Abstract
We present a new clinically practical method for online co-registration of 3D quantitative coronary angiography (QCA) and intravascular ultrasound (IVUS) or optical coherence tomography (OCT). The workflow is based on two modified commercially available software packages. Reconstruction steps are explained and compared to previously available methods. The feasibility for different clinical scenarios is illustrated. The co-registration appears accurate, robust and induced a minimal delay on the normal cath lab activities. This new method is based on the 3D angiographic reconstruction of the catheter path and does not require operator’s identification of landmarks to establish the image synchronization.
1
Introduction
Numerous reports of the deficiencies of coronary angiography as the gold standard imaging modality to perform percutaneous coronary intervention (PCI) have been published for more than 20 years, and yet luminology remains the primary guidance in all cath lab around the world , with Japan as an exception where intravascular ultrasound (IVUS) is performed in a majority of cases. Many factors limit routine IVUS usage, and even more optical coherence tomography (OCT), among them principally cost-constraints and debatable results of randomized studies that have tried to demonstrate hard clinical end-point benefits . Other factors might also play an important role, among them a lack of trained technicians to run the imaging consoles and perform the measurements, or insufficient training in IVUS/OCT image interpretation with difficulties to find back on the angiogram important intravascular findings. To this end, we believe that any help in the guidance of an IVUS or OCT examination and an increase in the user-friendliness in interpretation would improve the diagnostic and therapeutic return-of-investment in opening the box of an IVUS or OCT catheter during a PCI. Because it is sometime difficult to localize where an intravascular measurement/region-of-interest comes from in a planar two-dimensional silhouette of a coronary artery, we sought to develop an on-line co-registration between IVUS or OCT pullbacks and 3D reconstructed angiograms. It is always possible to go back to “live” imaging with an IVUS catheter re-advanced to the region of interest visualized during the pullback and then make an angiogram to see the position of the tip of the transducer. However, newer high-speed imaging modalities such as OFDI OCT with pullback speed up to 20 mm/s that also require flushing the lumen with saline or contrast make this approach impracticable and co-registration appears even more important for OCT.
In the following paragraphs the disadvantages of current solutions and the development of our new online co-registration method are explained and illustrated by several example cases showing the feasibility for different clinical scenario with IVUS or OCT.
2
Materials and methods
Two commercially available software packages were modified to communicate and exchange information including dynamic position for co-registration of angiography and IVUS/OCT. Specific additional hardware was necessary to acquire the IVUS data in the co-registration workstation without unacceptable delays for transferring large files. This is crucial in a clinical setting in which the procedure should be kept as simple and short as possible to avoid complications. The latest 4.0 version of the echoPlaque software developed by INDEC Medical systems Inc. (Santa Clara, USA) features acquisition, analysis and review of IVUS and OCT image data . It is installed on the co-registration workstation ( Fig. 1 ) equipped ideally with a two-monitor output to run simultaneously with the CAAS 5.10.2 QCA3D software of Pie Medical Imaging BV (Maastricht, The Netherlands). The CAAS QCA3D utilizes conventionally 2 two-dimensional angiographic images to calculate a 3D reconstruction of the vessel as illustrated on Fig. 2 . Reconstructing the vessel in 3D space overcomes limitations of standard QCA such as foreshortening and out-of-plane magnification . Elimination of these factors facilitates calculation of cross-sectional areas and vessel length with high accuracy, as validated previously . A similar 3D QCA methodology has been reported by Tu and collaborators . We will present a new, more robust approach, based on the real-time automatic co-registration of IVUS or OCT with the 3D angiographic imaging catheter path. Co-registration relies on echoPlaque and CAAS transferring positioning information from one software to the other. For application in the cath lab the co-registration must be accurate, robust and have a minimal delay on the normal cath lab activities. The workflow was created so that there is no delay on cath lab activities due to transferring large IVUS files from the console to the co-registration workstation burning DVDs. DICOM OCT files are transferred directly to the co-registration workstation.
2.1
Workflow using conventional 3D angiography
Two standard angiograms ( Fig. 2 ) are recorded and sent as DICOM XA modality files to the co-registration workstation. The user selects end-diastolic frames, indicates the vessel part of interest in both angiograms and a 3D model of the vessel is automatically computed. True length of this vessel part is calculated without foreshortening or out-of-plane magnification errors. Creating the 3D model can be performed during the time in which the IVUS or OCT pullback is performed.
