Deep Learning-Based Image Registration in Dynamic Myocardial Perfusion CT Imaging.

Registration of dynamic CT image sequences is a crucial preprocessing step for clinical evaluation of multiple physiological determinants in the heart such as global and regional myocardial perfusion. In this work, we present a deformable deep learning-based image registration method for quantitative myocardial perfusion CT examinations, which in contrast to previous approaches, takes into account some unique challenges such as low image quality with less accurate anatomical landmarks, dynamic changes of contrast agent concentration in the heart chambers and tissue, and misalignment caused by cardiac stress, respiration, and patient motion. The introduced method uses a recursive cascade network with a ventricle segmentation module, and a novel loss function that accounts for local contrast changes over time. It was trained and validated on a dataset of n = 118 patients with known or suspected coronary artery disease and/or aortic valve insufficiency. Our results demonstrate that the proposed method is capable of registering dynamic cardiac perfusion sequences by reducing local tissue displacements of the left ventricle (LV), whereas contrast changes do not affect the registration and image quality, in particular the absolute CT (HU) values of the entire CT sequence. In addition, the deep learning-based approach presented reveals a short processing time of a few seconds compared to conventional image registration methods, demonstrating its application potential for quantitative CT myocardial perfusion measurements in daily clinical routine.

Can clinical CT data improve forensic reconstruction?

In accidents resulting in severe injuries, a clinical forensic examination is generally abandoned in the initial phase due to high-priority clinical needs. However, in many cases, data from clinical computed tomography (CT) examinations are available. The goals of this prospective study were (a) to evaluate clinical CT data as a basis for forensic reconstruction of the sequence of events, (b) to assess if forensic radiological follow-up reading improves the forensic diagnostic benefit compared to the written clinical radiological reports, and (c) to evaluate if full data storage including additional reconstructed 0.6-mm slices enhances forensic analysis. Clinical CT data of 15 living individuals with imaging of at least the head, thorax, and abdomen following polytrauma were examined regarding the forensic evaluation of the sequence of events. Additionally, 0.6-mm slices and 3D images were reconstructed for forensic purposes and used for the evaluation. At the forensic radiological readings, additional traumatic findings were observed in ten of the 15 patients. The main weakness of the clinical reports was that they were not detailed enough, particularly regarding the localization of injuries and description of wound morphology. In seven cases, however, forensic conclusions were possible on the basis of the written clinical reports, whereas in five cases forensic reconstruction required specific follow-up reading. The additional 0.6-mm slices were easily available and with improved 3D image quality and forensic diagnostics. In conclusion, the use of clinical CT data can considerably support forensic expertise regarding reconstruction issues. Forensic follow-up reading as well as the use of additional thin slices for 3D analysis can further improve its benefit for forensic reconstruction purposes.

Computer tomography aided 3D analysis of the distal dorsal radius surface and the effects on volar plate osteosynthesis.

The aims of this study were to measure the size of Lister's Tubercle, the extent of the extensor pollicis longus (EPL) groove and the dihedral angle of the distal dorsal radius. Computer tomography scans of 30 forearms were performed by using a 64-slice Siemens SOMATOM Sensation CT system (Resolution 0.6 mm). DICOM raw data were calculated to 3D by MIMICS software (Materialise, Leuven, Belgium). The size of Lister's Tubercle varied from 1.4 to 6.6 mm (average 3.3 mm) in height radial to the tubercle, and from 5.6 to 18.6 mm (average 13.2 mm) in length. The depth of the EPL groove varied from 0.6 to 3.2 mm (average 1.6 mm). The height on the ulnar side, between the depth of the groove and the tip of the tubercle, varied from 2.2 to 5.8 mm (average 3.4 mm). The dihedral angle of the distal dorsal radius varied from 110 degrees to 135 degrees (average 123 degrees). The variations in height of Lister's Tubercle and in depth of the EPL groove are considerable. This needs to be taken into account when performing volar plating of distal radius fractures otherwise screws may inadvertently penetrate the dorsal cortex of the radius potentially leading to EPL rupture.

