| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||
ABSTRACT |
Thursday, May 10, 10:00 AM–12:00 PM
Abstracts 228-238
Moderator(s): Sanjeev Bhalla and Mylene Truong
10:00 AM
Keynote Address: Non-thromboembolic Pulmonary Embolism
Sanjeev Bhalla, Mallinckrodt Institute of Radiology, St. Louis, MO
10:10 AM
228. CT Pulmonary Angiogram: Clot Burden Index and 12-Month Mortality
Mandrekar J.1; Damon B.2; Subramaniam R.2* 1. Mayo Clinic, Rochester, MN; 2. Waikato Hospital, Hamilton, New Zealand
Address correspondence to R. Subramaniam (subramaniam.rathan{at}mayo.edu)
Objective: To establish the prognostic impact of CT pulmonary angiogram clot burden index and 12 month mortality.
Materials and Methods: 535 consecutive patients (Emergency Department and Hospital in-patients) who had CT pulmonary angiogram were recruited prospectively from March 2003 to October 2004. A total of 12 patients were excluded from the study for incomplete follow-up data (8) or indeterminate CT pulmonary angiogram (4). Hence the study population was 523 patients. 12-month follow-up was completed in October 2005. Each CTPA examination was read by two consultant radiologists independently. There were 105 patients with positive CT pulmonary angiogram examinations. Clot burden index was calculated in all of these patients. A 12-month post CTPA follow-up was done in all patients by telephone interview, interview with General Practitioners, hospital records and post mortem reports.
Results: The inter-rater agreement of calculation of clot burden score between the two radiologists was moderate (kappa 0.72). There were 13 (12.5%) patients died at 12-month follow-up among the PE-positive cohort. The mean clot burden index for those patients who were alive was 9.58 (SD 6.80) and for those died was 8.84 (SD 6.27) which was not significant (p = 0.73), among the PE positive cohort.
Conclusion: There was no association between the clot burden index and 12-month all-cause mortality in patients with PE.
229. Utility of Hounsfield Unit Measurements in Diagnosing Pulmonary Embolism
Irani N.*; Dash N.; Kim C.; Ilkhanipour K.; Lupetin A. Allegheny General Hospital, Pittsburgh, PA
Address correspondence to N. Irani (nirani{at}wpahs.org)
Objective: One out of ten CT scans for pulmonary embolism are reported to be equivocal or of suboptimal quality. We explored the feasibility of using Hounsfield units to distinguish artifacts from lesions concordantly read as pulmonary embolism (PE) on CT pulmonary angiography.
Materials and Methods: This prospective investigation was performed on 107 scans done during the 51-day study period. Filling defects read as positive for PE by four radiologists were compared with equivocal lesions thought to represent PE by only one of the four readers. Hounsfield unit (HU) measurements were made using a region of interest (ROI) including the filling defect on three consecutive slices for each of these two classes of lesions. The average of these three measurements obtained from equivocal lesions was compared with measurements obtained from filling defects read by all observers as PE.
Results: Twenty-five lesions were identified as consensus PE in
this series and had average HU = 49.9 ±18.5. Six
lesion were considered equivocal for PE with average HU = 121.2
± 35.7. The difference between these two groups of lesions was
very significant (p < 0.001). Interobserver variability among our
readers on positivity of scans was 17% 
= 0.77).
Conclusion: A filling defect with average Hounsfield units < 80 which otherwise fulfills visual criteria for PE is more likely to represent pulmonary embolism for multiple readers than artifacts which usually have average Hounsfield units >80.
230. Assessment of Second Generation Computer-Aided-Detection Software for the Detection of Pulmonary Embolism at CT Angiography
Walsham A.*; Ng Y.; Patsios D.; Kashani H.; Colak E.; Roberts H. University Health Network, Toronto, Canada
Address correspondence to A. Walsham (Anna.Walsham{at}uhn.on.ca)
Objective: To evaluate a second generation computer-aided detection (CAD) tool for the detection of intravascular filling defects in pulmonary computed tomography angiography (CTA), and to assess its benefit for readers of different levels of experience.
