A detection-driven and sparsity-constrained deformable model for fascia lata labeling and thigh inter-muscular adipose quantification

Quantification of the thigh inter-muscular adipose tissue (IMAT) plays a critical role in various medical data analysis tasks, e.g., the analysis of physical performance or the diagnosis of knee osteoarthritis. Several techniques have been proposed to perform automated thigh tissues quantification. However, none of them has provided an effective method to track fascia lata, which is an important anatomic trail to distinguish between subcutaneous adipose tissue (SAT) and IMAT in the thigh. As a result, the estimates of IMAT may not be accurate due to the unclear appearance cues, complicated anatomic, or pathological characteristics of the fascia lata. Thus, prior tissue information, e.g., intensity, orientation and scale, becomes critical to infer the fascia lata location from magnetic resonance (MR) images. In this paper, we propose a novel detection-driven and sparsity-constrained deformable model to obtain accurate fascia lata labeling. The model's deformation is driven by the detected control points on fascia lata through a discriminative detector in a narrow-band fashion. By using a sparsity-constrained optimization, the deformation is solved from errors and outliers suppression. The proposed approach has been evaluated on a set of 3D MR thigh volumes. In a comparison with the state-of-the-art framework, our approach produces superior performance.