Ming Chen1,2, Hailong Li1, Jinghua Wang3, Nehal A. Parikh4, and Lili He1
1Department of Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, 2Department of Electronic Engineering and Computing Science, University of Cincinnati, Cincinnati, OH, United States, 3Deep MRI Imaging Inc., Lewes, DE, United States, 4The Perinatal Institute, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States
1Department of Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, 2Department of Electronic Engineering and Computing Science, University of Cincinnati, Cincinnati, OH, United States, 3Deep MRI Imaging Inc., Lewes, DE, United States, 4The Perinatal Institute, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States
A novel multi-filter convolutional neural
network achieved an area under the receiver operating characteristic curve of
0.78 on the identification of high-risk infants for cognitive deficits at 2
years corrected age using brain structural connectome in very preterm infants.
Figure 1. An overview of proposed multi-filter
convolutional neural network for early prediction of cognitive deficits in very
preterm infants. The number of vector-shape filters is listed next to each
convolutional layer.
Table 2. Performance
comparison of our proposed multi-filter convolutional neural network (CNN) model vs. other peer CNN models for
the identification of high-risk infants for cognitive deficits in very preterm
infants at 2-years corrected age.