Materials and Methods: 100 consecutive CTA exams performed to diagnose or exclude pulmonary embolism (PE) were retrospectively evaluated by a staff chest radiologist for the presence and location of intravascular filling defects, serving as the reference standard. Subsequently, exams were analyzed using commercially available 2nd generation CAD software (ImageChecker CT, version 2.1 R2 Technology Inc., Sunnyvale, CA). The staff radiologist assessed all CAD marks and classified them as true positive (TP): CAD mark representing a filling defect; false positive (FP): CAD mark that was not a filling defect; false negative (FN): filling defect not marked by CAD. On a case-basis, a true negative (TN) was defined as a negative exam without CAD marks. Additionally, the CT scans were interpreted by three readers of different experience, both without and then with CAD, noting the time for reading, the diagnosis and the confidence level.
Results: 22 of the 100 exams were judged to be positive for PE by the staff radiologist. In all 22 of these cases, CAD correctly marked filling defects (22 TP, 0 FN exams). Of the 78 exams judged negative for PE, 16 had no CAD marks (TN) and 62 had at least one CAD mark (FP). On a case-basis, CAD sensitivity and negative predictive value were 100% (22/22 and 16/16, respectively), specificity 21% (16/78), and positive predictive value 26% (22/84). In the 22 PE positive cases, 70 individual PE were identified by the staff radiologist, of which 55 were correctly marked by CAD (sensitivity 76%). Analyzing individual marks, CAD generated a total of 259 FP marks (average 2.6/case) with a PPV of 18% (55/314). Many FP CAD marks (>40%) were in perihilar locations and easily dismissible. For the most inexperienced reader, 9 PE not detected on initial review were correctly identified following CAD input, in cases that had been correctly identified as positive for PE, overall not changing the interpretation. For more experienced readers, CAD did not alter the assessment for PE.
Conclusion: This 2nd generation CAD software is improved in sensitivity, and in the number and position of FP marks. CAD has a high negative predictive value on a case basis. It may have a useful role as a second reader for inexperienced interpreters.
231. Automated Calculation of Diameter, Area, and Volume of Lung Metastases with MDCT: Influence on Therapeutic Management
Pauls S.2*; Kuerschner C.2; Dharaiya E.1; Muche R.2; Schmidt S. A.2; Krueger S.2; Brambs H.2; Aschoff A. J.2 1. CT Clinical Science, Philips Medical Systems, Cleveland, OH; 2. University of Ulm, Ulm, Germany
Address correspondence to S. Pauls (sandra.pauls{at}uniklinik-ulm.de)
Objective: The goal of this study was to evaluate the influence of automated measurement of diameter, area, and volume from chest CT scans on therapeutic decisions of lung nodules as compared to manual 2D measurements.
Materials and Methods: The study involved 25 patients with 75 lung metastases. Contrast-enhanced CT scans (16 row) of the lung were performed three times during chemotherapy with a mean time interval of 67.9 days between scans. In each patient, 3 metastases were evaluated (n = 225). Automatic mea-surements were compared to manual assessment for the following parameters: diameter, area, and density. The influence on the therapeutic decisions was evaluated using the RECIST criteria.
Results: The maximum diameter measured by the automatic application was on average 27% (SD 39; CI 0.22-0.32; p < 0.0001) higher than the maximum diameter with manual assessment, and the differences depended on metastases size. Based on diameter calculation, manual and automated assessment disagreed in up to 32% of therapeutic decisions. Volumetric assessment tended towards more changes in therapy as compared to diameter calculation. The calculation of mean transversal area of metastases was 36% (SD 0.305; CI - 0.40 to -0.32; p < 0.0001) less with automated measurement. Therapeutic strategy would be changed in up to 25.7% of nodules using automated area calculation. Automated assessment of nodules' area and volume could influence the therapeutic decisions in up to 51.4% of all nodules. Density of the nodules was not validated to determine the influence on therapeutic decisions.
