5D Image Guided Cardiac Ablation

We propose to construct a computer system that will accept all of the dynamic during-procedure data streams for a CEP ablation procedure, and to devise a software system for integrating these signals into immediate, dynamic anatomic images suitable for guiding the procedure. These signals include fluoroscope and ultrasound video streams, fluoroscope gantry angle encoders, and various switch inputs. Cardiac electrophysiology signals will be digitized and analyzed by a commercial CEP machine, and the reduced data communicated to the visualization system via Ethernet, as will the coordinate data from tracked catheters.

The hardware system (see Figure 1) is composed of a cluster of 4 single CPU Intel-x86 compute servers communicating via gigabit Ethernet for non-mission critical, non-deterministic data transfer and an ultra-low latency SCRAMnet (networked shared memory) for mission critical, deterministic communication. Computational tasks are distributed across the cluster in the form of small intercommunicating applications dedicated to specific tasks. This allows for the simultaneous processing of a variety of data inputs in an efficient manner as possible while limiting the complexity of any single application. By reducing the complexity of the component applications, development and subsequent maintenance costs should be reduced.