Training models of anatomic shape variability

Abstract

Learning probability distributions of the shape of anatomic structures requires fitting shape representations to human expert segmentations from training sets of medical images. The quality of statistical segmentation and registration methods is directly related to the quality of this initial shape fitting, yet the subject is largely overlooked or described in an ad hoc way. This article presents a set of general principles to guide such training. Our novel method is to jointly estimate both the best geometric model for any given image and the shape distribution for the entire population of training images by iteratively relaxing purely geometric constraints in favor of the converging shape probabilities as the fitted objects converge to their target segmentations. The geometric constraints are carefully crafted both to obtain legal, nonself-interpenetrating shapes and to impose the model-tomodel correspondences required for useful statistical analysis. The paper closes with example applications of the method to synthetic and real patient CT image sets, including same patient male pelvis and head and neck images, and cross patient kidney and brain images. Finally, we outline how this shape training serves as the basis for our approach to IGRT/ART.

Non-Euclidean classification of medically imaged objects via s-reps

“…S-reps were fitted using shape statistics drawn from the set where both schizophrenic cases and control cases were pooled together. Detailed description of the actual s-rep fitting procedure can be found in (Schulz et al, 2013b; Merck et al, 2008). …”

Value of an Electronic Tutorial for Image Interpretation in Ultrasound-Guided Regional Anesthesia

“…16 When expertise in pattern recognition progresses, small anatomical variations will not confuse the novices. 17 We considered participants as inexperienced for up to 30 UGRA procedures. It is still unknown how many procedures are needed to achieve any competence in UGRA.…”

Femoral Head Shape Differences During Development May Identify Hips at Risk of Degeneration

“…28 The m-rep was deformed to fit the binary image of an SCO by working from the largest to the smallest scale. 22,26 The smaller scale refined the fit at the larger scale. In the initialization stage, the template m-rep was geometrically aligned with and scaled to the binary SCO image via landmarks placed by a single researcher with expertise in femoral head morphology (WVBF).…”

“…27 A fixed medial topology was deformed to model the shape of each individual SCO. 22,[25][26][27] The fixed topology mitigates sensitivity to boundary uncertainty, incorporates a priori knowledge of the object shape, and provides a consistent basis for comparison across a population of similar objects. The m-reps consist of a mesh of linked medial primitives called atoms (Figs.…”