Conclusion: Automatic calculation of nodule parameters can significantly influence the therapeutic decisions.
232. Computer-aided Detection (CAD) of Pulmonary Nodules: Effect of Reconstruction Algorithm and Slice Thickness on CAD Performance
Bayanati H.*; Sitartchouk I.; Patsios D.; Pereira A.; Dong Z.; Kashani H.; Paul N.; Roberts H. University Health Network, Toronto, Canada
Address correspondence to H. Bayanati (Hamid.Bayanati{at}uhn.on.ca)
Objective: To evaluate the effect of slice thickness and reconstruction algorithm on the performance of computer-aided detection (CAD) software for pulmonary nodule detection.
Materials and Methods: Computed tomography (CT) scans of 50 patients with lung cancer or pulmonary metastases were prospectively reconstructed at 1 mm and 5 mm, both with lung and soft tissue algorithms, resulting in 4 series of images. These were read by two chest radiologists, and analyzed using a commercially available CAD software (LungCAR 1.2, Medicsight plc, London, UK). All CAD marks were reassessed and assigned as true positive TP (solid nodule found by CAD and confirmed by the radiologist) or false positive FP (any CAD mark that was not a pulmonary nodule). Any nodule not marked by CAD was regarded as false negative (FN). Sensitivity was computed for all nodules, nodules smaller than 5 mm and nodules between 5 mm and 15 mm, and nodule larger than 15 mm regardless of their density. The McNemar test has been applied to assess significance of differences between the series.
Results: Overall, 472 lung nodules were found by radiologists
and/or CAD in the different CT series and used as reference standard.
Two-hundred and fifty-six nodules were smaller than 5 mm. Thirty-seven were
not seen during the radiologists' read but were found by CAD in at least one
of the series, all were smaller than 5 mm. Overall, 415 nodules were
parenchymal and 57 subpleural; 298 were in the right lung, 174 in the left;
246 were in the upper lobes, middle lobe or lingula, 226 in the lower lobes.
CAD sensitivity was highest in the 1-mm soft tissue series, for all nodules
(83.1%), for nodules smaller than 5 mm (94.9%), for nodules between 5 mm and
15 mm (67.9%) and for nodules larger than 15 mm (75.9%). CAD sensitivity was
lowest in the 5-mm reconstructed series. The sensitivity in detecting 5 mm to
15 mm nodules in the 1-mm soft tissue set was statistically significantly
higher than in any other series (p 
0.0001). The rate of false
positives was highest in the 1-mm soft tissue series (7.5 FP/case), compared
to an average of 3.4 to 3.9 FP in the other series. The majority of FPs were
vessels (45.5%) and easily dismissible.
Conclusion: CAD performance is influenced by the reconstruction algorithm and the slice thickness. It is important to understand these interactions when using CAD in clinical routine. In this study 1-mm slice thickness reconstructed with a soft tissue algorithm proved to be the most sensitive set of images to be analyzed on CAD.
233. Detection Sensitivity of a Commercial Lung Nodule CAD System in a Series of Pathologically Proven Lung Cancers
Godoy M. C.2*; Mayo J. R.3; Cooperberg P. L.2; Maizlin Z. V.2; McWilliams A.1; Yuan R.3; Lam S.1 1. British Columbia Cancer Agency, Vancouver, Canada; 2. Saint Paul's Hospital-UBC, Vancouver, Canada; 3. Vancouver General Hospital-UBC, Vancouver, Canada
Address correspondence to M. Godoy (migbarco{at}gmail.com)
Objective: To evaluate the performance of a commercially available computer-aided detection (CAD) system in a series of pathologically proven lung cancers.