Coronary artery plaque burden and perioperative cardiac risk.

BACKGROUND: Electron-beam computed tomography-derived coronary calcium score correlates with the morphologic severity of coronary artery disease, reflecting both global atherosclerotic plaque formation and coronary artery luminal narrowing. The current study examines the impact of coronary atherosclerotic plaque burden, measured by coronary calcium score, on the potential for perioperative myocardial cell injury, as assessed by cardiac troponin T elevations in patients undergoing elective vascular surgery. The authors further investigated whether perioperative myocardial cell injury in those patients adversely affects noninvasive measures of left ventricular systolic function, such as ejection fraction and wall motion score. METHODS: Fifty-one consecutive patients scheduled for vascular surgery were enrolled in this prospective study. In addition to standard preoperative evaluation, including patient history and physical examination, electron-beam computed tomography scan, 12-lead electrocardiography, and transthoracic echocardiography were performed on the day before surgery. Subsequent evaluations on postoperative days 2 and 7 included transthoracic echocardiography and 12-lead electrocardiography. Cardiac troponin T determinations were performed on the day before surgery, immediately preoperatively, and on postoperative days 1, 2, 3, and 7. RESULTS: The median coronary calcium score of the 51 patients was 997.0 (25th percentile, 202.5; 75th percentile, 1,949.5). Cardiac troponin T elevations exclusively occurred in patients with a coronary calcium score greater than 1,000. The six patients (12%) with perioperative cardiac troponin T elevations had a 2.5-fold higher coronary calcium score than those without cardiac troponin T elevation (P = 0.021). In these patients, the ejection fraction decreased from 61 +/- 10% to 52 +/- 13% (mean +/- SD) on postoperative day 2 and was 54 +/- 16% on postoperative day 7 (P = 0.022). CONCLUSION: A high electron-beam computed tomography coronary calcium score, reflecting substantial coronary plaque burden, carries an increased risk for myocardial cell injury after vascular surgery. In these patients, myocardial damage may result in deterioration of global systolic left ventricular function.

Left ventricular architecture after valve replacement due to critical aortic stenosis: an approach to dis-/qualify the myth of valve prosthesis-patient mismatch?

OBJECTIVES: Left ventricular hypertrophy in patients with critical aortic stenosis (AS) is an adaptive process that compensates for high intracavitary pressure and reduces systolic wall stress followed by an increase in myocardial masses. In the present prospective clinical trial, we investigated long-term compensatory changes in left ventricular geometry and function after aortic valve replacement using mechanical bileaflet prostheses with the main emphasis on the small-sized aortic annulus and valve prosthesis-patient mismatch. METHODS: A total of 58 patients with critical AS were assigned to the following groups according to the predictive value of prosthetic valve area index (VAI): group EXMIS: 29 patients (VAI < or =0.99), expected mismatch; group NOMIS: 29 patients (VAI < or =0.99), no mismatch. At controls T(0) (before operation/operation (OP), T(1) and T(2) (4 and 20 months after OP) the left ventricular geometry was recorded by means of Imatron electron beam tomography and the transprosthetic velocities were measured by echocardiography. RESULTS: Statistical analysis showed a consistent reduction in the absolute (P=0.04) and indexed (P=0.04) left ventricular myocardial mass for both cohorts; furthermore, there was a significant difference between EXMIS and NOMIS patients concerning the factors, time and mass reduction (P=0.005), because of distinct baselines. A logistic regression report revealed preoperative cardiac output, absolute left ventricular myocardial mass, perfusion, body surface area and the native valve orifice area as predicting coefficients and factors for a minimum mass reduction of 25%. We explain a mathematical formula that turned out to be the most sensitive for correctly classified factors. CONCLUSIONS: The left ventricular geometry and transprosthetic velocities resulted in the same postoperative recovery for both EXMIS and NOMIS patients. The presented data showed that valve prosthesis-patient mismatch had no influence in several stepwise logistic regression models. We conclude that modern mechanical bileaflet prostheses allow both acceptable hemodynamics and recovery of left ventricular hypertrophy, even in small aortic annuli.

