Automated analysis of intima-media thickness: analysis and performance of CARES 3.0.

OBJECTIVES: In recent years, the use of computer-based techniques has been advocated to improve intima-media thickness (IMT) quantification and its reproducibility. The purpose of this study was to test the diagnostic performance of a new IMT automated algorithm, CARES 3.0, which is a patented class of IMT measurement systems called AtheroEdge (AtheroPoint, LLC, Roseville, CA). METHODS: From 2 different institutions, we analyzed the carotid arteries of 250 patients. The automated CARES 3.0 algorithm was tested versus 2 other automated algorithms, 1 semiautomated algorithm, and a reader reference to assess the IMT measurements. Bland-Altman analysis, regression analysis, and the Student t test were performed. RESULTS: CARES 3.0 showed an IMT measurement bias ± SD of -0.022 ± 0.288 mm compared with the expert reader. The average IMT by CARES 3.0 was 0.852 ± 0.248 mm, and that of the reader was 0.872 ± 0.325 mm. In the Bland-Altman plots, the CARES 3.0 IMT measurements showed accurate values, with about 80% of the images having an IMT measurement bias ranging between -50% and +50%. These values were better than those of the previous CARES releases and the semiautomated algorithm. Regression analysis showed that, among all techniques, the best t value was between CARES 3.0 and the reader. CONCLUSIONS: We have developed an improved fully automated technique for carotid IMT measurement on longitudinal ultrasound images. This new version, called CARES 3.0, consists of a new heuristic for lumen-intima and media-adventitia detection, which showed high accuracy and reproducibility for IMT measurement.

Prostate tissue characterization/classification in 144 patient population using wavelet and higher order spectra features from transrectal ultrasound images.

In this work, we have proposed an on-line computer-aided diagnostic system called "UroImage" that classifies a Transrectal Ultrasound (TRUS) image into cancerous or non-cancerous with the help of non-linear Higher Order Spectra (HOS) features and Discrete Wavelet Transform (DWT) coefficients. The UroImage system consists of an on-line system where five significant features (one DWT-based feature and four HOS-based features) are extracted from the test image. These on-line features are transformed by the classifier parameters obtained using the training dataset to determine the class. We trained and tested six classifiers. The dataset used for evaluation had 144 TRUS images which were split into training and testing sets. Three-fold and ten-fold cross-validation protocols were adopted for training and estimating the accuracy of the classifiers. The ground truth used for training was obtained using the biopsy results. Among the six classifiers, using 10-fold cross-validation technique, Support Vector Machine and Fuzzy Sugeno classifiers presented the best classification accuracy of 97.9% with equally high values for sensitivity, specificity and positive predictive value. Our proposed automated system, which achieved more than 95% values for all the performance measures, can be an adjunct tool to provide an initial diagnosis for the identification of patients with prostate cancer. The technique, however, is limited by the limitations of 2D ultrasound guided biopsy, and we intend to improve our technique by using 3D TRUS images in the future.

Association between carotid artery plaque volume, composition, and ulceration: a retrospective assessment with MDCT.

OBJECTIVE: The purpose of this study was to evaluate the relationship between MDCT angiography-assessed carotid artery plaque volume and composition and the presence of ulceration. MATERIALS AND METHODS: Seventy consecutive patients (46 men and 24 women; mean age, 63 years; age range, 42-81 years) were studied using MDCT and were retrospectively analyzed. Component types of the carotid plaque were defined according to attenuation value ranges (lipid, < 60 HU; fibrous tissue, 60-130 HU; and calcification, > 130 HU). The plaque volumes of each component were calculated. Statistical analysis was performed using the receiver operating characteristic (ROC) statistic and Wilcoxon signed rank test to evaluate the association between the presence of ulceration and specific plaque components and their volume. RESULTS: Eighteen carotid arteries were excluded, and 16 ulcerated plaques were detected in the remaining 122 carotid arteries. Wilcoxon and ROC curve analysis showed a statistically significant association between increased relative lipid volume and ulceration (p = 0.0001; area under the ROC curve, 0.916). The total volume of the plaque did not show an association with the presence of ulceration (p = 0.0526). CONCLUSION: The results of our retrospective study suggest that there is no correlation between total carotid atherosclerotic plaque volume and ulcerations, whereas plaque relative lipid volume (using attenuation of < 60 HU) is associated with the presence of ulceration. This finding could indicate vulnerable plaques and increased risk for cerebrovascular events.

