Patient-specific three-dimensional torso models for analysing cardiac activity

Standard electrocardiogram (ECG) is routinely used for recording cardiac electrical activity but lacks 3D details. Body surface potential mapping techniques have been developed more recently. A major challenge related to this technique is the projection of body signals back to their original sources in the heart. For an accurate projection, one needs to take into account patient-specific electrophysiological data of tissues surrounding the heart. Much information regarding physiological variations, as well as the exact position of organs in the torso, can be obtained from magnetic resonance (MR) images. Here, a patient-specific methodology for building 3D torso models from transverse MR images is proposed. Torso contour detection is based on edge detection using a canny filter and indicating contour points with a polar coordinate system. Organ detection is performed using an interpolation technique with Active Contour modelling. Results show that accurate torso models can be constructed with short processing time.