Dimeric versus monomeric nonionic contrast agents in visualization of coronary arteries.

A cubital intravenous iodine contrast agent enhancement is used to visualize coronary arteries using EBT. The quality of the coronary artery visualization however is limited by the nearly simultaneous approximation of CT values in coronary arteries and myocardial tissue. The objective of the study was to evaluate if "under real clinical circumstances" the lower iodine concentration and the dimeric based characteristic of iodixanol may effect the kinetic of the applied contrast agent and the visualization of coronary arteries studied noninvasively by EBT. A double-blind, randomized, parallel study was performed in 111 cardiac patients, using iodixanol 270 mg I/ml or iohexol 300 mg I/ml. The kinetics of contrast enhancement was studied in the flow mode measuring following parameters: mean arrival time and mean time to reach peak CT values in the pulmonary trunk, transit time from the pulmonary trunk to the aorta as well as mean and maximum CT values in the left ventricular chamber and in the myocardium with respect to the body mass index. The mean difference of CT values in the left ventricular chamber and the myocardium was calculated. The length of the visualized coronary arteries was assessed and the diagnostic quality of coronary artery visualization scored on a visual analogue scale. Although iodixanol was used with a lower iodine concentration than iohexol there was no significant statistical difference between both groups with respect to the diagnostic visualization and length assessment of the coronary arteries as well as in the mean difference of CT values in the left ventricular chamber and the myocardium. This means that the advantageous dimeric characteristics of iodixanol may be used to reduce the amount of applicated iodine in contrast agents without loss of diagnostic image quality and information.

Imaging the coronary venous drainage system using electron-beam CT.

This study describes the appearance of the coronary sinus and its tributary veins as visualized on ECG-triggered electron-beam computed tomography (CT) and investigates their spatial relationship to other cardiac structures. Thirty-two patients were examined with ECG-triggered electron-beam CT (exposure time: 100 ms, slice thickness: 1.5 mm) after intravenous contrast agent administration. The entire heart was imaged; the appearance of the coronary sinus and its tributary veins were evaluated. In all 32 patients the anterior interventricular vein and the posterior interventricular vein drained into the coronary sinus. The small cardiac vein was visualized in five patients, a posterior vein of the left ventricle in three and the left marginal vein in eleven. The coronary sinus of all 32 patients had a average length of 30 mm +/- 10 mm (mean +/- SD), range: 21-40 mm and a diameter of 9 mm +/- 5 mm (mean +/- SD), range: 4-14 mm. The results of our work show that if the entire heart volume is scanned using ECG-triggered electron-beam CT, the delineation and the differentiation of the major cardiac veins is possible on transverse cross sections in a way which corresponds to the anatomical literature. Hence to the similar enhancement and similar diameter of coronary veins and arteries on contrast-enhanced electron-beam CT studies, the radiologist should be familiar with the cross-sectional anatomy of the major cardiac veins to prevent possible misinterpretation.

Noninvasive measurement of pulmonary vascular resistances by assessment of cardiac output and pulmonary transit time.

RATIONALE AND OBJECTIVES: Pulmonary vascular resistance is of special interest in many diseases. Usually it is determined invasively by catheterization, but cardiac output and pulmonary transit time can be ascertained by several noninvasive methods. METHODS: Fourteen heart recipients (age 34-71 years) were examined by electron-beam CT of the heart. Cine and flow studies were performed using a total of 60 mL of contrast and a breath-hold of 20 seconds. RESULTS: A mathematical model for calculating pulmonary vascular resistances from noninvasively measured cardiac outputs and pulmonary transit times was developed. Right-sided heart catheterization served as the reference method. CONCLUSIONS: The formula created seems to allow a clinically valid estimate of pulmonary vascular resistance from noninvasively acquired data.

Application of pharmacokinetics to electron-beam tomography of the abdomen.