Analysis of deep inferior epigastric perforator (DIEP) arteries by using MDCTA: Comparison between 2 post-processing techniques.

PURPOSE: Our purpose was to compare two post-processing techniques, Maximum-Intensity-Projection (MIP) and Volume Rendering (VR) for the study of perforator arteries. METHODS: Thirty patients who underwent Multi-Detector-Row CT Angiography (MDCTA) between February 2010 and May 2010 were retrospectively analyzed. For each patient and for each reconstruction method, the image quality was evaluated and the inter- and intra-observer agreement was calculated according to the Cohen statistics. The Hounsfield Unit (HU) value in the common femoral artery was quantified and the correlation (Pearson Statistic) between image quality and HU value was explored. RESULTS: The Pearson r between the right and left common femoral artery was excellent (r=0.955). The highest image quality score was obtained using MIP for both observers (total value 75, with a mean value 2.67 for observer 1 and total value of 79 and a mean value of 2.82 for observer 2). The highest agreement between the two observers was detected using the MIP protocol with a Cohen kappa value of 0.856. The ROC area under the curve (Az) for the VR is 0.786 (0.086 SD; p value=0.0009) whereas the ROC area under the curve (Az) for the MIP is 0.0928 (0.051 SD; p value=0.0001). CONCLUSION: MIP showed the optimal inter- and intra-observer agreement and the highest quality scores and therefore should be used as post-processing techniques in the analysis of perforating arteries.

Carotid artery stenosis at MSCT: is there a threshold in millimeters that determines clinical significance?

PURPOSE: The purpose of this work was to determine whether it is possible to identify a reliable carotid stenosis threshold--measured in millimeters (mm)--that is associated with cerebrovascular symptoms. METHODS: Written, informed consent was obtained for each patient; 149 consecutive patients (98 men; median age, 68 years) were studied for suspected pathology of the carotid arteries by using MDCTA. In each patient, carotid artery stenosis was quantified using the mm-method. Continuous data were described as the mean value ± standard deviation (SD), and they were compared by using the Student's t test. A ROC curve was calculated to test the study hypothesis and identify a specific mm-stenosis threshold. Logistic regression analysis was performed to include other MDCTA findings, such as plaque type and ulcerations. A P value < 0.05 was considered to indicate statistical significance. RESULTS: Twenty-six patients were excluded. Of those remaining, 75 patients suffered cerebrovascular symptoms (61%). There was a statistically significant difference (P = 0.0046) in the mm-carotid stenosis between patients with symptoms (1.31 ± 0.64 mm SD) and without symptoms (1.68 ± 0.79 mm SD). Multiple logistic regression analysis confirmed that symptoms were associated with increased luminal stenosis (P = 0.013) and with the presence of fatty plaques (P = 0.0491). Moreover, the ROC curve (Az = 0.669; ±0.051 SD; P = 0.0009) indicated that a threshold of 1.6 mm stenosis was associated with a sensitivity to symptoms of 76%. CONCLUSIONS: The results of our study suggest an association between luminal stenosis (measure in mm) and the presence of cerebrovascular symptoms. Luminal stenosis of 1.6 mm is associated, with a sensitivity of 76%, with cerebrovascular symptoms.

Imaging of the carotid artery.

In the study of carotid arteries, modern techniques of imaging allow to analyze various alterations beyond simple luminal narrowing, including the morphology of atherosclerotic plaques, the arterial wall and the surrounding structures. By using CTA and MRI it is possible to obtain three-dimensional rendering of anatomic structures with excellent detail for treatment planning. This paper will detail the role of various imaging methods for the assessment of carotid artery pathology with emphasis on the detection, analysis and characterization of carotid atherosclerosis.

Comparison between manual and automated analysis for the quantification of carotid wall by using sonography. A validation study with CT.

