[Show abstract][Hide abstract] ABSTRACT: Magnetic resonance (MR) diagnosis of regional left ventricular (LV) dysfunction relies on visual interpretation of cine images that suffers from wide inter-observer variability, especially when performed by readers not specifically trained in the assessment of LV wall motion. Quantitative analysis tools, though widely available, are rarely used because they provide large amounts of detailed information, the interpretation of which requires additional time-consuming processing. We tested the feasibility of fast automated interpretation of regional LV function using computer analysis of this wall motion information.
Dynamic, ECG-gated, steady-state free precession short-axis images were obtained in 6-10 slices in 28 subjects (10 normal volunteers; 18 patients). Images were reviewed by an expert cardiologist who provided "gold standard" grades (normal, abnormal) for regional wall motion and, independently, by four radiologists. Same images were then analyzed using custom software. Regional fractional area changes computed in normal volunteers were used to obtain the optimal segment- and slice-specific threshold values for automated classification of regional wall motion for each patient. The levels of agreement with the "gold standard" grades were compared between the radiologists and the automated interpretation.
While the visual interpretation required 2-5 minute per patient, the automated interpretation required < 1 sec, after endocardial border detection was complete. The automated interpretation resulted in higher sensitivity, specificity, and accuracy (84%, 77%, 79%, respectively) than the radiologists' grades (80%, 76%, 77%, respectively) and eliminated the high interobserver variability.
Once the endocardial boundaries are defined, computer analysis of the regional wall motion information allows accurate, fully automated, immediate, objective and experience-independent interpretation of regional LV function.
Journal of Cardiovascular Magnetic Resonance 07/2006; 8(3):427-33. · 4.44 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: To develop a technique for volumetric analysis of real time three dimensional echocardiography (RT3DE) data aimed at quantifying left ventricular (LV) mass and to validate the technique against magnetic resonance (MR) assumed as the reference standard.
RT3DE, which has recently become widely available, provides dynamic pyramidal data structures that encompass the entire heart and allows four dimensional assessment of cardiac anatomy and function. However, analysis techniques for the quantification of LV mass from RT3DE data are fundamentally two dimensional, rely on geometric modelling, and do not fully exploit the volumetric information contained in RT3DE datasets. Twenty one patients underwent two dimensional echocardiography (2DE), RT3DE, and cardiac MR. LV mass was measured from 2DE and MR images by conventional techniques. RT3DE data were analysed to semiautomatically detect endocardial and epicardial LV surfaces by the level set approach. From the detected surfaces, LV mass was computed directly in the three dimensional space as voxel counts.
RT3DE measurement was feasible in 19 of 21 patients and resulted in higher correlation with MR (r = 0.96) than did 2DE (r = 0.79). RT3DE measurements also had a significantly smaller bias (-2.1 g) and tighter limits of agreement (2SD = +/-23 g) with MR than did the 2DE values (bias (2SD) -34.9 (50) g). Additionally, interobserver variability of RT3DE (12.5%) was significantly lower than that of 2DE (24.1%).
Direct three dimensional model independent LV mass measurement from RT3DE images is feasible in the clinical setting and provides fast and accurate assessment of LV mass, superior to the two dimensional analysis techniques.
[Show abstract][Hide abstract] ABSTRACT: Real-time 3D echocardiographic (RT3DE) data sets contain dynamic volumetric information on cardiac function. However, quantification of left ventricular (LV) function from 3D echocardiographic data is performed on cut-planes extracted from the 3D data sets and thus does not fully exploit the volumetric information. Accordingly, we developed a volumetric analysis technique aimed at quantification of global and regional LV function.
RT3DE images obtained in 30 patients (Philips 7500) were analyzed by use of custom software based on the level-set approach for semiautomated detection of LV endocardial surface throughout the cardiac cycle, from which global and regional LV volume (LVV)-time and wall motion (WM)-time curves were obtained. The study design included 3 protocols. In protocol 1, time curves obtained in 16 patients were compared point-by-point with MRI data (linear regression and Bland-Altman analyses). Global LVV correlated highly with MRI (r=0.98; y=0.99x+2.3) with minimal bias (1.4 mL) and narrow limits of agreement (+/-20 mL). WM correlated highly only in basal and midventricular segments (r=0.88; y=0.85x+0.7). In protocol 2, we tested the ability of this technique to differentiate populations with known differences in LV function by studying 9 patients with dilated cardiomyopathy and 9 normal subjects. All calculated indices of global and regional systolic and diastolic LV function were significantly different between the groups. In protocol 3, we tested the feasibility of automated detection of regional WM abnormalities in 11 patients. In each segment, abnormality was detected when regional shortening fraction was below a threshold obtained in normal subjects. The automated detection agreed with expert interpretation of 2D WM in 86% of segments.
