Vessel tortuosity extraction from IVUS images

When looking at an Intravascular Ultrasound (IVUS) pullback image data, the lack of three-dimensional information does not allow to know the real shape of the vessel we are observing, and thus to infer whether the plaque is more probable to be deposited. The main objective of this work is to establish a mechanism to infer vessel extreme curvature points from the position of the IVUS catheter inside the blood vessel observed in the IVUS images without any external information. The obtained curvature is validated with its real 3D shape reconstructed using a system we developed before that fuse IVUS and angiography images[l]. We will see that the results are promising, partially revealing the extreme curvature points of the vessel. For the first time, vessel curvature can be partially extracted from IVUS images without any external information. This opens a wide range of possibilities like predicting plaque formation and the fusion with other modalities as optical computer tomography.