Materials and Methods: Sixty-nine chest computed tomography (CT) scans obtained in 12 subjects (8 women, 4 men, age 51-75 years, mean 63 years) with 15 pathologically proven lung cancers were retrospectively selected from 2156 entry and follow-up CT scans from a lung cancer screening program. CT scans were retrospectively analyzed using a commercially available CAD system for detecting lung nodules.
Results: When first detectable proven lung cancer nodules ranged in maximum diameter from 3 to 38 mm (10.4 ± 9.2 mm) with CAD detection sensitivity stratified by size: 0/2 (0%) = 3 mm, 4/7 (57.1%) 4-10mm, 2/3(66.7%) 11-15 mm, 0/0 16-20 mm, 3/3 (100%) >20 mm, overall sensitivity 9/15 (60%). The sensitivity for all CT scans (first detectable and follow-up), stratified by nodule size as above, were respectively 0/2, 18/25, 24/28, 6/9, 5/5, overall 53/69 (76.8%). Excluding nodules <4 mm and pure ground glass nodules, the sensitivity for all CT scans by size was 18/24 (75%) 4-10 mm, 21/22 (95.4%) 11-15 mm, 6/6 (100%) 16-20 mm, 5/5 (100%) >20 mm, overall 50/57 (87.7%). At resection (13) or biopsy (2) nodules were: adenocarcinoma (10), squamous cell carcinoma (3), small cell carcinoma (2).
Conclusion: The CAD system showed good sensitivity for solid and semisolid cancers equal or larger than 4 mm (sensitivity 87.7%) and excellent for those equal or larger than 11 mm (sensitivity >95.4%).
234. Evaluation of Inefficiencies in Lung Cancer Management in a Large Community Teaching Hospital and Recommendations for an Expedited Program
Shirkhoda A.*; Taberski N.; Fulton M. William Beaumont Hospital, Royal Oak, MI
Address correspondence to A. Shirkhoda (ashirkhoda{at}beaumont.edu)
Objective: The purpose of the study was to identify inefficiencies during the management of lung cancer in our 1000+ bed teaching hospital and assess their clinical implications as a result of potential delays in diagnosis and treatment planning. The ultimate goal of the study is to maximize reduction of the time from initial discovery of lung tumor to surgery.
Materials and Methods: One hundred patients with newly diagnosed, pathology-proven non-small cell lung cancers were analyzed as they were through routine steps of discovery, diagnosis, staging and surgery in 2005. Five time periods were established to analyze each patient from entry date into the hospital system to the date of surgery. Forty-five patients met all criteria and were evaluated for time intervals between various clinical, radiological, surgical and pathological values. The causes of delays between each value were identified and analyzed.
Results: One hundred percent (n = 45) of patients had their first CT within one month of abnormal chest film. 44% had a delay of 28-240 days from first CT to CT biopsy and 28-207 days from CT to PET. In 24% time from PET to CT biopsy was 28-157 days, in 15% between CT biopsy to surgical consult was 28-100 days and in 57% there was 28-467 days from surgical consult to surgery. Pathology turnaround time was 1-7 days.
Conclusion: The results of this retrospective analysis suggest a wide variation in the cause of delay from an abnormal chest X ray to surgery. Identification of these causes has been useful in expediting inefficiencies in diagnosis and management of lung cancer. As the result of this study and our recommendation to the cancer committee, a multidisciplinary pulmonary nodule clinic is being established at our institution.
235. Radiation Therapy for Stage III Lung Carcinomas
Hasezawa K.*; Fujisawa H.; Kusihashi T.; Nakajima H. Syowa University Northern Yokohama Hospital, Yokohama, Japan
Address correspondence to K. Hasezawa (hasezawa{at}med.showa-u.ac.jp)
Objective: With the progress of stereotactic radiotherapy (SRT) or other methods lung carcinomas with earlier stages can be controlled well. However, in advanced cases, survival rates are comparatively low with combination of any methods. Therefore we tried to control Stage III non small cell lung carcinomas (NSCLC) with new protocol of high dose concentration and step by step shrinking RT method.
