| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||
ABSTRACT |
Tuesday, May 2, 1:30 PM–3:30 PM
Abstracts 120–131
Moderators: Mylene Truong, MD and Smita Patel, MD
1:30 PM
120. Diagnostic Accuracy of 4-slice, 12-slice, 16-slice and 64-slice CT Systems in Noninvasive Coronary Angiography
Pugliese F.1*; Cademartiri F.1; Mollet N.R.1; Nieman K.2; Meijboom W.B.2; deFeyter P.J.2; Krestin G.P.1; 1. Department of Radiology, Erasmus Medical Center, Rotterdam, Netherlands; 2. Department of Cardiology, Erasmus Medical Center, Rotterdam, Netherlands.
Address correspondence to F. Pugliese (francesca.pugliese{at}libero.it)
Objective: To perform a comparison of the diagnostic accuracy of 4 different generations of multidetector CT scanners featuring 4, 12, 16 and 64 slices per rotation (in cardiac protocols) respectively in the assessment of coronary artery disease (CAD), with invasive coronary angiography as the reference standard.
Materials and Methods: Four groups were formed including the first 51 patients undergoing CT coronary angiography after the installation of each CT system. All patients (n = 204) were referred for conventional angiography for suspected CAD. Subjects with previous percutaneous angioplasty, stent placement and coronary artery bypass grafting were excluded. Gantry rotation times were 500 ms for the 4-slice, 420 ms for the 12-slice, 375 ms for the 16-slice and 330 ms for the 64-slice scanner.
Results: Heart rates and prevalence of CAD were comparable among the groups (mean HR = 57 bpm, distribution of disease = 1.3 lesion/patient). In the 4-slice group, 113/442 (26%) segments were excluded from analysis because of poor image quality, whereas all segments > 2 mm diameter were evaluable in the other groups. Sensitivity, specificity, PPV and NPV were 58%, 94%, 61% and 94% for 4-slice CT; 90%, 93%, 65% and 99% for 12-slice CT; 97%, 98%, 87% and 99% for 16-slice CT; 99%, 96%, 80% and 99% for 64-slice CT. In the 64-slice CT group, including the assessable < 2 mm diameter coronary branches (26%), the respective values were 99%, 95%, 76% and 99%.
Conclusion: Diagnostic accuracy of CT coronary angiography performed with 4-slice CT scanners is significantly lower than that allowed for by newer generation systems. Between the latters, a trend increase in diagnostic accuracy is seen. 64-slice CT has high diagnostic accuracy even when smaller coronary branches are included.
121. Coronary CTA with 64-slice MDCT: Assessment of Vessel Visibility
Pannu H.K.1*; Jacobs J.E.2; Lai S.3; Fishman E.K.1; 1. Radiology, Johns Hopkins Medical Institutions, Baltimore, MD; 2. Radiology, New York University Medical Center, New York, NY; 3. Pathology, Johns Hopkins Medical Institutions, Baltimore, MD.
Address correspondence to H.K. Pannu (hpannu1{at}jhmi.edu)
Objective: To evaluate the image quality of 64-slice MDCT for coronary angiography.
Materials and Methods: 50 consecutive MDCT coronary angiograms on a 64 slice scanner were independently reviewed by 2 readers. Segments were scored as no motion (score 1), minimal motion (score 2), moderate motion (score 3), respiratory motion (score 4) and vessel blurring (score 5). Opacification was graded as good or limited. Segments < 2 mm were graded as well seen or poorly seen/not seen. The scores for motion artifact, opacification and visibility were combined for overall vessel assessment. Segments with a motion score of 1 or 2 that had good opacification and were well seen were judged to be assessable.
Results: A total of 714 segments were analyzed in the 50 patients. 700 segments were assessed in all patients (segments 1–3, 11–20, 4 or 27) and 14 patients had a ramus inter-medius segment. Combining the scores for both readers: the average motion score was 1 for 617.5 segments (86.8%), the average motion score for all segments in an individual patient was 1.14 (range 1–3.35) and the average opacification score for all segments in a patient was 1.02 (range 1–1.38). There were a total of 374 segments less than 2 mm in diameter. Combining the scores for both readers: an average of 36 (5.0% of 714) segments could not be identified by the readers, 319.5 segments (85.4% of 374) were well seen and 18.5 segments (4.9% of 374) were poorly seen. Overall, an average of 631.5 (88.4%) segments were judged assessable by the readers. On a per patient basis, 10 or more vessel segments were judged assessable in 47 patients (94%).
