Explainable artificial intelligence model to predict acute critical illness from electronic health records.

Acute critical illness is often preceded by deterioration of routinely measured clinical parameters, e.g., blood pressure and heart rate. Early clinical prediction is typically based on manually calculated screening metrics that simply weigh these parameters, such as early warning scores (EWS). The predictive performance of EWSs yields a tradeoff between sensitivity and specificity that can lead to negative outcomes for the patient. Previous work on electronic health records (EHR) trained artificial intelligence (AI) systems offers promising results with high levels of predictive performance in relation to the early, real-time prediction of acute critical illness. However, without insight into the complex decisions by such system, clinical translation is hindered. Here, we present an explainable AI early warning score (xAI-EWS) system for early detection of acute critical illness. xAI-EWS potentiates clinical translation by accompanying a prediction with information on the EHR data explaining it.

Automatic Bone Marrow Cellularity Estimation in H&E Stained Whole Slide Images.

Bone marrow cellularity is an important measure in diagnostic hematopathology. Currently, the gold standard for bone marrow cellularity estimation is manual inspection of hematoxylin and eosin stained whole slide images (H&E WSI) by hematopathologists. However, these assessments are subjective and subject to interobserver and intraobserver variability. This may be reduced by using a computer-assisted estimate of bone marrow cellularity. The aim of this study was to develop a fully automated algorithm to estimate bone marrow cellularity in H&E WSI stains using bone marrow segmentation. Data consisted of eight bone marrow H&E WSIs extracted from eight subjects. An algorithm was developed to estimate the bone marrow cellularity consisting of biopsy segmentation, tissue classification, and bone marrow segmentation. Segmentations of the red and yellow bone marrow (YBM) were used to estimate the bone marrow cellularity within the WSI H&E stains. The DICE coefficient between automatic tissue segmentations and ground truth segmentations conducted by an experienced hematopathologist were used for validation. Furthermore, the agreement between the automatic and two manual cellularity estimates was assessed using Bland-Altman plots and intraclass correlation coefficients (ICC). The validation of the bone marrow segmentation demonstrated an average DICE of 0.901 and 0.920 for the red and YBM, respectively. A mean cellularity estimate difference of -0.552 and - 7.816 was obtained between the automatic cellularity estimates and two manual cellularity estimates, respectively. An ICC of 0.980 (95%CI: 0.925-0.995, P-value: 5.51 × 10-7 ) was obtained between the automatic and manual cellularity estimates based on manual annotations. The study demonstrated that it was possible to obtain bone marrow cellularity estimates with a good agreement with bone marrow cellularity estimates obtained from an experienced hematopathologist. © 2019 International Society for Advancement of Cytometry.

SHARQnet - Sophisticated harmonic artifact reduction in quantitative susceptibility mapping using a deep convolutional neural network.

Quantitative susceptibility mapping (QSM) reveals pathological changes in widespread diseases such as Parkinson's disease, Multiple Sclerosis, or hepatic iron overload. QSM requires multiple processing steps after the acquisition of magnetic resonance imaging (MRI) phase measurements such as unwrapping, background field removal and the solution of an ill-posed field-to-source-inversion. Current techniques utilize iterative optimization procedures to solve the inversion and background field correction, which are computationally expensive and lead to suboptimal or over-regularized solutions requiring a careful choice of parameters that make a clinical application of QSM challenging. We have previously demonstrated that a deep convolutional neural network can invert the magnetic dipole kernel with a very efficient feed forward multiplication not requiring iterative optimization or the choice of regularization parameters. In this work, we extended this approach to remove background fields in QSM. The prototype method, called SHARQnet, was trained on simulated background fields and tested on 3T and 7T brain datasets. We show that SHARQnet outperforms current background field removal procedures and generalizes to a wide range of input data without requiring any parameter adjustments. In summary, we demonstrate that the solution of ill-posed problems in QSM can be achieved by learning the underlying physics causing the artifacts and removing them in an efficient and reliable manner and thereby will help to bring QSM towards clinical applications.