Optimization of Deep Learning Based Brain Extraction in MRI for Low Resource Environments.

Brain extraction is an indispensable step in neuro-imaging with a direct impact on downstream analyses. Most such methods have been developed for non-pathologically affected brains, and hence tend to suffer in performance when applied on brains with pathologies, e.g., gliomas, multiple sclerosis, traumatic brain injuries. Deep Learning (DL) methodologies for healthcare have shown promising results, but their clinical translation has been limited, primarily due to these methods suffering from i) high computational cost, and ii) specific hardware requirements, e.g., DL acceleration cards. In this study, we explore the potential of mathematical optimizations, towards making DL methods amenable to application in low resource environments. We focus on both the qualitative and quantitative evaluation of such optimizations on an existing DL brain extraction method, designed for pathologically-affected brains and agnostic to the input modality. We conduct direct optimizations and quantization of the trained model (i.e., prior to inference on new data). Our results yield substantial gains, in terms of speedup, latency, through-put, and reduction in memory usage, while the segmentation performance of the initial and the optimized models remains stable, i.e., as quantified by both the Dice Similarity Coefficient and the Hausdorff Distance. These findings support post-training optimizations as a promising approach for enabling the execution of advanced DL methodologies on plain commercial-grade CPUs, and hence contributing to their translation in limited- and low- resource clinical environments.

Systematic Evaluation of Image Tiling Adverse Effects on Deep Learning Semantic Segmentation.

Convolutional neural network (CNN) models obtain state of the art performance on image classification, localization, and segmentation tasks. Limitations in computer hardware, most notably memory size in deep learning accelerator cards, prevent relatively large images, such as those from medical and satellite imaging, from being processed as a whole in their original resolution. A fully convolutional topology, such as U-Net, is typically trained on down-sampled images and inferred on images of their original size and resolution, by simply dividing the larger image into smaller (typically overlapping) tiles, making predictions on these tiles, and stitching them back together as the prediction for the whole image. In this study, we show that this tiling technique combined with translationally-invariant nature of CNNs causes small, but relevant differences during inference that can be detrimental in the performance of the model. Here we quantify these variations in both medical (i.e., BraTS) and non-medical (i.e., satellite) images and show that training a 2D U-Net model on the whole image substantially improves the overall model performance. Finally, we compare 2D and 3D semantic segmentation models to show that providing CNN models with a wider context of the image in all three dimensions leads to more accurate and consistent predictions. Our results suggest that tiling the input to CNN models-while perhaps necessary to overcome the memory limitations in computer hardware-may lead to undesirable and unpredictable errors in the model's output that can only be adequately mitigated by increasing the input of the model to the largest possible tile size.

Non-Boolean computing with nanomagnets for computer vision applications.

The field of nanomagnetism has recently attracted tremendous attention as it can potentially deliver low-power, high-speed and dense non-volatile memories. It is now possible to engineer the size, shape, spacing, orientation and composition of sub-100 nm magnetic structures. This has spurred the exploration of nanomagnets for unconventional computing paradigms. Here, we harness the energy-minimization nature of nanomagnetic systems to solve the quadratic optimization problems that arise in computer vision applications, which are computationally expensive. By exploiting the magnetization states of nanomagnetic disks as state representations of a vortex and single domain, we develop a magnetic Hamiltonian and implement it in a magnetic system that can identify the salient features of a given image with more than 85% true positive rate. These results show the potential of this alternative computing method to develop a magnetic coprocessor that might solve complex problems in fewer clock cycles than traditional processors.