Conclusion: On 64 slice MDCT, 88% of coronary artery segments are assessable. 10 or more vessel segments are assessable in 94% of patients.
122. ECG gated 64 Slice CT Angiography for the Differential Diagnosis of Acute Chest Pain
Johnson T.R.1*; Nikolaou K.1; Wintersperger B.J.1; Knez A.2; Rist C.1; Boekstegers P.2; Reiser M.F.1; Becker C.R.1; 1. Department of Clinical Radiology, University of Munich, Grosshadern Hospital, Munich, Bavaria, Germany; 2. Medical Clinic I, Department of Cardiology, University of Munich, Grosshadern Hospital, Munich, Bavaria, Germany.
Address correspondence to T.R. Johnson (thorsten.johnson{at}med.uni-muenchen.de)
Objective: The most important differential diagnoses of acute chest pain include myocardial infarction, aortic dissection and pulmonary embolism. The purpose of this study was to evaluate the potential of a new generation of 64-slice CT systems for the simultaneous assessment of pulmonary arteries, coronary arteries and the aorta within a single breath-hold in the differential diagnosis of acute chest pain.
Materials and Methods: ECG-gated CT-angiography of the whole thorax was performed with a 64-slice CT-scanner (Somatom Sensation 64, Siemens Medical Solutions, Erlangen, Germany) in 55 patients suffering from acute chest pain. Collimation was 64 x 0.6 mm with double z sampling, gantry rotation time was 0.33 sec. Volume and flow rate of contrast media injection were adapted to the scan length (mean, 135 ml at 4.6 ml/s). Intravascular density measurements and visual assessment of motion artifacts were performed to evaluate image quality. CT findings were correlated to reference imaging modalities, laboratory tests, physical examination and clinical followup. In 20 patients, findings of CT coronary angiography were compared to those of X-ray coronary angiography by two independent blinded readers.
Results: Adequate contrast enhancement of the pulmonary vessels, the coronary arteries and the aorta could be achieved in all individuals (mean 330 HU). There was only minor blurring in seven patients and one examination with non-diagnostic images of the coronary arteries, and there were slight motion artifacts of the pulmonary arteries seen in two patients. An adequate diagnosis causing the chest pain was found in 37 patients. Diagnoses included pulmonary embolism (n = 9), coronary stenosis (n = 8) and aortic aneurysm or dissection (n = 6). The specific cause of chest pain remained unclear at follow-up in fourteen patients. Comparing coronary CTA to conventional coronary angiography, sensitivity, specificity and accuracy on a per-patient basis were 94%, 77% and 87%, respectively.
Conclusion: With current techniques, an ECG gated CT angiography of the whole chest is feasible with very good image quality. The protocol proved to be helpful in the differential diagnosis of acute chest pain.
123. Aortic Root Catheter-Directed Coronary CT-Angiography in Swine: Coronary Enhancement with Minimum Iodinated Contrast Volumes
Kumar A.*; Bis K.G.; Shetty A.N.; Vyas A.; Anderson A.; Stein W.; O'Neill W.; Radiology, Beaumont Hospital, Royal Oak, MI.
Address correspondence to A. Kumar (arun2786{at}hotmail.com)
Objective: To evaluate the minimum amount of contrast required for coronary imaging with aortic root (AR) catheter-directed enhancement and multidetector (64-slice) computed tomographic angiography (MDCTA).
Materials and Methods: A 64 slice CT scanner (Siemens Medical Solutions, Inc.) was used after IRB approval in 4 ventilated and sedated swine. Heart rate reduction to 65 bpm was achieved with PO atenolol and IV cardizem. Common femoral arterial access was obtained with a 5-Fr micropuncture kit using ultrasound guidance. A diffusion tip (640 side-holes) 5 Fr pigtail catheter (Medrad, Inc.) was positioned in the aortic root on the CT table with c-arm fluoroscopy and connected to an arterial power injector (Medrad, Inc). ARMDCTA (retrospective ECG-gating, collimation 0.6mm, tube rotation time 0.3 sec, scan time 10–12 sec, tube voltage 120 Kvp, effective mAs 850, pitch 0.2, FOV 109–123 mm, slice thickness/increment 0.6 mm/0.3 mm) commenced 1 second after injection of a 100 cc volume of various Visipaque concentrations (10%, 20%, 30%, 40%) at 10cc/second. Coronary mean and peak densities, 3D-MIP's and 4D projections were obtained.
