Background Because effective prediction of survival time can be highly beneficial for the treatment of glioblastoma patients, the relationship between survival time and multiple patient characteristics has been investigated. In this paper, we investigate whether the predictive power of a survival model based on clinical patient features improves when MRI features are also included in the model. Methods The subjects in this study were 82 glioblastoma patients for whom clinical features as well as MR imaging exams were made available by The Cancer Genome Atlas (TCGA) and The Cancer Imaging Archive (TCIA). Twenty-six imaging features in the available MR scans were assessed by radiologists from the TCGA Glioma Phenotype Research Group. We used multivariate Cox proportional hazards regression to construct 2 survival models: one that used 3 clinical features (age, gender, and KPS) as the covariates and 1 that used both the imaging features and the clinical features as the covariates. Then, we used 2 measures to compare the predictive performance of these 2 models: area under the receiver operating characteristic curve for the 1-year survival threshold and overall concordance index. To eliminate any positive performance estimation bias, we used leave-one-out cross-validation. Results The performance of the model based on both clinical and imaging features was higher than the performance of the model based on only the clinical features, in terms of both area under the receiver operating characteristic curve (P < .01) and the overall concordance index (P < .01). Conclusions Imaging features assessed using a controlled lexicon have additional predictive value compared with clinical features when predicting survival time in glioblastoma patients.
In this work we investigate the use of deep inverse models (DIMs) for designing artificial electromagnetic materials (AEMs) – such as metamaterials, photonic crystals, and plasmonics – to achieve some desired scattering properties (e.g., transmission or reflection spectrum). DIMs are deep neural networks (i.e., deep learning models) that are specially-designed to solve ill-posed inverse problems. There has recently been tremendous growth in the use of DIMs for solving AEM design problems however there has been little comparison of these approaches to examine their absolute and relative performance capabilities. In this work we compare eight state-of-the-art DIMs on three unique AEM design problems, including two models that are novel to the AEM community. Our results indicate that DIMs can rapidly produce accurate designs to achieve a custom desired scattering on all three problems. Although no single model always performs best, the Neural-Adjoint approach achieves the best overall performance across all problem settings. As a final contribution we show that not all AEM design problems are ill-posed, and in such cases a conventional deep neural network can perform better than DIMs. We recommend that a deep neural network is always employed as a simple baseline approach when addressing AEM design problems. We publish python code for our AEM simulators and our DIMs to enable easy replication of our results, and benchmarking of new DIMs by the AEM community.
We demonstrate transfer learning as a tool to improve the efficacy of training deep learning models based on residual neural networks (ResNets). Specifically, we examine its use for study of multi‐scale electrically large metasurface arrays under open boundary conditions in electromagnetic metamaterials. Our aim is to assess the efficiency of transfer learning across a range of problem domains that vary in their resemblance to the original base problem for which the ResNet model was initially trained. We use a quasi‐analytical discrete dipole approximation (DDA) method to simulate electrically large metasurface arrays to obtain ground truth data for training and testing of our deep neural network. Our approach can save significant time for examining novel metasurface designs by harnessing the power of transfer learning, as it effectively mitigates the pervasive data bottleneck issue commonly encountered in deep learning. We demonstrate that for the best case when the transfer task is sufficiently similar to the target task, a new task can be effectively trained using only a few data points yet still achieve a test mean absolute relative error of 3 % with a pre‐trained neural network, realizing data reduction by a factor of 1000.
Modern deep neural networks (DNNs) are highly accurate on many recognition tasks for overhead (e.g., satellite) imagery. However, visual domain shifts (e.g., statistical changes due to geography, sensor, or atmospheric conditions) remain a challenge, causing the accuracy of DNNs to degrade substantially and unpredictably when testing on new sets of imagery. In this work, we model domain shifts caused by variations in imaging hardware, lighting, and other conditions as non-linear pixel-wise transformations, and we perform a systematic study indicating that modern DNNs can become largely robust to these types of transformations, if provided with appropriate training data augmentation. In general, however, we do not know the transformation between two sets of imagery. To overcome this, we propose a fast real-time unsupervised training augmentation technique, termed randomized histogram matching (RHM). We conduct experiments with two large benchmark datasets for building segmentation and find that despite its simplicity, RHM consistently yields similar or superior performance compared to state-of-the-art unsupervised domain adaptation approaches, while being significantly simpler and more computationally efficient. RHM also offers substantially better performance than other comparably simple approaches that are widely used for overhead imagery.