Retraining Convolutional Neural Networks for Specialized Cardiovascular Imaging Tasks: Lessons from Tetralogy of Fallot.

Ventricular contouring of cardiac magnetic resonance imaging is the gold standard for volumetric analysis for repaired tetralogy of Fallot (rTOF), but can be time-consuming and subject to variability. A convolutional neural network (CNN) ventricular contouring algorithm was developed to generate contours for mostly structural normal hearts. We aimed to improve this algorithm for use in rTOF and propose a more comprehensive method of evaluating algorithm performance. We evaluated the performance of a ventricular contouring CNN, that was trained on mostly structurally normal hearts, on rTOF patients. We then created an updated CNN by adding rTOF training cases and evaluated the new algorithm's performance generating contours for both the left and right ventricles (LV and RV) on new testing data. Algorithm performance was evaluated with spatial metrics (Dice Similarity Coefficient (DSC), Hausdorff distance, and average Hausdorff distance) and volumetric comparisons (e.g., differences in RV volumes). The original Mostly Structurally Normal (MSN) algorithm was better at contouring the LV than the RV in patients with rTOF. After retraining the algorithm, the new MSN + rTOF algorithm showed improvements for LV epicardial and RV endocardial contours on testing data to which it was naïve (N = 30; e.g., DSC 0.883 vs. 0.905 for LV epicardium at end diastole, p < 0.0001) and improvements in RV end-diastolic volumetrics (median %error 8.1 vs 11.4, p = 0.0022). Even with a small number of cases, CNN-based contouring for rTOF can be improved. This work should be extended to other forms of congenital heart disease with more extreme structural abnormalities. Aspects of this work have already been implemented in clinical practice, representing rapid clinical translation. The combined use of both spatial and volumetric comparisons yielded insights into algorithm errors.

Development of functional trait biomarkers for trace metal exposure in freshwater clams (Musculium spp.).

Exposure to trace metals typically causes oxidative stress; these consequences are better-characterized in estuarine and marine species than in freshwater species. How cellular-level responses to metal pollution influence whole-organism and population-level traits is poorly understood. We tested whether exposure to single metals (zinc and cadmium) and to metal mixtures (water in equilibrium with sediment from a highly polluted lake) alters two ecologically-relevant traits in freshwater clams, locomotion and reproduction. Fingernail clams (Musculium spp.) from unimpacted habitats were exposed to single metals and the metal mixture for up to 49days. The single metal doses (≤5mg/L Zn and ≤20µg/L Cd) were not toxicologically meaningful as clam survival, burial, and climbing activity did not differ across treatments. Water in equilibrium with the lake sediment contained cadmium, copper, lead, and zinc. Clams exposed to this metal mixture had decreased climbing activity but no change in burial activity. Metal-exposed clams had lower fecundity (number of shelled juveniles extruded by adult clams) and patterns in metal accumulation corresponded with lake sediment dose and clam activity. In contrast to the functional traits, stress protein expression and whole-clam glycogen content did not vary across treatment groups. These results indicate that fingernail clams of the genus Musculium are appropriate for development as sentinel species for metal pollution and can serve as a model for determining how metal pollution alters metabolic allocation patterns in freshwater organisms.

Electrocardiogram Belt guidance for left ventricular lead placement and biventricular pacing optimization.

BACKGROUND: Patients with ischemic cardiomyopathy, non-left bundle branch block, or QRS duration <150 ms have a lower response rate to cardiac resynchronization therapy (CRT) than did other indicated patients. The ECG Belt system (EBS) is a novel surface mapping system designed to measure electrical dyssynchrony via the standard deviation of the activation times of the left ventricle. OBJECTIVES: The objectives of this study were to evaluate the efficacy of the EBS in patients less likely to respond to CRT and to determine whether EBS use in lead placement guidance and device programming was superior to standard CRT care. METHODS: This was a prospective randomized trial of patients with heart failure and EBS-guided CRT implantation and programming vs standard CRT care. The primary end point was relative change in left ventricular end-systolic volume from baseline to 6 months postimplantation. RESULTS: A total of 408 patients from centers in Europe and North America were randomized. Although both patients with EBS and control patients had a mean improvement in left ventricular end-systolic volume, there was no significant difference in relative change from baseline (P = .26). While patients with a higher baseline standard deviation of the activation times derived greater left ventricular reverse remodeling, improvement in electrical dyssynchrony did not correlate with the extent of reverse remodeling. CONCLUSION: The findings of the present study do not support EBS-guided therapy for CRT management of heart failure with reduced ejection fraction.

Predictor-Response Analysis of Fiber Optic Enzymatic Biosensors Constructed with Nonmodified E. coli BL21 (DE3) pGELAF+ Sensing 1,2-Dichloroethane.

Biosensor technology can lack methods to iteratively validate system outputs (i.e., signals) concomitantly with the development of mathematical models. We evaluated a nonmodified fiber optic enzymatic biosensor (FOEB-Escherichia coli BL21 (DE3) pGELAF+) sensing dichloroethane with a predictor-response statistical form. The linear regression technique applied with MATLAB functions correlated FOEB parameters to sensing responses that could be used to identify system characteristics and interactions. A FOEB specific metric (i.e. normalized sensitivity) is shown to be significant as a mixed sensing correlation metric suggesting that similar development parameters could be related to engineering design paradigms for biosensor (or whole cell biosensor) systems.