Unbiased multiple-subject alignment of left ventricles

Image based quantitative stratification of the left ventricular (LV) pathology helps in understanding the structure-function symbiosis of the heart. However, this requires unbiased, referenceless multi-subject LV alignment that accounts for the physioanatomic variations associated with the pathology. This paper achieves this hitherto elusive alignment by adopting the conventional risk stratification strategies routinely followed by clinicians. The individual LV shape models are independently reoriented to an " attitudinally consistent orientation" that captures the physioanatomic and pathologic variations of the LV morphology. The proposed algorithm is computationally efficient and provides an efficient framework for constructing pathology stratified cardiac atlas.