Ran Vas

Application of information theory to clinical diagnostic testing. The electrocardiographic stress test.

The inherent imperfection of clinical diagnostic tests introduces uncertainty into their interpretation. The magnitude of diagnostic uncertainty after any test result may be quantified by information theory. The information content of the electrocardiographic ST-segment response to exercise, relative to the diagnosis of angiographic coronary artery disease, was determined using literature-based pooled estimates of the true- and false-positive rates for various magnitudes of ST depression from < 0.5 mm to ⩾ 2.5 mm. This analysis allows three conclusions of clinical relevance. First, the diagnostic information content of exercise-induced ST-segment depression, interpreted by the standard 1.0-mm criterion, averages only 15% of that of coronary angiography. Second, there is a 41% increase in information content when the specific magnitude of STsegment depression is analyzed, as opposed to the single, categorical 1-mm criterion. Third, the information obtained from ECG stress testing is markedly influenced by the prevalence of disease in the population tested, being low in the asymptomatic and typical angina groups and substantially greater in groups with nonanginal chest pain and atypical angina.The quantitation of information has broad relevance to selection and use of diagnostic tests, because one can analyze objectively the value of different interpretation criteria, compare one test with another and evaluate the cost-effectiveness of both a single test and potential testing combinations.

Digital quantification eliminates intraobserver and interobserver variability in the evaluation of coronary artery stenosis.

A leading problem with subjective interpretation of coronary angiography is high intraobserver and interobserver variability. Four experienced angiographers independently determined percent diameter narrowing of 36 stenoses using 3 methods: by subjective analysis of single-frame cine film images (film), by subjective analysis of digitized nonenhanced single-frame images (digital), and by using a semiautomated digital caliper quantification system (Corona). The reproducibility of interpretations was assessed by comparison of estimated intraclass correlation coefficients. Digital and Corona readings correlated well with subjective interpretation of film (r greater than 0.85 for both). In contrast to Corona, the angiographers systematically overestimated the magnitude of stenoses in the intermediate (50 to 75%) range. Corona markedly improved intraobserver (p less than 0.005) and interobserver (p less than 0.001) reproducibility. Corona less frequently misclassified individual observations than did film when categories of less than 50%, 50 to 75% and more than 75% diameter stenosis were used (3.7% vs 31.5%, p less than 0.001). Our results suggest that digitization of a coronary angiogram in a 512 X 512 matrix has no significant adverse effects on the perception and quantification of stenosis by angiographers. Additionally, automatic measurement of coronary stenosis has 2 major advantages: It is accurate compared with a group of experienced angiographers and for the practical purpose of clinical decision-making, it eliminates intraobserver and interobserver variability.

Noninvasive diagnosis of coronary artery disease: the cardiokymographic stress test.

Stress–induced abnormalities of regional left ventricular wall motion were assessed by cardiokymography (CKG) during the course of maximal treadmill exercise tests in 157 patients, of whom 122 subsequently underwent coronary angiography. Seventy patients had significant angiographic coronary artery disease and 52 were normal. Forty–one of the 70 patients developed >0.1 mV ST–segment depression (ECG sensitivity 59%) and 52 of 70 patients developed abnormal systolic outward motion by CKG (CKG sensitivity 74%). Among the 52 normals, 36 had negative ECG stress tests (ECG specificity 69%) and 49 had normally sustained systolic inward motion by CKG (CKG specificity 94%). The stress CKG was normal in 15 of the 16 false–positive stress ECGs; the stress ECG was correctly normal in two of the three false–positive stress CKG tests. Only one normal patient had concordantly false–positive ECG and CKG tests. The predictive accuracy of concordant ECG and CKG interpretations was, therefore, higher than either test alone.These data suggest that regional wall motion abnormalities, which are sensitive and specific markers of myocardial ischemia, may be detected noninvasively by CKG. We concluded that CKG helps identify falsepositive and false–negative ECG stress tests and improves the diagnostic accuracy of stress testing for detection of coronary artery disease.