Volumetric analysis of RT3DE data is clinically feasible and allows fast, semiautomated, dynamic measurement of LVV and automated detection of regional WM abnormalities.
[Show abstract][Hide abstract] ABSTRACT: Our goals were to: (1) develop a technique for 3-dimensional (3D) direct, model-independent quantitative assessment of left ventricular (LV) volume and ejection fraction based on semiautomated detection of LV endocardial surface from transthoracic near real-time full matrix-array 3D echocardiographic (FM3DE) imaging; (2) evaluate the accuracy of LV volumes obtained with this technique, using cardiac magnetic resonance imaging (MRI) measurements as the reference for comparison; and (3) determine the effects of contrast enhancement on the accuracy of FM3DE measurements. A total of 46 patients underwent 2-dimensional echocardiography, FM3DE, and cardiac MRI. End-diastolic volume, end-systolic volume, and ejection fraction were derived from endocardial borders manually traced from 2-dimensional echocardiographic images and from semiautomatically detected LV cavity from FM3DE data. In 14 patients, FM3DE was also acquired with contrast. All measurements were compared with MRI values using linear regression and Bland-Altman analyses. FM3DE was feasible in 44 of 46 patients with LV volumes < 345 mL. LV volumes and ejection fraction computed from FM3DE resulted in higher levels of agreement with MRI than conventional 2-dimensional echocardiography, with lower interobserver variability. With contrast enhancement, FM3DE significantly underestimated LV volumes and showed increased interobserver variability. Semiautomated LV endocardial surface detection from FM3DE images is feasible and results in fast and accurate assessment of LV function.
Journal of the American Society of Echocardiography: official publication of the American Society of Echocardiography 08/2005; 18(8):779-88. · 2.98 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Quantification of left ventricular (LV) function from the analysis of real-time 3D echocardiographic (RT3DE) data does not fully exploit the dynamic volumetric information contained in these data. Accordingly, we developed a volumetric analysis technique aimed at quantification of global and regional LV function. RT3DE images were analyzed using custom software to extract global and partial volumes and regional wall motion (RWM) throughout the cardiac cycle. RT3DE were compared point-by-point with magnetic resonance (MR) data (linear regression and Bland-Altman analysis) and the feasibility of automated detection of RWM abnormalities was tested by comparison with expert reading of 2D echocardiographic images. LV volumes correlated highly with MR with minimal bias and narrow limits of agreement. RWM correlated highly in basal and mid segments. The automated detection of RWM agreed with expert interpretation in 86% of LV segments. Volumetric analysis of RT3DE data is clinically feasible and allows fast and accurate assessment of cardiac function and automated detection of RWM abnormalities
[Show abstract][Hide abstract] ABSTRACT: The reproducibility of left ventricular (LV) volume and mass measurements based on subjective slice-by-slice tracing of LV borders is affected by image quality, and volume estimates are biased by geometric modeling. The authors developed a technique for volumetric surface detection (VoSD) and quantification of LV volumes and mass without tracing and geometric approximations. The authors hypothesized that this technique is accurate and more reproducible than the conventional methodology.
Images were obtained in 24 patients in 6 to 10 slices from LV base to apex (GE 1.5 T, FIESTA). Volumetric data were reconstructed, and endocardial and epicardial surfaces were detected using the level set approach. LV volumes were obtained from voxel counts and used to compute ejection fraction (EF) and mass. Conventional measurements (MASS Analysis) were used as a reference to test the accuracy of VoSD technique (linear regression, Bland-Altman). For both techniques, measurements were repeated to compute inter- and intra-observer variability.
VoSD values resulted in high correlation with the reference values (EDV: r = 0.98; ESV: r = 0.99; EF: r = 0.91; mass: r = 0.98), with no significant biases (8 ml, 5 ml, 0.2% and -9 g) and narrow limits of agreement (SD: 13 ml, 10 ml, 6% and 9 g). Inter-observer variability of the VoSD technique was lower (range 3 to 5%) than that of the reference technique (5 to 11%; p < 0.05). Intra-observer variability was also lower (1 to 3% vs. 7 to 10%; p < 0.05).
VoSD technique allows accurate measurements of LV volumes, EF, and mass, which are more reproducible than the conventional methodology.
