3 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;48:1570-1577.

Score-based generative models can produce high quality image samples comparable to GANs, without requiring adversarial optimization. However, existing training procedures are limited to images of low resolution (typically below 32 × 32), and can be unstable under some settings. We provide a new theoretical analysis of learning and sampling from score models in high dimensional spaces, explaining existing failure modes and motivating new solutions that generalize across datasets. To enhance stability, we also propose to maintain an exponential moving average of model weights. With these improvements, we can effortlessly scale score-based generative models to images with unprecedented resolutions ranging from 64 × 64 to 256 × 256. Our score-based models can generate high-fidelity samples that rival best-in-class GANs on various image datasets, including CelebA, FFHQ, and multiple LSUN categories.

In radiomics studies, researchers usually need to develop a supervised machine learning model to map image features onto the clinical conclusion. A classical machine learning pipeline consists of several steps, including normalization, feature selection, and classification. It is often tedious to find an optimal pipeline with appropriate combinations. We designed an open-source software package named FeAture Explorer (FAE). It was programmed with Python and used NumPy, pandas, and scikit-learning modules. FAE can be used to extract image features, preprocess the feature matrix, develop different models automatically, and evaluate them with common clinical statistics. FAE features a user-friendly graphical user interface that can be used by radiologists and researchers to build many different pipelines, and to compare their results visually. To prove the effectiveness of FAE, we developed a candidate model to classify the clinical-significant prostate cancer (CS PCa) and non-CS PCa using the PROSTATEx dataset. We used FAE to try out different combinations of feature selectors and classifiers, compare the area under the receiver operating characteristic curve of different models on the validation dataset, and evaluate the model using independent test data. The final model with the analysis of variance as the feature selector and linear discriminate analysis as the classifier was selected and evaluated conveniently by FAE. The area under the receiver operating characteristic curve on the training, validation, and test dataset achieved results of 0.838, 0.814, and 0.824, respectively. FAE allows researchers to build radiomics models and evaluate them using an independent testing dataset. It also provides easy model comparison and result visualization. We believe FAE can be a convenient tool for radiomics studies and other medical studies involving supervised machine learning.

We introduce a new generative model where samples are produced via Langevin dynamics using gradients of the data distribution estimated with score matching. Because gradients might be ill-defined when the data resides on low-dimensional manifolds, we perturb the data with different levels of Gaussian noise and jointly estimate the corresponding scores, i.e., the vector fields of gradients of the perturbed data distribution for all noise levels. For sampling, we propose an annealed Langevin dynamics where we use gradients corresponding to gradually decreasing noise levels as the sampling process gets closer to the data manifold. Our framework allows flexible model architectures, requires no sampling during training or the use of adversarial methods, and provides a learning objective that can be used for principled model comparisons. Our models produce samples comparable to GANs on MNIST, CelebA and CIFAR-10 datasets, achieving a new state-of-theart inception score of 8.91 on CIFAR-10. Additionally, we demonstrate that our models learn effective representations via image inpainting experiments.Preprint. Under review.

Reconstructing medical images from partial measurements is an important inverse problem in Computed Tomography (CT) and Magnetic Resonance Imaging (MRI). Existing solutions based on machine learning typically train a model to directly map measurements to medical images, leveraging a training dataset of paired images and measurements. These measurements are typically synthesized from images using a fixed physical model of the measurement process, which hinders the generalization capability of models to unknown measurement processes. To address this issue, we propose a fully unsupervised technique for inverse problem solving, leveraging the recently introduced score-based generative models. Specifically, we first train a score-based generative model on medical images to capture their prior distribution. Given measurements and a physical model of the measurement process at test time, we introduce a sampling method to reconstruct an image consistent with both the prior and the observed measurements. Our method does not assume a fixed measurement process during training, and can thus be flexibly adapted to different measurement processes at test time. Empirically, we observe comparable or better performance to supervised learning techniques in several medical imaging tasks in CT and MRI, while demonstrating significantly better generalization to unknown measurement processes.