PURPOSE: The purpose of this paper was to compare manual and automated analysis for the quantification of carotid wall obtained with sonography by using the computed tomography as validation technique. MATERIAL AND METHODS: 21 consecutive patients underwent MDCTA and ultrasound analysis of carotid arteries (mean age 68 years; age range 59-81 years). The intima-media-thickness (IMT) of the 42 carotids was measured with novel and dedicated automated software analysis (called AtheroEdge™, Biomedical Technologies, Denver, CO, USA) and by four observers that manually calculated the IMT. The carotid artery wall thickness (CAWT) was also quantified in the CT datasets. Bland-Altman statistics was employed to measure the agreement between methods. A Student's t-test was used to test the differences between the IMT values of AtheroEdge™. The study obtained the IRB approval. RESULTS: The correlation between automated AtheroEdge™ measurements and those of the human experts were equal to 95.5%, 73.5%, 88.9%, and 81.7%. The IMT coefficient of variation of the human experts was equal to 11.9%. By using a Student's t-test, the differences between the IMT values of AtheroEdge™ and those of the human experts were not found statistically significant (p value=0.02). On comparing AtheroEdge™ (using Ultrasound) with CAWT (using CT), the results suggested a very good concordance of 84.96%. CONCLUSIONS: Data of this preliminary study indicate that automated software AtheroEdge™ can analyze with precision the IMT of carotid arteries and that the concordance with CT is optimal.

Stenosis Asymmetry Index (SAI) between symptomatic and asymptomatic patients in the analysis of carotid arteries. A study using CT angiography.

PURPOSE: Extracranial carotid artery stenosis is accepted as a significant risk factor for cerebrovascular events. The purpose of this paper was to evaluate whether the Stenosis Asymmetry Index (SAI) between carotid arteries (in symptomatic and asymptomatic patients) can be considered a further parameter in the stroke risk stratification. MATERIALS AND METHODS: 60 consecutive symptomatic (males 36; median age 64) patients and 60 non symptomatic patients matched for gender and age, were analyzed using a 40-detector-row CT angiography. Each patient was analyzed by injecting 80 mL of contrast material at a 5 mL\s flow rate. Stenosis degree of 240 carotids was calculated according to NASCET method. For each patient, the ratio between the most severe stenosis and the contralateral was calculated to obtain the SAI. Multiple logistic regression analysis was performed and ROC curve was also calculated. RESULTS: Results of our study indicate a mean SAI of 1.48 (± 0.35 SD) in the asymptomatic group and a mean SAI of 1.69 (± 0.53 SD) in the symptomatic group with a statistically significant difference (p value=0.0204). The multiple logistic regression analysis did not find statistically significant association between SAI and symptoms. The ROC curve analysis indicated that an SAI value of 1.8 has a specificity of 84.31% presence of cerebral symptoms whereas using a 1.2 SAI we obtained a sensitivity of 88.24%. CONCLUSION: Results of our study suggest that a SAI>1.8 has a good sensitivity in identifying the association with cerebrovascular events.

MRI and "tenderness guided" transvaginal ultrasonography in the diagnosis of recto-sigmoid endometriosis.

PURPOSE: To compare the diagnostic accuracy of MRI and "tenderness-guided" transvaginal ultrasonography (tg-TVUS) in the identification of recto-sigmoid endometriosis. MATERIALS AND METHODS: Institutional Review Board approval for this study was obtained, and written informed consent was given by all patients. This study is compliant with the STARD (Standards for Reporting of Diagnostic Accuracy) method. Fifty-nine patients (mean age, 33 years; range, 21-44 years) with clinical suspicion of deep pelvic endometriosis were prospectively enrolled. They underwent tg-TVUS and MRI before surgery. The characteristics of the MRI signal were analyzed. Mapping of recto-sigmoid endometriosis was performed and tg-TVUS and MR imaging results were compared with surgical and pathological findings. Sensitivity, specificity, and the positive and negative likelihood ratio (LR+ and LR-) were calculated. Inter-technique concordance was assessed using the Cohen statistic, and receiver operating characteristic (ROC) curves were obtained. Logistic regression analysis was performed. RESULTS: The prevalence of recto-sigmoid endometriosis was 51%. The specificity, sensitivity, and LR+ and LR- were 90%, 73%, 7.089 and 0.297, respectively, for MRI and 86%, 73%, 5.317 and 0.309, respectively, for tg-TVUS. The presence of a high T1 signal spot was an excellent specific finding (100%) but was associated with a low sensitivity (30%). Inter-technique concordance using the Cohen statistic indicated a kappa value of 0.658 (± 0.098 SD). According to the logistic regression equation obtained, the use of both tg-TVUS and MRI allows optimal diagnostic performance. CONCLUSION: MRI and tg-TVUS show similar results in the identification of recto-sigmoid endometriosis. The Cohen kappa value suggests that these methods may have complementary roles in the identification of recto-sigmoid endometriosis, depending on the site affected.