RATIONALE AND OBJECTIVES: The purpose of this study was to evaluate the pharmacokinetics of abdominal time-attenuation curves obtained at electron-beam tomography. MATERIALS AND METHODS: Computed tomographic enhancement data of the aorta, portal vein, vena cava, liver, spleen, and pancreas were obtained in 25 patients after injection of 50 mL of contrast medium. These data were used to calculate pharmacokinetic parameters such as half-lives, mean residence times, and areas under the curve with a computer program. RESULTS: Maximal enhancement was observed in the aorta 24 seconds +/- 5 (mean +/- standard deviation) after starting the injection of contrast medium (178 HU +/- 56), in the portal vein after 42 seconds +/- 14 (60 HU +/- 17), in the vena cava after 35 seconds +/- 7 (66 HU +/- 23), in the liver after 58 seconds +/- 15 (24 HU +/- 6), in the spleen after 35 seconds +/- 12 (42 HU +/- 16), and in the pancreas after 39 seconds +/- 15 (42 HU +/- 10). Half-lives of the last phase observed were 108 seconds +/- 123 in the aorta, 33 seconds +/- 30 in the portal vein, 49 seconds +/- 40 in the vena cava, 50 seconds +/- 54 in the liver, 62 seconds +/- 33 in the spleen, and 22 seconds +/- 27 in the pancreas. The computer program allowed for excellent fitting curves to the measured attenuation values and for subsequent calculation of pharmacokinetic parameters. New dosage regimens also could be simulated successfully. CONCLUSION: The pharmacokinetic parameters evaluated might be useful in the optimization of dosing and scanning parameters of the abdomen for ultrafast and helical CT.

[Electron beam computerized tomography (EBCT) of the heart]. / Electron-beam-Computertomographie (EBCT) des Herzens.

Using electron-beam computed tomography (EBCT) with short exposure times of 100 or 50 ms and the capability of acquiring up to 2 x 17 images/s it is possible to study most of the important morphological and functional determinants of the heart. Various examples of studies in acute and chronic cardiac diseases are shown to demonstrate the use of EBCT to determine quantitatively left ventricular volumes (ml), myocardial mass (g), wall thickness changes over the cardiac cycle (mm/s), myocardial perfusion (ml/ 100 g/min) and the extent of coronary calcification (calcium score) and qualitatively the state of the proximal 4-6 cm of the subepicardial coronary arteries. The knowledge of these determinants seems very useful in excluding cardiac dysfunction, in the early recognition of cardiac disease and in the evaluation of the haemodynamic severity of coronary artery stenotic lesions. Further interdisciplinary studies are necessary to assess the clinical validity of these cardiac determinants, especially myocardial perfusion, using this advanced CT technology.

[Quantitative determination of left ventricular myocardial perfusion with electron beam computerized tomography]. / Quantitative Bestimmung der linksventrikulären Myokardperfusion mittels EBCT.