Results: The mean proximal, mid and distal coronary attenuation values [HU ± SD(standard deviation)] for the right (RCA), left anterior descending (LAD) and left circumflex (LCx) coronary arteries at various concentrations were as follows: 10% RCA (270.0±42.3, 321.5±10.6, 329.0±14.1), LAD (240.5±31.6, 197.8±24.0, 147.5±15.1), LCx (248.5±78.5, 200.0±17.7, 196.2±9.7) (n = 4). 20% RCA (335.0±81.1, 257.0±43.4, 409.8±111.5), LAD (351.0±19.6, 336.5±31.5, 246.5±46.3), LCx (391.5±61.5, 329.3±18.2, 236.8±42.2) (n = 4). 30% RCA (461.5±133.5, 362.3±172.2, 536.8±53.3), LAD (439.3±118.2, 424.5±157.1, 345.3±75.2), LCx (458.3±189.2, 490.8±264.9, 293.3±29.9) (n = 4). 40% RCA (582.0±48.4, 737.3±241.9, 769.0±164.7), LAD (678.0±35.6, 637.3±107.7, 411.8±112.3), LCx (739.0±75.9, 706.3±56.8, 599.0±67.2) (n = 4). Coronary imaging with AR-enhanced MDCTA was feasible at all concentrations. The definition and clarity of tributary and distal anatomy improved with increasing concentration of contrast. The results obtained with the minimum amount of contrast are comparable to those currently obtained with peripheral IV MDCTA using 10–20% the contrast volume.
Conclusion: AR-enhanced MDCTA can depict the coronary arteries with as little as 10–20cc of contrast. This would provide an alternative means of coronary evaluation for patients in whom a minimal contrast load is desirable such as those with renal insufficiency.
124. Segmented 2D Bright-Blood Cine MRI for Measuring Aortic Root Dimensions: Robustness, Agreement and Correlation with Two-dimensional Transthoracic Echocardiography
Stolpen A.H.1*; Aggarwala G.2; Iyengar N.2; Thedens D.R.1; Scholz T.D.3; Willing M.C.4; Weiss R.M.2; Sonka M.5; 1. Radiology, University of Iowa, Iowa City, IA; 2. Internal Medicine, Division of Cardiovascular Diseases, University of Iowa, Iowa City, IA; 3. Pediatrics, Division of Cardiology, University of Iowa, Iowa City, IA; 4. Pediatrics, Division of Genetics, University of Iowa, Iowa City, IA; 5. Electrical and Computer Engineering, University of Iowa, Iowa City, IA.
Address correspondence to A.H. Stolpen (alan-stolpen{at}uiowa.edu)
Objective: Aortic root dilatation is an ominous complication of many inherited connective tissue disorders. High-risk patients benefit from accurate and reproducible serial imaging to measure aortic root dimensions and establish the need for surgery. 2D transthoracic echocardiography (Echo) is the primary modality for imaging the aortic root. However, Echo quality suffers when the acoustic window is poor. MRI offers an attractive alternative. The purpose of this study was to compare the performance of MRI and Echo for measuring the aortic root.
Materials and Methods: Over a 6.6 yr period, 200 consecutive MRI with aortic root measurements were performed. Each MRI was matched with a contemporaneous Echo (when available) that also included aortic root measurements (actual or attempted). Aortic root measurements were obtained in diastole from parasternal long axis Echo and MR images at the annulus, sinuses of Valsalva, sinotubular junction and proximal ascending aorta. MRI was performed at 1.5T using a breath-hold, segmented 2D cine SSFP or GRE sequence with 20 cardiac phases. For each modality, the study quality and the number of successful measurements at each aortic root location were recorded. Bland-Altman and regression analysis were used to assess the agreement and correlation, respectively, between Echo and MRI measurements.
