Computer Reads Heart Condition From X-ray Images

During a cardiac catheterisation procedure, X-ray film images are taken of the coronary arteries providing the heart muscle with blood, and of the left heart chamber, which is the major pump of the heart providing the blood supply to the entire body. The condition of the heart can be determined by studying the wall motion of the ventricle from these images. This is usually done by outlining the contours of the heart chamber at the phase of maximal filling of the heart chamber with blood and at the phase of minimal filling, i.e. at two phases during a cardiac cycle. From these data the overall function of the heart can be determined, as well as the regional wall motion.

However, drawing the heart contours is a time-consuming process, which is also hampered by large variabilities between observers. During the research carried out in this project, a statistical model was developed, derived from seventy case studies. This statistical model was subsequently translated into a software application.

After a slight interaction from the user, being the manual definition of the aorta position and the lowest point of the heart, the computer automatically determines the contours in these two phases of the cardiac cycle. Subsequently, the regional wall motion and the global function of the heart are derived. These data are of great importance for the proper diagnosis by the physician and is also used by the surgeons in cases of bypass operations.

The results of this research have been incorporated into a commercial software application developed by Medis. The company participated in the study right from the start as a result of which the knowledge could be applied and developed at an early stage. The result is the recently demonstrated measuring software.

The research was monitored and partly funded by Technology Foundation STW.