Intima Media Thickness Variability (IMTV) and its association with cerebrovascular events: a novel marker of carotid therosclerosis?

PURPOSE: Increased intima-media thickness (IMT) is an early marker of atherosclerotic disease and several prospective studies have demonstrated that IMT is a strong predictor of cerebrovascular complications. In this paper we propose a novel method to assess IMT, called "intima media thickness variability" (IMTV) and evaluate its relationship with the development of cerebrovascular events. MATERIAL AND METHODS: The study was approved by the local IRB. Twenty consecutive patients underwent ultrasound analysis of the carotid arteries (mean age 68 years; age range 59-81 years). The IMT and IMTV of the 40 carotid vessels was assessed. Bland-Altman statistics were employed to measure the inter-observer variability, and ROC analysis was used to assess the association with cerebrovascular events. RESULTS: In all cases the ROC area under the curve was higher for IMTV than IMT. There was a statistical significant association between IMTV and cerebrovascular events (P=0.018), but no relationship between IMT and events. The results for inter-observer variability showed a systematic error between 0.04 and 0.08 mm. In the Person Rho correlation analysis in no case a statistical association between IMT and symptoms was detected whereas in 2 cases (observer 1 and observer 2), a statistically significant association between IMTV and symptoms was found. CONCLUSIONS: Data of this preliminary study suggest that IMTV may represent a powerful method to assess carotid atherosclerotic disease with a significant association to cerebrovascular events.

The multidetector computed tomography angiography (MDCTA) in the diagnosis of splenic artery aneurysm and pseudoaneurysm.

Splenic artery aneurysm is the most frequent visceral artery aneurysm and rupture of the aneurysm is associated with a high mortality rate. It is important to discriminate between a true aneurysm and a pseudoaneurysm that may be caused by pancreatitis, iatrogenic and postoperative causes, trauma and peptic ulcer disease. Multidetector-row CT angiography (MDCTA) allows detailed visualization of the vascular anatomy and may allow identification of aneurysms and pseudoaneurysms that affect the splenic artery. The objective of this article is to provide a review of the general characteristics of splenic artery aneurysms and pseudoaneurysms and to describe the findings of MDCTA.

Comparison between postprocessing techniques in the analysis of hepatic arteries using multi-detector-row computed tomography angiography.

PURPOSE: Our purpose was to compare 4 different postprocessing techniques (maximum-intensity projection [MIP], multiplanar reconstruction, curved planar reconstruction, and volume rendering [VR]) for the study of hepatic arteries. METHODS: One hundred thirty-seven patients who underwent multi-detector-row computed tomography angiography between August 2009 and January 2010 were retrospectively analyzed. For each patient and for each reconstruction method, the image quality was evaluated and the interobserver and intraobserver agreement was calculated according to Cohen statistics. RESULTS: The Pearson r between the observers for the common hepatic artery measurement (Hounsfield unit) was good (r = 0.88). The VR showed a Cohen κ value of 0.78, and the highest image-quality score was obtained using MIP (total value, 384; mean value, 3.01) for observer 1 and using VR and MIP for observer 2 (mean value of 2.94). CONCLUSIONS: Maximum-intensity projection and VR showed the optimal interobserver and intraobserver agreement and the highest quality scores and therefore should be used as postprocessing techniques when analyzing the hepatic arteries.

Anatomic variations of arterial liver vascularization: an analysis by using MDCTA.