Results: The study group consisted of 170 Echo-MRI pairs on 137 patients (50F, 87M, ages 7–80 yrs). The median time between Echo and MRI was 32 d. Fifty Echoes (29%) and 3 MRI (2%) were technically difficult. Measurements were successfully acquired by Echo vs. MRI at the annulus in 78% vs. 96%, at the sinuses in 85% vs. 99%, at the sinotubular junction in 74% vs. 98% and at the proximal aorta in 57% vs. 96%. All differences were significant at p < 0.0001 (McNemar's test). MRI showed a small negative bias (mean difference ± SD) at the annulus (–1.5 ± 4.4 mm, n = 130), sinuses (–0.1 ± 3.8 mm, n = 144), sinotubular junction (–3.5 ± 4.4 mm, n = 125) and proximal aorta (–2.8 ± 5.4 mm, n = 94). The correlation coefficients (r) for Echo and MRI measurements were 0.607 at the annulus, 0.903 at the sinuses, 0.795 at the sinutubular junction and 0.705 at the proximal aorta.
Conclusion: Cine MRI is a highly robust technique for measuring the aortic root. Echo and MRI measurements correlate well. MRI exhibits a small, clinically acceptable negative bias. However, uniformly large limits of agreement (bias ± 2SD) raise questions about the variability of one or both modalities. This will require further study.
125. Coronary, Aortic and Pulmonary Enhancement Using a Biphasic Single Injection 64-slice CT-Angiography Protocol
Bis K.G.1*; Vrachliotis T.G.2; Haidary A.1; Kosuri R.1; Balasubramaniam M.1; Oneil B.3; O'neill W.4; 1. Radiology, William Beaumont Hospital, Royal Oak, MI; 2. Radiology, Henry Dunant Hospital, Athens, Greece; 3. Emergency Medicine, William Beaumont Hospital, Royal Oak, MI; 4. Cardiology, William Beaumont Hospital, Royal Oak, MI.
Address correspondence to K.G. Bis (kbis{at}beaumont.edu)
Objective: To evaluate a CT-angiography (CTA) protocol which enhances the coronary, aortic, and pulmonary vasculature with a single-breath-hold acquisition.
Materials and Methods: With IRB approval and informed consent, 50 Emergency Department (ED) patients with atypical chest pain were imaged with a Siemens 64-slice CT scanner following oral atenolol and/or IV metoprolol. Bolus timing in the mid-ascending aorta with 20 ml contrast (Visipaque) (5 ml/sec followed by a 50 ml saline flush at 5 ml /sec) preceded thoracic CTA which commenced 4–5 seconds after aortic peak density (100 ml Visipaque at 5 ml/sec, followed by 30 ml Visipaque at 3 ml/sec and saline flush with 50 ml saline at 3 ml/sec) with a caudal-to-cranial acquisition: Retrospective ECG-gating, collimation 0.6 mm, tube rotation 0.3 sec, tube voltage 120 mV, effective mAs 750–850, pitch 0.2. Reconstruction FOV, slice thickness/recon increment and kernels were: coronary arteries (15–22 cm, 0.6 mm/0.3 mm, B25 f smooth), mediastinum/soft tissues(35–42 cm, 2 mm/1mm, B31 f medium smooth) and lungs (35–42 cm, 3 mm/2mm, B70 f very sharp). Vascular density (HU), cardiac venous enhancement grading and right atrial streak artifact were assessed.
Results: Mean attenuation values [HU±SD (standard deviation)] were: RCA P (299.0 ± 62.9) M (328.4 ± 61.8) D (308.6 ± 61.3), LAD P (287.0 ± 57.6) M (264.5 ± 45.2) D (198.1 ± 43.7), LCx P (305.2 ± 54.9) M (290.0 ± 35.1) D (286.1 ± 68.9) (RCA–Right coronary artery, LAD–Left anterior descending artery, LCx-Left circumflex artery, P-proximal, M-middle, D-distal). LUL (299 ± 86.4), TA (327 ± 79.8), LLLS (331 ± 81.3), RLLS (315 ± 84.1), RM (310 ± 73.3), LM (315 ± 74.9) (LUL-Left upper lobe, TA-truncus anterior, LLLS-left lower lobe segmental, RLLS-right lower lobe segmental, LM-left main, RM-right main). Mid ascending aorta (331.6 ± 61.2), proximal descending aorta (321.8 ± 58.6) and distal descending aorta (347.5 ± 61.0). Cardiac venous enhancement grading was: Middle cardiac vein [0 (none): 36%, 1 (mild): 62%, 2 (moderate): 2%, 3 (=arterial): 0%], Great cardiac vein [0:2%, 1:43%, 2:55%, 3:0%] and Coronary sinus [0:2%, 1:76%, 2:21%, 3:0%]. Right atrial streak artifact was present in 88% of studies, but focally traversed the RCA in only 1-patient. Cardiac venous enhancement and right atrial streak artifact did not interfere with interpretation of coronary arterial anatomy.
