A hybrid multidimensional image segmentation algorithm is proposed, which combines edge and region-based techniques through the morphological algorithm of watersheds. An edge-preserving statistical noise reduction approach is used as a preprocessing stage in order to compute an accurate estimate of the image gradient. Then, an initial partitioning of the image into primitive regions is produced by applying the watershed transform on the image gradient magnitude. This initial segmentation is the input to a computationally efficient hierarchical (bottom-up) region merging process that produces the final segmentation. The latter process uses the region adjacency graph (RAG) representation of the image regions. At each step, the most similar pair of regions is determined (minimum cost RAG edge), the regions are merged and the RAG is updated. Traditionally, the above is implemented by storing all RAG edges in a priority queue. We propose a significantly faster algorithm, which additionally maintains the so-called nearest neighbor graph, due to which the priority queue size and processing time are drastically reduced. The final segmentation provides, due to the RAG, one-pixel wide, closed, and accurately localized contours/surfaces. Experimental results obtained with two-dimensional/three-dimensional (2-D/3-D) magnetic resonance images are presented.

A method for extraction and labeling of the coronary arterial tree (CAT) using minimal user supervision in single-view angiograms is proposed. The CAT structural description (skeleton and borders) is produced, along with quantitative information for the artery dimensions and assignment of coded labels, based on a given coronary artery model represented by a graph. The stages of the method are: 1) CAT tracking and detection; 2) artery skeleton and border estimation; 3) feature graph creation; and iv) artery labeling by graph matching. The approximate CAT centerline and borders are extracted by recursive tracking based on circular template analysis. The accurate skeleton and borders of each CAT segment are computed, based on morphological homotopy modification and watershed transform. The approximate centerline and borders are used for constructing the artery segment enclosing area (ASEA), where the defined skeleton and border curves are considered as markers. Using the marked ASEA, an artery gradient image is constructed where all the ASEA pixels (except the skeleton ones) are assigned the gradient magnitude of the original image. The artery gradient image markers are imposed as its unique regional minima by the homotopy modification method, the watershed transform is used for extracting the artery segment borders, and the feature graph is updated. Finally, given the created feature graph and the known model graph, a graph matching algorithm assigns the appropriate labels to the extracted CAT using weighted maximal cliques on the association graph corresponding to the two given graphs. Experimental results using clinical digitized coronary angiograms are presented.

Background: Fetal cardiovascular MRI complements ultrasound to assess fetal cardiovascular pathophysiology.Purpose: To develop a free‐breathing method for retrospective fetal cine MRI using Doppler ultrasound (DUS) cardiac gating and tiny golden angle radial sampling (tyGRASP) for accelerated acquisition capable of detecting fetal movements for motion compensation.Study Type: Feasibility study.Subjects: Nine volunteers (gestational week 34–40). Short‐axis and four‐chamber views were acquired during maternal free‐breathing and breath‐hold.Field Strength/Sequence: 1.5T cine balanced steady‐state free precession.Assessment: A self‐gated reconstruction method was improved for clinical application by using 1) retrospective DUS gating, and 2) motion detection and rejection/correction algorithms for compensating for fetal motion. The free‐breathing reconstructions were qualitatively and quantitatively assessed, and DUS‐gating was compared with self‐gating in breath‐hold reconstructions. A scoring of 1–4 for overall image quality, cardiac, and extracardiac diagnostic quality was used.Statistical Tests: Friedman's test was used to assess differences in qualitative scoring between observers. A Wilcoxon matched‐pairs signed rank test was used to assess differences between breath‐hold and free‐breathing acquisitions and between observers' quantitative measurements.Results: In all cases, 111 free‐breathing and 145 breath‐hold acquisitions, the automatically calculated DUS‐based cardiac gating signal provided reconstructions of diagnostic quality (median score 4, range 1–4). Free‐breathing did not affect the DUS‐based cardiac gated retrospective radial reconstruction with respect to image or diagnostic quality (all P > 0.06). Motion detection with rejection/correction in k‐space produced high‐quality free‐breathing DUS‐based reconstructions [median 3, range (2–4)], whereas free‐breathing self‐gated methods failed in 80 out of 88 cases to produce a stable gating signal.Data Conclusion: Free‐breathing fetal cine cardiac MRI based on DUS gating and tyGRASP with motion compensation yields diagnostic images. This simplifies acquisition for the pregnant woman and thus could help increase fetal cardiac MRI acceptance in the clinic.Level of Evidence: 2Technical Efficacy Stage: 1 J. Magn. Reson. Imaging 2020;51:260–272.

Supervised Machine Learning techniques have shown significant potential in medical image analysis. However, the training data that need to be collected for these techniques in the field of MRI 1) may not be available, 2) may be available but the size is small, 3) may be available but not representative and 4) may be available but with weak labels. The aim of this study was to overcome these limitations through advanced MR simulations on a realistic computer model of human anatomy without using a real MRI scanner, without scanning patients and without having personnel and the associated expenses. The 4D-XCAT model was used with the coreMRI simulation platform for generating artificial short-axis MR-images for training a neural-network to automatic delineate the LV endocardium and epicardium. Its performance was assessed on real MR-images acquired from eight healthy volunteers. The neural-network was also trained on real MR-images from a publicly available dataset and its performance was assessed on the same volunteers' data. The proposed solution demonstrated a performance of 94% (endocardium) and 90% DICE (epicardium) in real mid-ventricular slices, whereas a 10% addition of real MR-images in the artificial training dataset increased the performance to 97% DICE. The use of artificial MR-images that cover the entire LV yielded 85% (endocardium) and 88% DICE (epicardium) when combined with real MR data with an 80%-20% mix respectively. This study suggests a low-cost solution for constructing artificial training datasets for supervised learning techniques in the field of MR by using advanced MR simulations without the use of a real MRI scanner, without scanning patients and without having to use specialized personnel, such as technologists and radiologists.