Materials and Methods: From June 2002 to December 2003, we treated thirteen cases of Stage III NSCLC with new protocol of RT. This method we made three times CT planning for shrinking irradiation region with the progress of RT, and total RT dose reached to 75 gray. We analyzed these cases from the point of survival, patient status, and radiation morbidities, etc. Main characteristics of this therapy is making small high dose region and comparatively large low dose region. And we spare spinal cord, vertebra and esophagus from the irradiation field as far as possible. With the combination methods of three dimensional conformal and multi-time shrinking field techniques, dose area over twenty-five gray were diminished about fifty percent compared with normal opposite two port RT. Chemotherapy were performed for 61.5% patients.
Results: Disease-free three-year survival rate from RT is 69.2%, and survival rate is 76.9%. All survival patients spend their life at home, not in hospital. Complete remission (CR) rate is 84.6%. No meaning difference between with or without chemotherapy. Nevertheless of high dose irradiation, no severe side effects occurred. Radiation morbidities of RTOG score grade 0, I or II were observed at skin, lung and WBC.
Conclusion: Metastasis may decide mainly the survival period of cancer patients, and therefore RT rarely performed for advanced stage patients as a curative therapy. But these results suggest that main tumor and broad area lymph nodes control may prevent metastasis. Observation period was rather short because our hospital is a newly constructed one. But considering that no severe damage was observed and most patients need no hospitalization, this RT method seems safety and quite effective for the tumor control and may contribute on survival of lung NSCLC patients.
236. Increasing Incidence of Cavitation of Pulmonary Metastases in Colon Cancer: Possible Association with Targeted Anti-tumor Therapies
Balcombe J. N.*; Rosen M. A.; Hewitt M.; Gambino P. F.; Sun W.; O'Dwyer P. J.; Torigian D. The Hospital of the University of Pennsylvania, Philadelphia, PA
Address correspondence to J. Balcombe (jonathan.balcombe{at}uphs.upenn.edu)
Objective: To investigate increased observance of cavitation of pulmonary metastases in patients with colon cancer, and to determine the potential relationship between tumor cavitation and the introduction of novel targeted anti-tumor therapies recently approved for treatment of metastatic colon cancer.
Materials and Methods: Departmental records were accessed by ICD9 code to identify the number of patients with colon cancer referred for chest CT from January, 1999 through June, 2006. A keyword search for "colon" and "cavitation" for all chest CT reports for this time period was performed. For reports meeting these criteria, CT images were evaluated to confirm findings compatible with metastatic colon cancer to the lungs. Medical records for patients with confirmed cavitary metastases were reviewed to identify potential causative agents.
Results: Between January, 2003 and June, 2006, 333 patients with colon cancer were referred for chest CT, and twenty (6.0%) were identified with at least one cavitary metastatic lesion. Between January, 1999 and December, 2002, 127 patients with colon cancer underwent chest CT, and two (1.6%) had cavitary metastases. Eight patients whose cavitary lesions were present on the first available CT study were excluded from further analysis. Medical record evaluation has been completed on 6 of the remaining 14 patients. In 5/6 patients, treatment with a targeted biological agent (bevacizumab [n = 2] or cetuximab [n = 3]) was initiated prior to the development of cavitation. The median time interval between start of targeted therapy and development of cavitation was 2.5 months (range 1-9 months). In 4/6 cases, the targeted agent was the only newly introduced therapeutic agent in that time period.
Conclusion: Although cavitation of lung metastases has previously been considered to be an uncommon event, the incidence of cavitation of pulmonary metastases from colon cancer has increased approximately four-fold since 2003, roughly correlating to the introduction of targeted therapies (cetuximab and bevacizumab) for colon cancer treatment. Our review suggests a potential relationship between the lesion cavitation and treatment with targeted chemotherapy. If supported by future studies, these findings may aid radiologists in recognizing cavitation as a manifestation of tumor response to targeted chemotherapeutic agents in colon cancer.