Conclusion: This CTA protocol provides adequate enhancement of coronary, aortic and pulmonary vasculature in a single breath-hold and is currently used for ED patients with atypical chest pain.
126. Time-resolved MR Angiography for Pulmonary Embolism: Preliminary Results
Ersoy H.1*; Goldhaber S.Z.2; Skorstad R.B.1; Cai T.3; Rosebrook J.L.1; Rybicki F.J.1; 1. Department of Radiology, Brigham and Women's Hospital, Boston, MA; 2. Department of Cardiology, Brigham and Women's Hospital, Boston, MA; 3. Department of Biostatistics, Harvard Medical School, Boston, MA.
Address correspondence to H. Ersoy (hersoy{at}partners.org)
Objective: Although pulmonary 3D Gd-MRA has high accuracy for pulmonary embolism (PE) diagnosis, routine implementation has been limited by technical and practical factors such as venous contamination from suboptimal bolus timing and the need of high spatial resolution to evaluate small arterial segments. The purpose of this study is to determine the reproducibility of high quality PE images by using 3D TRICKS (Time-Resolved Imaging of Contrast Kinetics).
Materials and Methods: 12 consecutive patients (5M, 7F; mean = 60y) with clinically suspected PE and contraindication for iodinated contrast underwent MRA (1.5T Signa GE, 8-ch phased array coil). Parameters: TR/TE = 3.5/1.3 ms, FA = 35°, FOV = 34 cm, matrix = 256 x 192, slice thickness = 3.0 mm (interpolated to 1.5 mm), 32 slices on the coronal plane. Acq time was 40s with a scan delay time of 3–4s; 40 mL Gadopentetate dimeglumine with an injection rate of 3 mL/s. 9 phases acquired with a temporal output rate of 1 phase per 3.5s. For analysis, pulmonary arterial anatomy was divided into 26 parts: main, right and left, 5 lobar, and 18 segmental branches. 2 radiologists independently scored image quality using a 3 point scale (1-nonvisualized; 2-visualized but nondiagnostic; 3-diagnostic). For all diagnostic parts, the pres-ence (filling defect persisting through all phases) or absence of PE was assessed. Interrater agreement (weighted kappa, kw) was analyzed for image quality and the presence of PE. MR findings were correlated with CT for the 4/12 patients for whom the contraindication for iodinated contrast was corrected after the MR.
Results: Artery-only images were constantly acquired in all patients without complication. For both readers, the main, right, and left pulmonary arteries were diagnostic in all studies. Readers independently scored 96% of the lobar artery images as diagnostic. Segmental artery images were diagnostic in 90% (reader 1) and 94% (reader 2). Interrater image quality agreement was very good for lobar (kw = 0.85; SE:0.05; 95% CI: 0.56,1.14) and good for segmental (kw = 0.77; SE:0.03; 95% CI:0.70,0.83) arteries. Both readers independently diagnosed PE in 4/12 cases in identical parts. For the 4/12 patients who subsequently underwent CTA, 1 had PE by CT; for all 4 patients, the MR findings were identical to the CTA findings with respect to all 26 parts.
Conclusion: 3D TRICKS provides a simple, rapid (9 phases, 1 phase / 3.5s), and robust (artery-only, high spatial resolution) pulmonary MRA strategy, enabling confident diagnosis of PE through the segmental branches.
127. Evaluation of Image Postprocessing and ECG Gating in MDCT of the Right Ventricle in Patients with Acute Pulmonary Embolism
Lu M.T.1*; Ersoy H.1; Whitmore A.G.1; Cai T.2; Meadows J.L.3; Goldhaber S.Z.3; Rybicki F.J.1; 1. Cardiovascular Imaging Section, Brigham and Women's Hospital, Boston, MA; 2. Biostatistics, Harvard School of Public Health, Boston, MA; 3. Medicine, Cardiovascular Division, Brigham and Women's Hospital, Boston, MA.