Computer enhancement of direct and venous-injected left ventricular contrast angiography

Following peripheral venous injection of radiopaque contrast material, a new on-line automatic computer image enhancement technique was employed to delineate and left ventricular (LV) endocardial silhouette in 10 dogs and 8 patients. This technique employs a very fast analog-to-digital conversion system capable of digitizing video frames on-line. By averaging into digital image memory the first 30 video frames and then subtracting each incoming frame from this memory, most of the background is eliminated, leaving only the contrast-filled ventricle. Since the technique employs conventional fluoroscopic exposure rates rather than cineangiography, there is marked reduction in x-ray exposure. An in vitro study using the Rando whole body phantom demonstrated that a 5 mm object with 2% contrast could be imaged within the complex chest anatomy with an incident exposure rate of only 30 mR/sec, using digital subtraction followed by contrast enhancement. In vivo studies were performed to assess the relative accuracy of ventricular border definition using this new technique by comparison to the unenhanced images in eight patients. The difference in planimetered area of the two cardiac silhouettes was 13 +/- 4 mm2 (mean difference +/- 3.4%). In four patients both direct and peripheral venous LV angiograms were obtained. There was a small (2% to 7%) systematic difference between calculated end-diastolic and end-systolic LV volume, with peripheral venous volumes invariably being smaller. Differences in calculated ejection fraction (EF) were of smaller magnitude; the maximum absolute difference in EF was 2%. We conclude that this technique is applicable to angiographic studies involving either cardiac or peripheral vascular injection of contrast material, and allows high quality images to be obtained at approximately seven-fold reduction in radiation dose (5 mA, 65 to 85 kv).

Fibrinolytic therapy of St. Jude valve thrombosis under guidance of digital cinefluoroscopy

Fibrinolytic therapy is an alternative to urgent reoperation for patients with St. Jude prosthetic valve thrombosis, but requires an accurate method for repeated assessment of prosthetic function. Since the St. Jude valve is not well visualized by conventional cinefluoroscopy, digital subtraction techniques were developed that improved visualization of the value and allowed assessment of leaflet separation and velocity. A 74 year old woman with prosthetic valve thrombosis 5 years after St. Jude aortic valve placement had an opening angle of 58 degrees (normal range 10 to 13; n = 8) with a maximal opening velocity of 1.37 degrees/ms (normal range 2.46 to 2.93). The closing angle was 125 degrees (normal range 120 to 127) with a maximal closing velocity of 1.38 degrees/ms (normal range 2.24 to 3.60). The patient received 250,000 U of streptokinase intravenously, then 100,000 U/h for 72 hours. Improvement in auscultatory findings occurred at 12 hours; repeat digital cinefluoroscopy showed an opening angle of 20 degrees with a maximal velocity of 2.77 degrees/ms, and a closing angle of 126 degrees with a maximal velocity of 1.91 degrees/ms. Digital cinefluoroscopy 4 weeks after discharge on warfarin and dipyridamole therapy was unchanged. There have been no thromboembolic complications after 6 months of follow-up. Thus, digital cinefluoroscopy is a new noninvasive technique that permits accurate measurement of normal and abnormal St. Jude leaflet function. Intravenous streptokinase dissolution of prosthetic valve thrombosis under digital cinefluoroscopic guidance may be an acceptable alternative to emergency reoperation. The frequency and significance of residual subclinical leaflet dysfunction after fibrinolytic therapy and the indications for elective reoperation require further evaluation.

Computer-enhanced digital angiography: Correlation of clinical assessment of left ventricular ejection fraction and regional wall motion

We compared computer-enhanced digital angiography (CEDA) following pulmonary injection of 20 ml Renografin-76 (5 ml/sec) to conventional directly injected left ventriculography (LV) in 13 patients undergoing routine diagnostic catheterization. Left ventricular ejection fraction (LVEF) was determined by planimetry from end-diastolic and end-systolic images by two independent angiographers. The correlation coefficient for LVEF (CEDA vs. LV) was r = 0.75 (p less than 0.005) for observer 1 and r = 0.85 (p less than 0.0005) for observer 2. The interobserver variability for LVEF was very low, resulting in a high correlation coefficient (r = 0.91, p less than 0.0005). Three angiographers independently reviewed both the conventional and CEDA images in a random order for assessment of anterior, apical, and inferior regional wall motion, using a 6-point subjective grading system (198 determinations). The interobserver correlation for subjective assessment of regional wall motion by both LV and CEDA was poor (49% for LV and 59% for CEDA, p = NS). These poor correlations were not improved by excluding any region or angiographer from the analysis. The agreement of regional motion assessments between the two techniques was only 40%. To improve reproducibility of wall motion interpretation, an automated analysis program was developed. First the range of normal contraction was defined from pooled literature data. The movement of any segment of the left ventricular wall could then be determined in millimeters and referenced to the normal range. This method eliminated interobserver variability. In the absence of an acceptable standard of segmental wall motion to which this measurement can be compared, the accuracy of this objective format could not be determined. We conclude that CEDA images allow accurate determination of ejection fraction and that the large interobserver variability of subjective regional wall motion analysis can be overcome by employing more objective formats.

