Computer analysis of left ventricular cineangiograms

The heart is in continuous and rhythmic motion. For centuries this unique property has fascinated clinicians and scientists, although their attempts to get a better insight in the mechanisms which govern heart wall motion were seriously hampered by the almost complete absence of effective measurement tools. A dramatic change took place in the course of this century thanks to the technological revolution which changed cardiology more than any other of the biomedical sciences. The introduction of roentgen rays into medical practice enabled physicians for the first time to observe the beating heart in the intact human being. In cases of myocardial infarction, assessed by the new electrocardiographic technique, the fluoroscopy screen revealed abnormal motion of the epicardial contour. Numerous tools and methods were designed and developed to preclude the subjectivity and inaccuracy inherent with naked eye observation of heart wall motion. However, as only the epicardial contour could be determined, the information thus obtained remained limited. In 1929 Forssman introduced a catheter into his right heart. This act set the pace for the development of angiocardiography which after a hesitating start became one of the fastest growing areas in the medical field. As a consequence of the rapidly increasing number of cardiac catheterizations paralleled by an explosive growth of the amount of data obtained during each procedure, automated data processing became indispensible. In 1967 cardiologists of Stanford University in Palo Alto, California, described their flrst experiences with the use of a computing device in the cardiac catheterization procedure. This example was soon followed by cardiologists and technicians of the Thoraxcenter who developed a computer system, tailored to their catheterization laboratory, which became operative in 1973 and comprised facilities to process pressure signals on line and under control of the operator. To pursue the automation of data processing in the catheterization laboratory two more systems had to be developed: a system to detect automatically coronary artery contours and an analysis system for left ventriculograms. For this automated detection of the left ventricular endocardial contour from cineangiograms, a hard-wired system was designed and constructed. It was called the "Contouromat'. The aim of this thesis is to describe work to evaluate and validate this tool and its applicability in clinical practice and research. The first question which arises is whether an automated angie processing system has any benefit. Indeed in some clinics left ventriculograms continue to be judged in a purely qualitative way, although Herman et al's semiquantitative classification into norma-, hypo-, and dyskinesia is now applied in most centers. Since the angiogram contains more information than the human eye can detect by simple inspection, and the human observation often is inaccurate and inconsistent and strongly dependent upon the experience of the observer, there is much to be said for some form of standardization. The same applies to hand-drawn contours from end diastolic and end systolic cineframes which are currently employed to quantify the respective volumes. In conclusion, a weB designed angie processing system which produces consistent and reproducible quantitative data at a rate man cannot match, may help us to manage the diagnostic problems of our patients, whose interest - in spite of all emphasis on the technical tool we use - remains the primary aim of our efforts.