237. Comparison of PET/CT and MDCT for Diagnosis of Pleural Metastasis in Patients with Lung Cancer
Gun P.*; Jin G.; Lee Y.; Sohn M.; Han Y. Chonbuk National University, Medical School, Jeonju, South Korea
Address correspondence to P. Gun (greem{at}freechal.com)
Objective: To compare diagnostic accuracy of PET/CT with that of MDCT for detection and differentiation of pleural metastasis in patients with lung cancer.
Materials and Methods: This study was retrospective analysis of eligible 107 patients of lung cancer who underwent PET/CT and contrast-enhanced chest MDCT. 25 patients had cytologic examination of pleural fluid or pleural biopsy because they have pleural effusion or thickening and 16 of 25 patients proved as pleural metastasis. Retrospective interpretation of CT images of 25 patients was made by consensus of two radiologists. Interpretation of PET/CT was made by two nuclear medicine physicians. Criteria for pleural metastasis on CT image were circumferential pleural thickening, enhancing nodule within pleural effusion, >1 cm in pleural thickness, mediastinal pleural nodule. Pleural metastasis on PET/CT was defined as area of increased 18F-FDG activity, of intensity higher than that of surrounding tissues, not related to normal physiologic 18F-FDG uptake. Statistical analysis of data was performed by use of chi square test.
Results: 11 of 25 (44%) pleural metastases were detected on PET/CT and 10 of 25 (40%) pleural metastases were detected on MDCT. The accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of PET/CT were 72%, 68.8%, 77.8%, 84.6%, and 58.3%. The accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of MDCT were 64%, 62.5%, 66.7%, 76.9%, and 50%.
Conclusion: PET/CT is slightly superior to MDCT for diagnosis of pleural metastasis in patients with lung cancer. Especially, PET/CT is useful to detect a small pleural nodule with large pleural effusion.
238. Laser Guidance in Interventional CT: Technique Description and Assessment of Benefits
Pereira A. M.*; Paul N.; Mehdizadeh-Kashani H.; Patsios D.; Doyle D. University Health Network/Mount Sinai Hospital, Toronto, Canada
Address correspondence to A. Pereira (andre.pereira{at}uhn.on.ca)
Objective: To evaluate the utility of a low-cost laser guidance device in CT-guided percutaneous lung biopsies, with a detailed description of the technique.
Materials and Methods: Two hundred patients (100M, 100F), were enrolled into this study. All patients were referred for a CT-guided percutaneous lung biopsy. After positioning the patient, we used a tripod mounted laser level with a beam splitter to accurately reproduce the angle of needle entry into the chest wall as determined from the axial planning soft copy CT image. The patients were analyzed in 2 groups: Group A, 100 consecutive CT-guided lung biopsies using laser guidance and Group B, 100 consecutive CT-guided lung biopsies without laser guidance. We documented the number of repeat CT acquisitions for needle position verification prior to crossing the parietal pleura, the procedure time and the experience of the operator.
Results: The number of needle position verifications was as follows: Group A: a mean of 1.48 CT acquisitions (1-4); 1 acquisition in 67%, 2 acquisitions in 22% and 3 or more acquisitions in 11% of patients, Group B: a mean of 2.11 CT acquisitions (1-12); 1 acquisition in 49%, 2 acquisitions in 25% and 3 or more acquisitions in 26% of patients. Non-square angles were easily achieved and the procedure time in Group A patients was significantly reduced. The greatest incremental benefit gained by using the laser level was to relatively inexperienced radiology residents.
Conclusion: Laser guidance is a valuable and inexpensive tool in CT-guided lung biopsies, decreasing the total number of CT acquisitions for needle position verification and resulting in a shorter procedure time and a reduction in radiation dose to the patient. Radiologists in training gained the most benefit from laser guidance.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