Journal of Cardiovascular Magnetic Resonance 02/2005; 7(3):595-602. · 4.44 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: The accuracy of M-mode and 2D echocardiographic measurements of LV mass is limited. We hypothesized that real-time three-dimensional (RT3D) imaging may allow more accurate measurements. LV mass was calculated from 2D and RT3D images obtained in 21 consecutive patients. The RT3D data resulted in LV mass that correlated with cardiac magnetic resonance measurements better (r=0.90) than 2D (r=0.79). The 2D technique underestimated LV mass (bias 39%). RT3D showed only minimal bias (3%), and reduced the interobserver (37% to 7%) and intraobserver (19% to 8%) variability. RT3D imaging provides the basis for accurate and reliable measurement of LV mass.
[Show abstract][Hide abstract] ABSTRACT: Cardiac MRI diagnosis of regional LV dysfunction relies on subjective interpretation of cine images that suffers from wide inter-observer variability, especially when performed by readers not trained in the assessment of regional wall motion. We developed a method for objective, automated detection of regional wall motion abnormalities. Cine images (GE 1.5 T, FIESTA) were obtained in 6-10 short axis slices in 10 pts. Images were reviewed by an expert cardiologist whose grades were used as the gold standard. Images were then analyzed using custom software. In each slice, endocardial border was detected semiautomatically and divided into 6 segments. Regional fractional area change (RFAC) was computed and displayed as "bull's eyes". RFAC<50% was considered abnormal. The "bull's eyes" showed dark areas in segments judged as abnormal by the expert reader. The automated detection had sensitivity 79%, specificity 78%, accuracy 78%. Analysis of endocardial motion from MRI images provides the basis for accurate, automated and objective interpretation of regional wall motion.
[Show abstract][Hide abstract] ABSTRACT: Left ventricular (LV) mass is an important predictor of morbidity and mortality, especially in patients with systemic hypertension. However, the accuracy of 2D echocardiographic LV mass measurements is limited because acquiring anatomically correct apical views is often difficult. We tested the hypothesis that LV mass could be measured more accurately from real-time 3D (RT3D) data sets, which allow offline selection of nonforeshortened apical views, by comparing 2D and RT3D measurements against cardiac MR (CMR) measurements.
Echocardiographic imaging was performed (Philips 7500) in 21 patients referred for CMR imaging (1.5 T, GE). Apical 2- and 4-chamber views and RT3D data sets were acquired and analyzed by 2 independent observers. The RT3D data sets were used to select nonforeshortened apical 2- and 4-chamber views (3DQ-QLAB, Philips). In both 2D and RT3D images, LV long axis was measured; endocardial and epicardial boundaries were traced, and mass was calculated by use of the biplane method of disks. CMR LV mass values were obtained through standard techniques (MASS Analysis, GE). The RT3D data resulted in significantly larger LV long-axis dimensions and measurements of LV mass that correlated with CMR better (r=0.90) than 2D (r=0.79). The 2D technique underestimated LV mass (bias, 39%), whereas RT3D measurements showed only minimal bias (3%). The 95% limits of agreement were significantly wider for 2D (52%) than RT3D (28%). Additionally, the RT3D technique reduced the interobserver variability (37% to 7%) and intraobserver variability (19% to 8%).
RT3D imaging provides the basis for accurate and reliable measurement of LV mass.
[Show abstract][Hide abstract] ABSTRACT: High spectral and spatial resolution (HiSS) MR data were acquired at 1.5 T using echo-planar spectroscopic imaging from patients with suspicious breast lesions. The water resonances in small voxels are inhomogenously broadened and often have distinct components. Images were calculated with intensity proportional to the Fourier components of the water resonance in each voxel at different offsets from the peak frequency. The results demonstrate that in breast the off-peak Fourier component images of water are qualitatively different from those derived from the peak height of the water resonance. These differences most likely reflect underlying anatomy or physiology. In conventional images, the superposition of the various Fourier components of the water signal may cause loss of detail. The synthesis of water Fourier component images from high spectral and spatial resolution data may provide a new form of contrast, and increase sensitivity to subvoxel physiology and anatomy.
