Accurate and consistent segmentation plays an important role in the diagnosis, treatment planning, and monitoring of both High Grade Glioma (HGG), including Glioblastoma Multiforme (GBM), and Low Grade Glioma (LGG). Accuracy of segmentation can be affected by the imaging presentation of glioma, which greatly varies between the two tumor grade groups. In recent years, researchers have used Machine Learning (ML) to segment tumor rapidly and consistently, as compared to manual segmentation. However, existing ML validation relies heavily on computing summary statistics and rarely tests the generalizability of an algorithm on clinically heterogeneous data. In this work, our goal is to investigate how to holistically evaluate the performance of ML algorithms on a brain tumor segmentation task. We address the need for rigorous evaluation of ML algorithms and present four axes of model evaluation—diagnostic performance, model confidence, robustness, and data quality. We perform a comprehensive evaluation of a glioma segmentation ML algorithm by stratifying data by specific tumor grade groups (GBM and LGG) and evaluate these algorithms on each of the four axes. The main takeaways of our work are—(1) ML algorithms need to be evaluated on out-of-distribution data to assess generalizability, reflective of tumor heterogeneity. (2) Segmentation metrics alone are limited to evaluate the errors made by ML algorithms and their describe their consequences. (3) Adoption of tools in other domains such as robustness (adversarial attacks) and model uncertainty (prediction intervals) lead to a more comprehensive performance evaluation. Such a holistic evaluation framework could shed light on an algorithm's clinical utility and help it evolve into a more clinically valuable tool.
Deep Neural Networks (DNNs) can provide clinicians with fast and accurate predictions that are highly valuable for high-stakes medical decision-making, such as in brain tumor segmentation and treatment planning. However, these models largely lack transparency about the uncertainty in their predictions, potentially giving clinicians a false sense of reliability that may lead to grave consequences in patient care. Growing calls for Transparent and Responsible AI have promoted Uncertainty Quantification (UQ) to capture and communicate uncertainty in a systematic and principled manner. However, traditional Bayesian UQ methods remain prohibitively costly for large, million-dimensional tumor segmentation DNNs such as the U-Net. In this work, we discuss a computationally-efficient UQ approach via the partially Bayesian neural networks (pBNN). In pBNN, only a single layer, strategically selected based on gradient-based sensitivity analysis, is targeted for Bayesian inference. We illustrate the effectiveness of pBNN in capturing the full uncertainty for a 7.8-million parameter U-Net. We also demonstrate how practitioners and model developers can use the pBNN's predictions to better understand the model's capabilities and behavior.
28 February 2023 in ACM Transactions on Computer-Human Interaction
Computational models that formalize complex human behaviors enable study and understanding of such behaviors. However, collecting behavior data required to estimate the parameters of such models is often tedious and resource intensive. Thus, estimating dataset size as part of data collection planning (also known as Sample Size Determination) is important to reduce the time and effort of behavior data collection while maintaining an accurate estimate of model parameters. In this article, we present a sample size determination method based on Uncertainty Quantification (UQ) for a specific Inverse Reinforcement Learning (IRL) model of human behavior, in two cases: (1)pre-hocexperiment design—conducted in the planning stage before any data is collected, to guide the estimation of how many samples to collect; and (2)post-hocdataset analysis—performed after data is collected, to decide if the existing dataset has sufficient samples and whether more data is needed. We validate our approach in experiments with a realistic model of behaviors of people with Multiple Sclerosis (MS) and illustrate how to pick a reasonable sample size target. Our work enables model designers to perform a deeper, principled investigation of the effects of dataset size on IRL model parameters.
Infrastructure in Detroit cannot handle flooding so it backs up to the Arsenal … There is a worry of research and mission capacity (assets) loss during flooding events [44A].Precipitation in the Un...