Myocardial perfusion is one of the most important functional parameters of the heart. Presently various indirect methods are used to determine coronary blood flow or myocardial perfusion as inertgas-, thermodilution-, Doppler catheter- and radiopharmacological techniques. Electron-beam-computed-tomographical technology is able to perform CT data acquisition with a very short exposure time of 50 ms. Using this method it is not only possible to determine left ventricular volumes but also to measure myocardial perfusion in ml/100 g/min. The measurement of the left myocardial perfusion is performed using the short axis view. This position is obtained by moving the table 25 degrees to the patient's right and 15 degrees caudally. To determine the position of the left ventricle, a localization scan is obtained in multi-slice-mode using all for target-rings, thus obtaining 8 tomographic levels over 68 mm (each tomographic level having a slice thickness of 7 mm, with an interslice gap of 4 mm between each two adjacent tomographic levels). In this short axis position, using the multi slice flow mode with 3 target-rings and after administration of 50 ml of contrast medium intravenously with a flow of 3 ml/s, 6 tomographic levels are imaged. Each tomographic level is obtained 13 times at 80% of the R-R-interval at each 2 or 3 heart beat (ECG-gated). The left ventricular myocardial contrast enhancement is measured by drawing manually the outline of the left ventricular myocardium using time-density-software of the Imatron workstation. For calculation of the myocardial perfusion the so-called "slope method" is used and the results are expressed as the maximum slope of enhancement of the myocardium divided by the difference of the precontrast and peak CT-value in the left ventricle. The global myocardial perfusion is calculated as a mean of all evaluated tomographic levels. In this study left ventricular volumes as enddiastolic volume endsystolic volume and stroke volume were measured and ejection fraction and cardiac output calculated. The measurements were performed in the log axis view. This view is obtained by moving the table 15 degrees to the patients left in a horizontal position. In this long axis position 6 tomographic levels are imaged using the multi-slice-cine-mode with 3 target-rings after administration of 50 ml of contrast medium intravenously with a flow of 3 ml/s. Each tomographic level is obtained 13 times starting at 0% of the R-R-interval (ECG-triggering). The exposure time is 50 ms with an interscan time delay of 8 ms. In 9 studied patients of whom one had 3 significant coronary artery stenotic lesions (> 50%), 2 patients had each 2 non significant stenotic lesions (< 50%) and 6 revealed nearly normal coronary angiograms. The mean global myocardial perfusion was 70 ml/100 g/min (min.32 and max. 116 ml/100 g/min). This mean value of 70 ml/100 g/min is reflecting 5% of the cardiac output supposing that the mean heart weight of these patients was 300 g. In this study the mean of the left ventricular muscle mass determined by the use of EBCT was 130 g. A comparative evaluation of coronary angiographic findings in these patients with the measured myocardial perfusion values revealed, that is not sufficient to look only at the absolute values of the measured myocardial perfusion. Furthermore it seems to be necessary to interpret these perfusion values with respect to the calculated cardiac output. Additional studies of well defined patients groups are necessary to determine normal values of myocardial perfusion at rest in patients with and without coronary artery disease. This seems to be important as comparative analysis of myocardial scintigraphic and EBCT-studies is difficult because of methodical inherent differences. The results of this study suggest that despite the presence of some beam hardening artifacts it is possible to measure myocardial perfusion using EBCT in patients with suspected coronary artery disease in the

[Functional studies of swallowing with electron beam computerized tomography]. / Funktionelle Schluckaktuntersuchungen mit der Elektronenstrahl-computertomographie.

Functional evaluation of swallowing disorders requires rapid imaging modalities. Videofluorography and cinematography are the gold standard, but they have their limitations: no transverse plane imaging is achieved and structural resolution for exact topographic analysis is limited. Three cases preselected by videofluorography were studied to evaluate whether electron beam tomography (EBT) permits more detailed dynamic imaging of swallowing disorders focusing on the mesonasopharyngeal segment, the hypopharynx and the upper esophageal sphincter (UES). Immediately after videofluorographic examination of the oropharyngeal deglutition, EBT is performed. The patient is in a supine position and while the patient swallows a 20 ml bolus of water or diluted iodine containing contrast agent, a sequence of 20 images per level is scanned. The levels, which are determined by using the scout view, are oriented parallel to the hard palate either at the level of the hard palate to image the mesonasopharyngeal segment or just above the hyoid bone to focus on the hypopharynx or at the location of the UES. The scan technique is a single-slice cinemode with a slice thickness of 3 mm (exposure time 100 ms, interscan delay 16 ms, 130 kV, 620 mA). The following structural interactions that we have so far been unable to image can be clearly demonstrated with EBT: (1) during normal swallowing, the mesonasopharyngeal segment is completely and symmetrically closed by the soft palate and Passavant's cushion; (2) lateral hypopharyngeal pouches can be located more precisely; and (3) disorders of the UES can be differentiated into functional or morphologically caused disorders (e.g., goiter or cervical osetophytes). Videofluorography and cinematography are still the gold standard in functional evaluation of swallowing disorders. However, EBT permits dynamic imaging of pharyngeal deglutition in a preselected transverse plane and can give useful additional information concerning functional anatomical changes in the pharynx during swallowing. Further clinical evaluation is needed.

