Signal Processing/Automation for Cardiac Output Monitor

Thermal dilution is by far the most common method of measuring cardiac output. Over onemillion thermodilution pulmonary artery catheters are used each year in the U.S. We have appliedartificial neural networks to improve the accuracy of the thermal dilution cardiac output measurementand allow using small injectate volumes (2 ml). We will build an automated system using a modifiedsyringe pump and a solid state injectate cooler to automatically measure cardiac output once everyminute. Recently introduced devices automate this measurement and provide continuous measurement,however, these devices require special catheters costing $195.00 per patient and require long averagingperiods (10 min.). Our system would achieve the same accuracy and real-time data using a standardcatheter costing $40.00 per patient. We have tested the technique in five animals and have found thesystem capable of making cardiac output measurements using only 2 ml injectate volumes. The smallvolume technique gave measurements with accuracy similar to those obtained using the standard of 10ml iced injectate over wide range of cardiac outputs. We plan to build a proto-type automated systemand test it in 10 animals. Less than 12% errors in animal tests using the automated system will beconsidered success.