Analysis of regional ischemic left ventricular dysfunction by quantitative cineangiography.

The ability of left ventricular angiography to detect regional ischemic dysfunction was assessed in 10 closed-chest dogs during the course of acute balloon occlusion of the anterior descending coronary artery. During the 2-minute period of occlusion, serial cineangiography revealed a sequence of wall motion abnormalities over the anteroapical region almost identical to that observed using directly implanted gauges. This sequence consisted of progressive reduction in regional systolic shortening with eventual replacement by systolic expansion. These changes preceded both electrocardiographic ST-segment and hemodynamic alterations, and were readily observed by gross subjective inspection of the cineangiograms, but with an intraobserver variability of 22%. Frame-by-frame motional analysis of the ventricular perimeter relative to its centroid of mass allowed more precise characterization of regional dysfunction. These data are consistent with previous studies demonstrating that regional wall motion abnormalities are both sensitive and specific markers of acute ischemia, and support the use of computerized left ventricular angiography for the quantitative assessment of clinical ischemic dysfunction.

The influence of bias on the subjective interpretation of cardiac angiograms

Subjective interpretation of angiographic left ventricular regional wall motion is routinely performed with knowledge of the location and extent of coronary artery stenosis. We studied 100 patients with coronary artery disease in order to determine the accuracy of such wall motion assessment relative to a more objective standard based upon computer-assisted left ventricular (LV) ejection fraction and end-systolic fractional shortening referenced to the end-diastolic area centroid. Only 379 of 700 (54%) region-by-region comparisons of wall motion were in precise agreement. Computer-assisted wall motion analysis correlated significantly better with ejection fraction than did subjective analysis (r = 0.82 vs r = 0.61, p less than 0.002). In 56 patients, in whom major discordance was noted, subjective assessment of wall motion correlated significantly better with the presence of coronary artery stenosis (p less than 0.05), but objective assessment correlated significantly better with ejection fraction in these same patients (p less than 0.02). These data suggest that the accuracy of subjective assessment of regional wall motion, relative to global ejection fraction, can be adversely biased by knowledge of the patients coronary anatomy. Because of the inherently reproducible nature of the algorithmic process, and in light of the better correlation with global function, computer-assisted analysis of regional wall motion might be preferable to conventional subjective assessment.

Assessment of the functional significance of coronary artery disease with atrial pacing and cardiokymography

Abstract Thirty-six patients were studied during the course of cardiac catheterization to assess the role of cardiokymography and atrial pacing in the functional evaluation of angiographic coronary arterial stenosis. Only 4 of 25 patients with greater than 50 percent diameter stenosis of at least one major vessel had 0.1 mv or greater S-T segment depression at a paced heart rate of 123 ± 25/min, and 2 of 11 normal patients revealed a similar response (P = not significant). In contrast, in 22 of 25 patients systolic outward motion developed as determined with cardiokymography during the same pacing period, whereas in only 1 of 11 normal patients a similar abnormality did develop (P These data are consistent with the view that regional wall motion abnormalities are highly sensitive and specific markers of ischemia and that such abnormalities may be detected noninvasively with cardiokymography. It is concluded that atrial pacing in conjunction with cardiokymography is applicable to the functional assessment of ischemic heart disease and may provide a means for objective evaluation of the significance of angiographically observed coronary stenosis.

Cardiokymographic wall motion pattern in patients with acute myocardial infarction

We studied 50 patients with acute myocardial infarction by cardiokymography to record anterior left ventricular wall motion. Systolic outward motion was observed in 46 of 50 patients (92%). Holosystolic outward motion was characteristic of acute anterior infarction (93%) and acute subendocardial infarction (89%) but was less common in acute inferior infarction (29%). Partial systolic outward motion was seen in 8 of the 14 patients with acute inferior infarction, but only in 2 of 36 patients with acute anterior and subendocardial infarction. Cardiokymographic abnormalities were seen in more precordial locations in acute anterior (80%) and subendocardial infarction (97%) than in inferior infarction (74%). Holosystolic outward motion was seen in 75% of all locations in acute anterior and subendocardial infarction and only in 23% in acute inferior infarction. Thus both the extent and the severity of abnormal systolic outward motion clearly separated inferior infarction from anterior and subendocardial infarction. Dynamic changes in wall motion contraction patterns were observed during the course of acute myocardial infarction; both improvement and deterioration were observed. Cardiokymography is a simple, non-invasive method to assess changes in left ventricular segmental wall motion in patients with acute myocardial infarction.