Learning Active Shape Models for Bifurcating Contours

Abstract

Statistical shape models are often learned from examples based on landmark correspondences between annotated examples. A method is proposed for learning such models from contours with inconsistent bifurcations and loops. Automatic segmentation of tibial and femoral contours in knee X-ray images is investigated as a step towards reliable, quantitative radiographic analysis of osteoarthritis for diagnosis and assessment of progression. Results are presented using various features, the Mahalanobis distance, distance weighted K-nearest neighbours, and two relevance vector machine-based methods as quality of fit measure.

Unsupervised segmentation of the vocal tract from real-time MRI sequences

“…This happened because such a model encompasses much more variability in the velar region which, considering the noise in the images, was often a problem. An approach based on bifurcating contours (Seise et al, 2007) might be suitable, but added complexity which we considered avoidable. We opted for building a different model for each situation (hereafter referred to as the nasal model, which contains a nasal cavity, and the oral model) and define rules to decide, at each image frame, which model to use.…”

Facial point detection using boosted regression and graph models

“…Cristinacce and Cootes [4] use GentleBoost regression within the Active Shape Model (ASM) search framework to detect 20 facial points. Seise et al [20] use the ASM framework together with a Relevance Vector Machine regressor to track the contours of lips. However, their approach was tested on only a single image sequence.…”

Enhanced Distal Radius Segmentation in DXA Using Modified ASM

“…Because a bone generally has a specific shape according to its type, segmentation based on the active shape model (ASM) has been widely accepted for extracting the shape of bones from digital radiographic images [5]. Given the statistical shape model of a bone computed from a set of labeled training images, ASM iteratively tries to fit the model to the contour of the bones in a set of test images.…”

“…Rebuffel et al [5] proposed a method for modeling and segmenting contours with inconsistent loops and bifurcations. It was observed that segmentation errors occur because the number of loops and the position of the bifurcation points on an object may vary in a complex way.…”