Spirometrically controlled quantitative CT for assessing diffuse parenchymal lung disease.

OBJECTIVE: Assessment of lung attenuation by CT reflects changes in the air-to-tissue ratio of the lung. We have analyzed the interdependence of intrathoracic gas volume, lung morphology, and functional disorder by high resolution CT (HRCT) to assess quantitative disease threshold in obstructive and restrictive diffuse lung disease. MATERIALS AND METHODS: Pulmonary HRCT was performed on 24 healthy volunteers, 11 patients with chronic obstructive pulmonary disease (COPD), and 16 patients with idiopathic lung fibrosis (IPF). HRCT measurement was standardized by taking three scans at the carina +/- 5 cm and by defining inspiration levels by percent vital capacity (VC) via spirometrically gating to the scanner. RESULTS: The mean lung density at 50% VC (DL50) for healthy subjects was -819 +/- 3.8 (mean +/- SEM) HU. In contrast, COPD DL50 was lower, averaging -861 +/- 6.4 HU, and the IPF DL50 was considerably higher (-731 +/- 17.7 HU), both significantly different (p < 0.001) compared with the control group. The accuracy of quantitative HRCT at different inspiration levels was evaluated by scanning the basal layer at 20, 50, and 80% VC. The control values were -747 +/- 5.6, -816 +/- 3.6, and -855 +/- 3.0 HU, respectively, which were significantly higher (p < 0.001) than those seen in COPD patients at 20 and 50% VC. Again, the IPF patients exhibited increased lung density (p < 0.001) at all inspiratory levels. Discrimination power was best among all cohorts at 20 and 50% VC. Position-dependent artifacts on lung density were quantified by the anteroposterior density gradient (APG). Irrespective of the underlying disease, APG at 50 and 80% VC was similar, but was up to twofold higher at 20% VC, indicating that quantitative estimates near RV may misrepresent mean lung density. CONCLUSION: Our data indicate that quantitative HRCT measurements should be performed not near full inspiration or expiration, but at an intermediate degree of lung inflation, e.g., 50% VC, for reasons of accuracy, intra- and intersubjective comparability, and feasibility. We conclude quantitative HRCT to be a sensitive tool for the evaluation of diffuse parenchymal lung disease.

Short-term and long-term effects of serial bronchoalveolar lavages in a nonhuman primate model.

Bronchoalveolar lavage (BAL) has gained widespread use as a tool for investigating human lung diseases. In certain cases, it can be useful to obtain BAL material in a serial manner. There is convincing evidence from experimental and clinical studies that BAL can cause influx of neutrophils into the bronchoalveolar space. However, conflicting data have been reported on whether this side effect of BAL also affects previously nonlavaged lung areas. In addition, there is little information available on whether multiple repetitive BAL procedures cause damage to lung tissue. To reexamine the short-term effects of serial BAL procedures, the left lung of 10 cynomolgus monkeys was lavaged with five 20-ml aliquots of saline four times at 24-h intervals (Group A). 72 h after the initial BAL, the right lung was lavaged as a control. The percentage of neutrophils increased significantly (p < 0.05), with the greatest effect seen at 48 h (30.7 +/- 5.8 versus 0.8 +/- 0.3%, mean +/- SEM). No significant changes were observed in the control BAL of the right lung at 72 h. A multidisciplinary approach was used to assess the long-term effects of multiple BAL procedures. BAL was performed 14 times over 26 mo at 2-mo intervals (Group B, n = 5). The right lung was lavaged as a control 25 mo after the initial BAL. In addition to standard cellular BAL parameters, the concentrations of fibronectin, procollagen III amino-terminal peptide-related antigen, total phospholipids, and lactate dehydrogenase activity were measured.(ABSTRACT TRUNCATED AT 250 WORDS)