Address correspondence to M.T. Lu (mlu4{at}partners.org)
Objective: Pressure overload of the right ventricle (RV) and subsequent RV dysfunction (RVD) causes significant morbidity and mortality in patients with acute pulmonary embolism (PE). The literature increasingly supports the prognostic value of CT analysis of the RV. However, there is a spectrum of complexity for CT measurements of the RV, ranging from nongated axial RV/LV diameter ratios (more simple) to ECG gated 4-chamber (4ch) RV/LV volume ratios (more complex). The purpose of this study is to test the hypothesis that measurement parameters derived from image post-processing and/or ECG gating significantly differ from simpler parameters obtained from routine CT angiography (CTA) performed without ECG gating.
Materials and Methods: Prospective analysis was performed on 65 consecutive patients (40F, mean = 63+/-14 y) diagnosed with acute PE by standard nongated pulmonary CTA. Spearman's Rank Correlation Coefficient (RCC) was computed for (1) axial vs 4ch RV diameter and (2) axial vs 4ch RV/LV diameter ratio. In 11 of these 65 patients (7F, mean = 62+/-14 y), the nongated CTA was complemented with ECG gated cardiac CT. For these 11 patients, the RCC was computed for (1) gated end-diastolic vs nongated axial RV/LV diameter ratio and (2) gated end-diastolic vs nongated 4ch RV/LV diameter ratio. In addition, for these 11 patients, the RCC was computed for the RV/LV volume ratio in end-diastole vs end-systole. A 0.90 confidence interval (CI) for each RCC was calculated via the bootstrap method. The study was approved by the institutional human research committee.
Results: For all 65 patients, the RCC [0.90 CI] for the axial vs 4ch RV diameter and axial vs 4ch RV/LV diameter ratio were 0.77 [0.74,0.90] and 0.77 [0.69,0.88], respectively. For the 11 gated studies, the RCC for the gated end-diastolic vs nongated axial RV/LV diameter ratio, gated end-diastolic vs nongated 4ch RV/LV diameter ratio, and RV/LV volume ratio in end-diastole vs end-systole were 0.94 [0.74,1.0], 0.75 [0.40,0.92], and 0.92 [0.69,1.0], respectively.
Conclusion: This study suggests that parameters derived from image postprocessing and additional dedicated ECG gated RV imaging do not significantly differ from parameters obtained from standard nongated axial CTA images alone. Consequently, prognostic RV data can be reported without subjecting patients to the risk from additional contrast material and radiation.
128. Computer-aided Detection of Peripheral Pulmonary Embolus on Multidetector Row CT: Initial Experience and Impact on Resident Diagnosis
Lake D.R.1*; Seekins J.M.1; Cluver A.R.1; Fernandez M.J.1; Ravenel J.G.1; Schoepf U.J.1; Salganicoff M.2; Liang J.2; Nguyen S.1; Dundar M.2; Wolf M.2; Costello P.1; 1. Department of Radiology, Medical University of South Carolina, Charleston, SC; 2. CAD Solutions Group, Siemens Medical Systems, Inc. USA, Malvern, PA.
Address correspondence to D.R. Lake (lake{at}musc.edu)
Objective: To determine the improvement in diagnosing segmental and subsegmental pulmonary embolism (PE) at CT pulmonary angiography, when a computer-aided detection (CAD) prototype for filling defect detection (FDD) is added to interpretation by resident readers.
Materials and Methods: IRB waiver was obtained. Contrast enhanced CT pulmonary angiograms [16-slice scanner (GE), 1.25 mm section thickness, 0.6 mm interval] of twenty-five patients with suspected pulmonary embolus were retrospectively analyzed. Four radiology residents independently reviewed all CT scans on a dedicated workstation and recorded sites of suspected PE on a segmental and subsegmental arterial level. The FDD CAD prototype (Siemens CAD Group, Malvern, PA) deploys detection marks on candidate filling defects, suspicious of PE. The same residents reanalyzed all studies, this time with the aid of the FDD CAD prototype. The resident's performance for diagnosis of PE with and without FDD CAD were compared to the expert read of a board-certified thoracic radiologist.