Magnetic Resonance in Medicine 08/2004; 52(1):193-6. · 3.27 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: The aim of this study was to show how evidence-based medicine (EBM) techniques can be applied to the appraisal of diagnostic radiology publications. A clinical scenario is described: a gastroenterologist has questioned the diagnostic performance of magnetic resonance cholangiopancreatography (MRCP) in a patient who may have common bile duct (CBD) stones. His opinion was based on an article on MRCP published in "Gut." The principles of EBM are described and then applied to the critical appraisal of this paper. Another paper on the same subject was obtained from the radiology literature and was also critically appraised using explicit EBM criteria. The principles for assessing the validity and strength of both studies are outlined. All statistical parameters were generated quickly using a spreadsheet in Excel format. The results of EBM assessment of both papers are presented. The calculation and application of confidence intervals (CIs) and likelihood ratios (LRs) for both studies are described. These statistical results are applied to individual patient scenarios using graphs of conditional probability (GCP). Basic EBM principles are described and additional points relevant to radiologists discussed. Online resources for EBR practice are identified. The principles of EBM and their application to radiology are discussed. It is emphasized that sensitivity and specificity are point estimates of the "true" characteristics of a test in clinical practice. A spreadsheet can be used to quickly calculate CIs, LRs and GCPs. These give the radiologist a better understanding of the meaning of diagnostic test results in any patient or population of patients.
European Radiology 06/2004; 14(5):915-22. · 4.34 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Cardiac magnetic resonance (MR) images are often reviewed by non-cardiologists who are not trained in the interpretation of regional left ventricular (LV) function. We hypothesized that the use of still-frame parametric MR images of wall motion could aid in the assessment of regional LV function.
Dynamic, electrocardiogram-gated, steady-state free precession (FIESTA) short-axis images were obtained in 6 to 10 slices in 18 consecutive patients. Each loop was used to automatically generate a still-frame image, in which each pixel is assigned a value equal to the amplitude of cyclic variation in local intensity, resulting in higher intensity in pixels that change between blood and tissue during the cardiac cycle. The dynamic images were reviewed by an expert cardiologist who provided gold standard grades for regional wall motion and by four radiologists. Then the radiologists reviewed and graded the same MR images in combination with parametric images. Grades assigned to each segment in the two sessions were compared with the gold standard.
According to expert interpretation, 6 patients had normal wall motion, and 12 had wall motion abnormalities. Parametric images showed a bright band in the area spanned by endocardial motion, with reduced brightness and thickness in areas of hypokinesis. The agreement between the radiologists' grades and the gold standard significantly improved by adding parametric images (from 77% to 81%), which also resulted in reduced interobserver variability (from 52% to 33%).
Still-frame parametric images aid in the assessment of regional wall motion by non-cardiologists who are required to interpret cardiac images.
Journal of Cardiovascular Magnetic Resonance 02/2004; 6(3):619-25. · 4.44 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Cardiac magnetic resonance imaging (MRI) allows accurate and noninvasive measurements of left ventricular (LV) volumes. Nonetheless, chamber volume quantification is based on manual tracing of endocardial boundaries, which is subjective and time-consuming. We developed a technique for semi-automated detection of 3D endocardial surfaces from MRI data and direct quantification of volumes, and validated it against conventional manual tracing. Eleven patients underwent cardiac MRI (GE, 1.5T FIESTA, short axis views). Custom software was used to semi-automatically detect the LV endocardial surfaces from the MRI data. End-systolic (ESV) and end-diastolic (ESV) volumes were computed directly from voxel counts. Linear regression and Bland–Altman analyses were used to compare semi-automated versus manually traced ventricular volumes and ejection fraction (EF). Semi-automated EDV and ESV showed high correlations with manual values both in EDV and ESV (r=0.97, 0.99, respectively) and EF (r=0.93). No biases and narrow limits of agreement in EDV, ESV and EF estimates were found and a small interobserver variability was calculated for both volumes and EF. The method results in an accurate detection of endocardial surfaces and direct measurement of ventricular volumes and EF.
CARS 2004. Computer Assisted Radiology and Surgery. Proceedings of the 18th International Congress and Exhibition, Chicago, USA, June 23-26, 2004; 01/2004
[Show abstract][Hide abstract] ABSTRACT: To evaluate the efficacy of minimal preparation computed tomography (MPCT) in diagnosing clinically significant colonic tumours in frail, elderly patients.
A prospective study was performed in a group of consecutively referred, frail, elderly patients with symptoms or signs of anaemia, pain, rectal bleeding or weight loss. The MPCT protocol consisted of 1.5 l Gastrografin 1% diluted with sterile water administered during the 48 h before the procedure with no bowel preparation or administration of intravenous contrast medium. Eight millimetre contiguous scans through the abdomen and pelvis were performed. The scans were double-reported by two gastrointestinal radiologists as showing definite (>90% certain), probable (50-90% certain), possible (<50% certain) neoplasm or normal. Where observers disagreed the more pessimistic of the two reports was accepted. The gold standard was clinical outcome at 1 year with positive end-points defined as (1) histological confirmation of CRC, (2) clinical presentation consistent with CRC without histological confirmation if the patient was too unwell for biopsy/surgery, and (3) death directly attributable to colorectal carcinoma (CRC) with/without post-mortem confirmation. Negative end-points were defined as patients with no clinical, radiological or post-mortem findings of CRC. Patients were followed for 1 year or until one of the above end-points were met.