PURPOSE: The knowledge of celiac trunk vascularization and hepatic arteries' configuration is extremely important in several areas like hepato-biliary pancreatic surgery, as well as in interventional radiological treatment. Our purpose was to evaluate the incidence of anatomic variation of arterial liver vascularization by using MDCTA in a large, homogeneous population. METHODS: Between January 2004 and December 2009, 1,910 patients (1,156 men; mean age, 62.7 years) who underwent MDCT were retrospectively analyzed. Hepatic arterial configuration was classified according to Michels' classification. Image quality was graded according to an ordinal scale. RESULTS: Of the 1,910 examined patients, 281 (15, 67%) were excluded because of sub-optimal image quality. Of the remaining 1,629 patients (986 men; mean age, 61.1 years ± 8 (SD); age range, 19-90 years), anatomic variations were detected in 631 patients (38.73%). CONCLUSIONS: MDCT was useful in analyzing arterial liver configuration and demonstrated the presence of a significant incidence of arterial variants (38.73%) in the examined population.

Magnetic resonance imaging of pontine capillary telangectasia.

Capillary telangectasia (CTS) is a vascular malformation, characterized by multiple thin-walled vascular channels, interposed between normal brain parenchyma. It has been hypothesized that CTS is an acquired lesion caused by other underlying venous anomalies but this theory is still debated. The clinical symptoms of CTS are described to be rather mild. In most cases CTS is not detectable by computed tomography and among imaging techniques, nowadays magnetic resonance imaging (MRI) represents the standard of reference in diagnosing this pathology. Usually, by analyzing the different signal characteristics it is possible to perform a differential diagnosis between CTS and the other pathologies that may affect the pons and the use of diffusion-weighted and Gradient-echo sequences increase the MRI specificity.

Learning curve in the detection of ovarian and deep endometriosis by using Magnetic Resonance: comparison with surgical results.

PURPOSE: Determining if Magnetic Resonance Imaging (MRI) accuracy in diagnosing endometriosis is related to radiologist's expertise. METHODS AND MATERIALS: Written informed consent was obtained from all patients. This study is compliant to STARD method. Thirty patients (mean age 34; range 21-45 years) who had undergone MRI study for suspected endometriosis underwent surgery were retrospectively evaluated. MRI at 1.5T was performed with SE and TSE sequences, T1 and T2-weighted with and without fat suppression. Four localizations were analyzed: ovary, uterosacral ligaments (USL), vaginal fornix and Rectum\Sigma\Douglas (R.S.D.). One radiologist evaluated each dataset; sensitivity, specificity, PPV and NPV, accuracy, LR+ and LR- were calculated according to the surgical results (first analysis). Dataset were then re-analyzed 12 months (second analysis) and 24 months (third analysis) later. McNemar test was applied to determine differences between the three analysis. RESULTS: Sensitivity, specificity and accuracy for the ovary at the first analysis were 88.9%, 87% and 88%, at the second 92.6%, 87% and 90% whereas at the third 92.6%, 91.3% and 92%. Sensitivity, specificity and accuracy for the USLs at the first analysis were 62.5%, 76.9% and 70%, at the second 72%, 80.8% and 76% whereas at the third 80%, 84.6% and 82%. Sensitivity, specificity and accuracy for the vaginal fornix at the first analysis were 63.2%, 64.5% and 64%, at the second 73.7%, 77.4% and 76% whereas at the third 73.7%, 83.9% and 80%. Sensitivity, specificity and accuracy for the R.S.D. at the first analysis were 39.1%, 81.5% and 62%, at the second 62.5%, 85.2% and 76% whereas at the third 73.9%, 88.9% and 82%. McNemar test indicated a significant statistical difference in sensitivity in detecting nodules of endometriosis in R.S.D. between first and third analysis (p=0.0215). The mean review time decreased (p=0.0001). CONCLUSIONS: Accuracy of MRI in diagnosing endometriosis increased with radiologist's expertise and the improvement was statistically significant in determining RSD involvement.

Vulnerable plaque: detection of agreement between multi-detector-row CT angiography and US-ECD.