Results: On reference reading, the 25 scans showed 107 confirmed peripheral pulmonary emboli. A critical distance of 10 mm along the vessel course was used to determine if resident marks coincided with reference emboli. Finding-level (on a vessel-by-vessel basis) and case-level (on a per-patient basis) analysis was performed. Finding-level reader sensitivity with and without FDD CAD was: reader 1 (46.7% vs 52.3%), reader 2 (57.9% vs 59.8%), reader 3 (55.1% vs 60.7%) and reader 4 (54.2% vs 62.6%). Case-level reader sensitivity before and after FDD CAD was: reader 1 (70.8% vs 83.3%), reader 2 (79.2% vs 87.5%), reader 3 (100% vs 100%) and reader 4 (91.7% vs 100%). Overall, mean detection of PE on a vessel-by-vessel basis by resident readers was significantly increased from 53.5 vs 58.9 (p < 0.028). On a per-patient basis, FDD CAD enabled correct PE diagnosis for an additional 4 patients overall. The mean false positive rate of FDD CAD was 2.4 per case versus the reference standard.
Conclusion: Use of the prototype FDD CAD system improved the residents' detection of peripheral pulmonary emboli at CT pulmonary angiography.
129. The Utility of Computer-aided Detection (CAD) for the Automated Evaluation of Pulmonary Arterial Filling Defects
Colak E.*; Kucharczyk M.; Kucharczyk S.; Zukotynski K.; Patsios D.; Sitartchouk I.; Roberts H.; Department of Medical Imaging, University Health Network, Toronto, ON, Canada.
Address correspondence to E. Colak (errol.colak{at}utoronto.ca)
Objective: Contrast-enhanced helical computed tomography (CT) angiography has become the standard tool for the detection and exclusion of acute pulmonary embolism. With the development of multi-row detector scanners, thin sections are acquired during the contrast bolus, resulting in several hundred images per CT angiogram. The purpose of this study is to validate a computer aided diagnosis (CAD) tool for the automated detection of pulmonary embolism.
Materials and Methods: 100 consecutive multi-detector row (MD)CT scans performed to rule out pulmonary embolism were retrospectively evaluated with a commercially available CAD software (ImageChecker CT, R2 Technology, Sunnyvale, CA). The software automatically highlights filling defects in segmental and subsegmental pulmonary arterial branches for further review and characterization. All CAD marks were re-assessed by a chest radiologist, and characterized as: true positive, TP: intravascular filling defect found by CAD and confirmed by the radiological re-assessment; false positive, FP: any CAD-identified entity that was not a vascular filling defect; false negative, FN: any vascular filling defect that was found by the radiologist, but not marked by CAD. If CAD did not mark an entry in a study negative for PE, this was defined as true negative (TN).
Results: Of the 100 CT scans, 56 were performed with 1 mm, 44 with 1.25 mm slice thickness. 78 were judged as optimal quality, 19 as good quality, 3 as suboptimal. 18 CT studies were positive for pulmonary embolism, 82 were negative. Of the 18 positive, CAD had marks in 12 (true positive), no marks in 6 (false negative) cases. Of the 82 negative CT scans, CAD had marks in 37 (false positive), no marks in 45 (true negative) cases. Analyzing the individual data entries, R2 yielded 537 marks in the 100 CT scans. 31 of those (6%) were true positive, 506 (91%) false positive; most of the FP marks were in pulmonary veins (372, 74% of all FP). 20 filling defects were missed by CAD.
Conclusion: First generation computer aided detection of pulmonary embolism has a high negative predictive value based on retrospective review of high-resolution MDCT pulmonary angiography. The majority of FP data entries are in pulmonary veins, which can be dismissed based on anatomy by the reading radiologist.
130. Additional Benefit of CT Venography for the Diagnosis of Thromboembolic Disease in an Acute Tertiary Care Setting
Rhee K.1*; Ko J.P.1; Iyer R.S.1; Cha S.1; Jacobowitz G.2; Naidich D.P.1; 1. Radiology, New York University Medical Center, New York, NY; 2. Vascular Surgery, New York University Medical Center, New York, NY.
Address correspondence to K. Rhee (rheek01{at}endeavor.med.nyu.edu)
Objective: To determine the additional benefit of performing lower extremity CT venography (CTV) with pulmonary CT angiography (CTA) in the diagnosis of thromboembolic disease.
Materials and Methods: Reports of all CTAs and CTVs over a 3 year
interval (Group I) and CTAs, CTVs, and lower extremity Doppler ultrasounds
(US) over a 1 
year subset (Group II) performed on inpatients and
emergency department patients who were assessed for pulmonary embolism (PE)
and deep venous thrombosis (DVT) at a tertiary care hospital were reviewed.