Seventy-two patients were included (mean age 81; range 62-93). One-year follow-up was completed in 94.4% (n=68). Mortality from all causes was 33% (n=24). Five histologically proven tumours were diagnosed with CT and there were two probable false-negatives. Results were analysed twice: assuming all CT lesions test positive and considering "possible" lesions test negative [brackets] (95% confidence intervals): sensitivity 0.88 (0.47-1.0) [0.75 (0.35-0.97)], specificity 0.47 (0.34-0.6) [0.87 (0.75-0.94)], positive predictive value 0.18 [0.43], negative predictive value 0.97 [0.96], positive likelihood ratio result 1.6 [5.63], negative likelihood ratio result 0.27 [0.29], kappa 0.31 [0.43]. Tumour prevalence was 12%. A graph of conditional probabilities was generated and analysed. A variety of unsuspected pathology was also found in this series of patients.
MPCT should be double-reported, at least initially. "Possible" lesions should be ignored. Analysis of the graph of conditional probability applied to a group of frail, elderly patients with a high mortality from all causes (33% in our study) suggests: (1) if MPCT suggests definite or probable carcinoma, regardless of the pre-test probability, the post-test probability is high enough to warrant further action, (2) frail, elderly patients with a low pre-test probability for CRC and a negative MPCT should not have further investigation, (3) frail, elderly patients with a higher pre-test probability of CRC (such as those presenting with rectal bleeding) and a negative MPCT should have either double contrast barium enema (DCBE) or colonoscopy as further investigations or be followed clinically for 3-6 months. MPCT was acceptable to patients and clinicians and may reveal significant extra-colonic pathology.
[Show abstract][Hide abstract] ABSTRACT: To evaluate the effect of varying spectral resolution on image quality of high spectral and spatial resolution (HiSS) images.
Eight women with suspicious breast lesions and six healthy volunteers were scanned using echo-planar spectroscopic imaging (EPSI) at 1.5 Tesla with 0.75- to 1-mm in-plane resolution and 2.3- to 2.6-Hz spectral resolution. Time domain data were truncated to obtain proton spectra in each voxel with varying (2.6-83.3 Hz) resolution. Images with intensity proportional to water signal peak heights were synthesized. Changes in water signal line shape following contrast injection were analyzed.
Fat suppression is optimized at approximately 10-Hz spectral resolution and is significantly improved by removal of wings of the fat resonance. This was accomplished by subtracting a Lorentzian fit to the fat resonance from the proton spectrum. The water resonance is often inhomogeneously broadened, and very high spectral resolution is necessary to resolve individual components. High spectral resolution is required for optimal contrast in anatomic features with very high T(2)* (e.g., within a lesion) and for detection of often subtle effects of contrast agents on water signal line shape.
Despite a trade-off between the spectral resolution and signal-to-noise ratio, it is beneficial to acquire data at the highest spectral resolution currently attainable at 1.5 Tesla.
Journal of Magnetic Resonance Imaging 11/2003; 18(4):442-8. · 2.57 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: To use evidence-based techniques to compare elective open surgical repair of abdominal aortic aneurysms with endovascular repair by means of stent placement.
A focused clinical question formed the basis of a literature search. Evidence-based criteria were used to appraise and assign a "level of evidence" to retrieved articles. The following data were determined from the best studies: systemic, local, and/or vascular complications; graft failure rates; blood loss; mortality; length of intensive care and/or hospital stay; mid- and long-term outcomes; cost of endovascular repair versus that of surgery; and eligibility for endovascular repair. Absolute risk reductions and/or increases and numbers needed to treat or harm were calculated.
The best current evidence came from 22 studies, which showed that there is slight, if any, difference between mortality rates of endovascular repair and surgery. Hospital and/or intensive care stay is shorter, blood loss less, and systemic complications fewer (numbers needed to treat, two to 12) with endovascular repair. Some authors reported a significant increase in local and/or vascular complications with endovascular repair (numbers needed to harm, two to six). Graft failure is significantly more common with endovascular repair (numbers needed to harm, four), and substantive adjunctive interventions are needed. Endovascular repair is more expensive than surgery.
Elective endovascular repair has short-term benefits compared with surgery. There is slight, if any, difference in mortality. Endovascular repair costs more than surgery. At follow-up, surgical grafts performed better.