The echoPlaque system has the unique feature of direct acquisition of IVUS image frames during pullback by means of digitizing the signals sent to the device monitor by the IVUS system. The normal IVUS screen remains available to the cardiologist for clinical decision during the pullback and in parallel there is in echoPlaque an automatic extraction of the image pixels that show the IVUS “tissue ball”. This feature is the cornerstone of the live co-registration. Otherwise, long pullback files that can be as large as a couple of Gb need to be transferred via DICOM XA modality. That would require several minutes, an unacceptable delay in the clinical setting. Here, the entire set of pullback frames is available for viewing in the workstation at the end of the pullback, saving significant time. Co-registration requires an automated pullback to ensure a constant frame-to-frame distance so that positions in the image set can be accurately related to physical positions in the artery.
After the pullback is finished, the user indicates one common anatomical position (landmark) on both the angiography-based 3D model (or on one of the two angiograms) and the IVUS pullback, e.g. the carina of a clearly visible bifurcation ( Fig. 3 ).
Based on this landmark, actual co-registration of both imaging modalities is performed. For each IVUS/OCT frame its position on the 3D angiographic model (and subsequently both angiograms) is determined using a distance mapping algorithm.
After the co-registration is performed the user can add bookmarks or other graphical indications of regions such as a stent with a predefined standard length so that the landing zone on both the angiograms and the IVUS/OCT pullback are automatically shown.
As an alternative to this live workflow, the workstation can also be used in a core-laboratory mode where angiograms and IVUS/OCT are read from CD/DVDs and the analysis is performed off-line.
2.2
Workflow using the 3D angiographic imaging catheter path
We found several disadvantages of the method based on the 3D QCA for co-registration. The 3D QCA requires one common angiographic landmark to be selected by the user and is sometime difficult to obtain due to overlap of coronary side branches or suboptimal visualization in two orthogonal views. Further, manual indication of one common anatomical position is needed in both QCA and IVUS/OCT modalities, with additional inter- and intraobserver variability. Moreover, two angiographic projections filled with contrast are needed.
Co-registration using the 3D catheter path of the imaging catheter/wire is as simple as ABC:
- A.
Advancing imaging catheter.
- B.
Biplane or two angiograms (preferably orthogonal, at least 30 degrees apart) are acquired.
- C.
Contrast injection in one of the angiograms used as the roadmap image. Angiograms are sent via DICOM to the co-registration workstation.
- D.
Digital transfer of the motorized IVUS/OCT pullback streamed real-time to the co-registration workstation or sent via DICOM.
In both angiograms the user indicates the 2D path of the catheter starting on the imaging tip in end-diastolic frames ( Fig. 4 ). Thereafter the system automatically calculates the true 3D catheter path length without foreshortening or out-of-plane magnification. IVUS/OCT co-registration is then based on the starting point of the trajectory (the ultrasound crystal or the tip of the OCT imaging wire) that will be matched to the first recorded IVUS or OCT frame. From this starting point a distance mapping algorithm co-registers each IVUS or OCT frame with the 3D catheter path.
2
Materials and methods
Two commercially available software packages were modified to communicate and exchange information including dynamic position for co-registration of angiography and IVUS/OCT. Specific additional hardware was necessary to acquire the IVUS data in the co-registration workstation without unacceptable delays for transferring large files. This is crucial in a clinical setting in which the procedure should be kept as simple and short as possible to avoid complications. The latest 4.0 version of the echoPlaque software developed by INDEC Medical systems Inc. (Santa Clara, USA) features acquisition, analysis and review of IVUS and OCT image data . It is installed on the co-registration workstation ( Fig. 1 ) equipped ideally with a two-monitor output to run simultaneously with the CAAS 5.10.2 QCA3D software of Pie Medical Imaging BV (Maastricht, The Netherlands). The CAAS QCA3D utilizes conventionally 2 two-dimensional angiographic images to calculate a 3D reconstruction of the vessel as illustrated on Fig. 2 . Reconstructing the vessel in 3D space overcomes limitations of standard QCA such as foreshortening and out-of-plane magnification . Elimination of these factors facilitates calculation of cross-sectional areas and vessel length with high accuracy, as validated previously . A similar 3D QCA methodology has been reported by Tu and collaborators . We will present a new, more robust approach, based on the real-time automatic co-registration of IVUS or OCT with the 3D angiographic imaging catheter path. Co-registration relies on echoPlaque and CAAS transferring positioning information from one software to the other. For application in the cath lab the co-registration must be accurate, robust and have a minimal delay on the normal cath lab activities. The workflow was created so that there is no delay on cath lab activities due to transferring large IVUS files from the console to the co-registration workstation burning DVDs. DICOM OCT files are transferred directly to the co-registration workstation.
