and Application of Deep Learning to Drug Development and Disease Classification

Anster Charles, University of Missouri
Webinar
CANDLE image

Predicting cancer type and drug response using histopathology images from the National Cancer Institute’s Patient-Derived Models Repository. Image: Argonne National Laboratory

Description: Radiotherapeutic pharmaceuticals are drugs that contain radioisotopes which emit ionizing radiation to kill malignant cells. Critical to the development of these drugs are the production and separation of medically relevant cyclotron and reactor produced radionuclides. With the help of various computational models, we have developed an efficient method for the production and purification of rhenium-186 and successfully reformulated a novel yttrium-90 based brachytherapy tumor agent which achieved tumor reduction in mouse models.

My research also explores the application of deep learning to drug development by using graphical representations of small molecules and graph convolutional networks to predict drug properties. For classification of melanoma image data, transfer learning was employed via the pre-trained DenseNet161 PyTorch model which was optimized for our dataset.