Reported results for each study were categorized as positive, negative,
suspicious or indeterminate for PE or DVT. When CTV and US results were
discrepant, medical records were reviewed for clinical management. Additional
benefit of CTV was assessed by Chi-square analysis.
Results: In Group I, in 909 concurrent CTA and CTVs, 737 (8.1%) CTAs were negative. For the 737 cases with negative CTAs, CTV increased the diagnosis of thromboembolic disease in 6.2% (46 positive or suspicious CTVs) or 5.3% (39 positive CTVs) (p <= 0.001). When considering all 909 combined CTA/ CTVs, the rate of thromboembolic disease increased from 13.0% to 17.3% or to 18% for positive studies only and positive and suspicious studies, respectively (p = 0.01). In Group II, 119 cases had CTA, CTV, and US, with 14 discordant CTVs and USs. In 11 of the 14 with clinical follow up, 2 of 119 (1.7%) CTVs were considered falsely negative (USs truly positive). Nine CTVs were suspicious (n = 4) or positive (n = 5) while USs were negative. Three of 4 suspicious CTVs were considered truly positive (US falsely negative); one case was accompanied by a positive CTA, so CTV affected management in 2 of 4. Three of 5 positive CTVs were treated as true positives; however, CTA was positive in one case, and therefore CTV affected management in 2 of 5. The 4 positive CTVs that altered management increased the rate of thromboembolism diagnosis from 25.2% to 28.6% (34/119), although not proven significant (p > 0.05).
Conclusion: Combined CTA/ CTV can increase diagnosis of thromboembolic disease in an acute care situation.
131. Comparison of Helical and Axial Mode Indirect CT Venography in Patients with Pulmonary Thromboembolism
Kalra N.*; Gupta A.; Bhalla A.; Khandelwal N.; Suri S.;. Department of Radiodiagnosis & Department of Internal Medicine, Postgraduate Institute of Medical Education and Research, Chandigarh, U.T., India.
Address correspondence to N. Kalra (navkal2004{at}yahoo.com)
Objective: To compare the diagnostic utility of multidetector helical and axial indirect CT venography for diagnosing deep venous thrombosis (DVT) of the lower extremities in patients proven to have pulmonary thromboembolism (PTE) on CT pulmonary angiography(CTPA).
Materials and Methods: 106 consecutive patients clinically suspected to have PTE underwent CTPA on a multidetector helical CT scanner. In 20 of these patients who had PTE on CTPA, indirect CT venography of the lower extremities was done. CT venography (CTV) was done 180 seconds after the start of the contrast injection. The patients were randomly assigned to the helical (5/2.5 mm) and the axial modes (5/15 mm). Volume rendered reconstructions were done for the helical mode. The radiation dose was calculated for each patient by using commercially available software. Doppler ultrasound of the deep venous system was also done within 24 hours of the indirect CT venography. The CT and Doppler studies were initially interpreted by different observers who were blinded to each other. Subsequently the diagnosis of DVT was made by consensus.
Results: 13 out of 20 patients had DVT with thrombus seen in 46 venous segments. 17 thrombosed venous segments were present in those patients who were subjected to helical CTV while 29 thrombosed venous segments were present in those patients who underwent axial CTV. Helical CTV diagnosed thrombosis in 14 out of 17 thrombosed venous segments while axial CTV diagnosed thrombosis in 28 out of 29 thrombosed venous segments. Helical CTV was found to have a sensitivity of 82.4% and specificity of 99.1%. False positive and false negative rates were 6.7% and 2.7%, respectively. Axial CTV was found to have a sensitivity of 96.6% and specificity of 100%. False positive and false negative rates were 0% and 0.98%, respectively. Excellent to fair opacification of the deep venous system was seen in 8 cases on helical CTV and in 9 cases on axial CTV. Fair to poor opacification was present in 2 cases on helical CTV and in 1 case on axial CTV. Mean radiation dose was 1,153.57mgy.cm for helical mode and 806.53 mgy.cm for axial mode. This difference was statistically significant. Color Doppler did not detect DVT in one patient who had a thrombus in the duplicated limb of the popliteal vein.
Conclusion: Indirect CTV with CTPA is a one stop examination for evaluating patients with PTE. Multidetector helical CTV does not offer any advantage over axial CTV and gives more radiation dose to the patient.
* Will present paper
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
