Rachel E Roca1, Joshua D Herman1, Alexandra G O'Neill1, Sajan G Lingala2, and Angel R Pineda1
1Mathematics Department, Manhattan College, Riverdale, NY, United States, 2Roy J. Carver Department of Biomedical Engineering, University of Iowa, Iowa City, IA, United States
1Mathematics Department, Manhattan College, Riverdale, NY, United States, 2Roy J. Carver Department of Biomedical Engineering, University of Iowa, Iowa City, IA, United States
Choosing
the parameters of a neural network for image reconstruction based on the
detection of signals may lead to artifacts which are not acceptable. Task-based optimization may not align with
artifact minimization.
Figure 4. Sample
2AFC trial for network with 64 channels and 0.3 dropout including artifacts
which do not affect the detection task.
The artifacts may be hyper-enhanced features of the true image.
Table 1. Results for the combinations of initial number
of channels and amount of dropout. The
choice that did consistently well across all metrics we considered was 64
channels with 0.1 dropout but the network with 64 channels and 0.3 dropout has very
similar human observer performance. All networks perform similarly for human
detection except for the network with 64 channels and 0 dropout. We also see that the approximation to the
ideal observer (LG-CHO) performs similarly except for the networks with 32
channels and 0.3 dropout and 64 channels and 0 dropout.
