KLINGER VS CENTERLINE
Category
A comparison between two methods of left ventricle wall motion analysis
OBJECTIVE ANALYSIS OF LEFT VENTRICULAR WALL MOTION
KLINGER VERSUS CENTERLINE
G. Bozzi, F. Casolo, F. Regalia, P.G. Duca, S. Porru, S. Todd
Klinger is a method of calculation of wall motion developed by the author at the end of the 80s after two years of study of the problem, using a little research or point method known in the medical field, but much used in other contexts ( for example by physicists). This method is based on the consideration that a complex problem, for which it is not found a unique solution, can be divided into a variable number of parts, each of which is subjected to logical-mathematical tests that produce each time a result assessable by an expert on how true/false (or best/ worst). Objective analysis of left ventricular wall motion seems one of these problems. As I explained in a previous article, the efforts of many authors in the 70s and 80s of last century to solve this problem have not produced satisfactory results and led me to adopt the research method I mentioned above. In about a year's work, I read more than 1500 coronary angiograms and left ventriculography in the Ospedale di Circolo di Varese archive ( note 1), looking for normal cases. After finding 12 normal cases, the standard deviations (SD) of the calculated points have stopped falling, so I stopped the search for normal cases, the number of 14. At the same time, I produced the algorithms for the execution of computer tests necessary for the study.
The following is a list of the main tests carried out at the time.
1.0 Percent reduction of silhouette components
1.1 radial segments starting from a supposed center of contraction
1.2 segments orthogonal to the longitudinal axis
1.3 as 1.2 times a coefficient computed on long axis shortening
*** 1.4 slices obtained dividing long axis by 10 equidistant segments
1.5 volumes applying Simpson's rule to 1.4
2.0 Long axis proximal end position
2.1 fixed point
2.1.1 mid-aortic point
2.1.2 boundary between mitral and aortic valves
2.2 changeable point
2.2.1 changeable case by case, fixed in single case
*** 2.2.2 always changeable, looking for symmetry of semi-silhouettes
3.0 Subdivision of long axis shortening among slices
*** 3.1 all slices (1-20)
3.2 apical slices (9, 10, 11, 12)
3.3 basal slices (1, 2, 19, 20)
4.0 Wall apposition at the apex
4.1 fixed 10% of long axis
*** 4.2 equalizations to aortic plane lowering
5.0 Exclusion of valve edges
5.1 automatic identification of mitral fornix
5.2 operator identifies mitral fornix
5.3 operator selects between mitral fornix and efflux tract
5.4 operator selects the point on both silhouettes
*** 5.5 operator selects a point on ED silhouette and algorithm identifies an equivalent point on ES silhouette
Table I. Algorithmic tests used to identify calculations to be used for objective evaluation of wall motion.
The Klinger method constructs a left ventricular contraction curve by calculating the percentage reduction of 20 areas orthogonal to the longitudinal axis of the ventricular cavity, obtained by end-diastolic silhouette ( ED ) and end-systolic ( ES ) filmed in the right anterior oblique projection ( RAO). The solution to the problem appeared when it was assumed that the angiographic shortening of this axis is due to the shortening of the real ventricular cavity added to a stretch of affixing of the walls at the apex. Since the cardiac action takes place in a clear condition of dynamic equilibrium, we have tested the hypothesis that the actual shortening of the cavity is deduced by the lowering of the aortic floor, as obtained by ventriculography in the RAO projection. The introduction of this data in the calculations resulted in 0 the standard deviation of the last longitudinal axis apical sector corresponding to the areas 10 & 11: in other words, in normal cases, the apex walls are always in contact with each other, making virtual a certain portion of the cavity ( see the fourth film of the series presented in this site).
The centerline method proposed with great fanfare by cardiologists at the University of Washington, Seattle, analyzes the motion of the heart walls using the two silhouettes ED and ES, identifies a center line between the two margins and track 100 equidistant chords, perpendicular to this centerline. The values of these chords are normalized by calculating a fractional shortening to cancel the differences due to the different size of images.
The centerline method was chosen for comparison with our because it appears in textbooks, in Geigy scientific tables and is implemented on General Electric ( GE ) angiographic system at our disposal in Radiology at the Hospital St. Carlo Borromeo directed by Prof. Filippo Casolo.
MATERIALS AND METHODS. From a series of 70 consecutive cine-angiographic studies performed in patients with coronary artery disease, we selected 50 cases whose left ventricle angiography (LVA) were considered to be of good quality. Ventricular silhouettes were directly subjected to computer analysis using the centerline method which is implemented on GE angiographic unit of San Carlo Hospital, Milan; corresponding laser printed hard copies were read through a graphic table (Calcomp Drawing Board III) connected to a PC (IBM 480) for calculation by the Klinger method. Cine-angiograms were separately examined by three experienced angiographers who didn't know the results of the two objective evaluations. According to CASS suggestions, five ventricular segments of the RAO profile were separately considered, and classified as normal, hypo-kinetic (hypo-kinesis being graded as moderate or severe), a-kinetic or dyskinetic. Each case was debated in successive meetings to reach a consensus diagnosis.
WALL SECTORS KLINGER SLICES CENTERLINE CHORDS
anterobasal 1 - 4 1 - 20
anterolateral 5 - 8 21 - 40
apical 9 - 12 41 - 60
diaphragmatic 11 - 16 61 - 80
posterobasal 17 - 20 81 - 100
Table II. Correspondences of slices of Klinger method and chords of centerline method and sectors of subdivision of cardiac silhouette (CASS).
In the centerline method, on each side of the curve, areas encompassed by values of +/- 1SD, 2SD, 3SD can be defined. The values between -2SD and -3SD define hypo-kinesis (without distinction between mild, moderate or severe), whereas those below -3SD correspond both akinesis and dyskinesis.
In the Klinger method, the area between (normal value - 2SD) and 10% defines hypo-kinesis (easily graded as mild, moderate or severe); the area between 10%(note 2) and 0 defines a-kinesis; negative values indicate dyskinesis.
STATISTICAL ANALYSIS. Having agreed upon diagnosis as a gold standard, the diagnostic results of the Klinger and the centerline methods were comparable by means of the following calculations: 1) Overall percentage of agreement with agreed upon diagnosis, on a segment by segment basis. The 95% C.I. has been computed by the Gaussian approximation to the binomial distribution. 2) Performing K statistics and its 95% C.I., to exclude chance expected agreement. 3) Sensitivity and specificity; the 95% C.I. were computed by the Gaussian approximation to binomial distribution (tab.III).
STATISTICS Klinger centerline
Average percentage of concordance 86,0 (95% CI 81,8 to 90,3) 57,6 (95% CI 51,1 to 63,7)
Cohen's K 0,76 (95% CI 0,67 to 9,85) 0,28 (95% CI 0,19 to 0,37)
sensitivity 93,7 (95% CI 89,5 to 97,9) 71,6 (95% CI 62,8 to79,4)
specificity 82,9 /95% CI 76,3 to 89,5) 60,2 (95% CI 51,6 to 68,8)
Table III. Statistical data.
All values in Table III indicate a statistically significant difference between the two methods: effective Klinger method, ineffective centerline method. In particular, the coefficient K of Cohen indicates a good correlation for Klinger (value between 0.6 and 0.8) and a poor correlation to the centerline (value between 0 and 0.4). The difference in diagnostic skills, presented in the example of Figure 1 and 2 is so obvious as not to require additional explanations.
Fig. 1. Coronary angiography of this patient showed a 70% stenosis of the proximal left anterior descending artery and the second stenosis about 50% just before the 2nd diagonal branch. A 70% stenosis was also present in the right posterior descending artery. Klinger contraction curve shows a well-defined hypo-kinesis involving anterolateral and apical sectors. Centerline contraction curve does not reach the level for hypo-kinesis of the anterolateral wall and shows normal apical contraction, whereas perceives an unexisting hypo-kinesis of the posterobasl chords.
Fig. 2. Coronary angiography of this patient showed a 70% stenosis before the crux and a 70% stenosis of mid anterior descending artery. The patient had had an inferior myocardial infarction. Klinger contraction curve shows a very light anterolateral and a mild apical hypo-kinesis, a more evident contraction compromise involving diaphragmatic and posterobasl sectors. Centerline contraction curve instead of hypokinesis, shows a hyperkinetic anterolateral wall, is unable to perceive apical hypo-kinesis and with difficulty reach the limit for hypo-kinesis beyond 85th chord.
DISCUSSION
The different diagnostic capacity of the two methods does not depend on the calculations, but from having considered that the heart during systole performs a complex movement due to the more components add up: a displacement of the ventricle walls to the inside for muscle contraction, a lowering of the heart caused by the third principle of dynamics and a partial twisting of the heart on itself caused by the arrangement of the muscle fibers of the myocardium. The traditional angiography shows a two-dimensional drawing of the sum of these movements, so contraction study should take account of this reality, must take advantage of the available data, and limit the damage of the loss of a dimension.
In our method was decisively calculating the shortening real cavity and apical obliteration tract, free positioning of the longitudinal axis proximal end which compensates in part the twisting movement of the heart, and the elimination from the calculations of the mitral valve margins. These factors are ignored in the centerline method ( but also in all other methods proposed up to now). The end result is the method of Klinger ability to produce diagnosis coincident with agreed upon subjective diagnosis, while no other method is able to achieve this result.
CLOSING REMARKS OF the FIRST AUTHOR
This work, in the most detailed and comprehensive form than that presented here, was sent to Circulation in the late ' 90s. The work was refused and I'm displeased not to be able to enclose the response of referees, probably over in a waste bin, but I remember very surprising: " ... the authors (i.e. I, prof. Casolo and dr. Regalia) have not understood anything about the centerline method ...... and have used a method not original, definitely not right". I leave it to the reader 's judgment on a similar response.
Years later, I am struck by the realization that my curves have been ignored by manufacturers of angiographic systems, which for years implement on their systems unnecessary calculation methods because they are incapable of formulating a diagnosis. But here the question probably complicated by an economic component (expensive upgrade programs of thousands of angiographic systems) and image ( justify an oversight rather serious from a scientific point of view ): In conclusion, better to pretend nothing has happened. In support of this hypothesis, the referees as well as the response of Circulation, I remember that an executive of GE Italy (which years ago I proposed the problem), told me that GE's foreign branches could sell equipment, not to suggest ideas to the parent. Furthermore, I remember that Italian officials who had sold a new Siemens angiography system in a hospital in northern Italy, they came to me and bought the hemodynamic program, requested by the doctors of that hospital. Then at least two important angiographic systems factories were aware of my contraction curves and have ignored. Now I expect that, sooner or later, someone, " rediscover " the problem. This is the reason for the which I opened my own website and published some articles: to assert the paternity of a left ventricle contraction curve, the only able to process diagnosis comparable to those of expert angiographers.
Recent Posts
The discovery of a withholding lost mail ignites a controversy
18/05/2016
An angiographic phenomenon not well explained up to now.
27/08/2016
The heart is a very efficient hydrodynamic pump
11/03/2016