OBJECTIVES: The purpose of this work was to evaluate the agreement between ultra-sound echo-color Doppler (US-ECD) and multi-detector-row CT angiography (MDCTA) in the characterization of vulnerable plaque. METHODS: From January 2004 to January 2007 658 patients who underwent both MDCTA and US-ECD for the study of carotid arteries, were retrospectively evaluated (453 males, 205 females). For all subjects the following parameters were analysed: plaque morphology (regular versus irregular), type of the plaque (fatty, mixed and calcified) and presence of ulcerations. Statistical analysis was performed to calculate concordance between the two techniques employed. RESULTS: In the definition of the type of plaque, the observed agreements were 77.2% and the kappa value was 0.657 (95% confidence interval: 0.615-0.699). The weighted kappa resulted 0.644. In the definition of ulceration plaque, the observed agreements were 88.4% but the kappa value was only 0.325 (95% confidence interval: 0.201-0.449). Agreement observed in the evaluation of plaque morphology was 78.3% with a kappa value of 0.513 (95% confidence interval: 0.452-0.574). CONCLUSION: We observed a good agreement between US-ECD and MDCTA in the assessment of plaque type whereas a poor agreement resulted in the evaluation of plaque ulceration. The use of US-ECD and MDCTA provides different results in the evaluation of plaque. Our results suggest that information deriving from US-ECD should be always critically compared with other diagnostic techniques.

Carotid artery stenosis quantification: concordance analysis between radiologist and semi-automatic computer software by using Multi-Detector-Row CT angiography.

PURPOSE: Carotid artery stenosis quantification is still considered a leading parameter in the choice of the therapeutic option. Our purpose was to asses the concordance between radiologist and a semi-automatic computer software in the stenosis quantification of carotid artery studied by using a Multi-Detector-Row CT angiography (MDCTA). METHODS AND MATERIAL: 45 patients studied by using a 40-detector row CT scanner were retrospectively analyzed. Carotid artery stenosis was quantified by one high experienced radiologist in vessel analysis and by using a dedicated software. Carotid artery stenosis was calculated according to the ECST method. Bland-Altman statistics was used to measure the inter- and intra-concordance between radiologist and software and correlation coefficient between measures were performed by using nonparametric Spearmann correlation statistic. A p value<0.05 was considered to mean statistical significance. RESULTS: A strength correlation according to linear regression (correlation Spearman'ρ coefficient=0.975; p<0.0001) between radiologist and software of vessel analysis was observed. Between first and second stenosis of carotid artery quantification performed by radiologist and software of vessel analysis we observed a Spearman'ρ coefficient=0.943 (p<0.0001) and a Spearman'ρ coefficient=0.9879; (p<0.0001) respectively. CONCLUSIONS: Our results indicated that there is a strength correlation according to linear regression between stenosis of carotid artery quantification performed by radiologist and semi-automatic software. Reproducibility between measurements performed by semi-automatic software are higher compared to radiologist.

Comparison between quantification methods of carotid artery stenosis and computed tomographic angiography.

PURPOSE: The purpose of this study was to compare 3 percentage carotid artery measurement methods (the North American Symptomatic Carotid Endarterectomy Trial [NASCET], the European Carotid Surgery Trial [ECST], and the Carotid Stenosis Index [CSI]) and 1 millimeter method (direct millimeter measurement) to evaluate the difference and correlation between them. MATERIALS AND METHODS: Seven hundred ninety-two patients (591 men; age: mean, 63 years; range, 32-91 years) studied by using a multi-detector row computed tomographic scanner for a total of 1584 carotid arteries were retrospectively analyzed. Each carotid stenosis was measured according to 4 measurement methods (the NASCET, the ECST, the CSI, and the direct millimeter measurement). Carotid arteries with near-occlusion condition were excluded. The Kolmogorov-Smirnov Z test was used to test the normality of continuous variable groups. Comparison of derived ratio-percent methods was performed by using the Bland-Altman plots, and receiver operating characteristic curves were calculated. Correlation coefficients were also calculated by using a nonparametric Spearman correlation. A P < 0.05 was considered to mean statistical significance. RESULTS: Four hundred sixteen carotid arteries were excluded, and in the remaining 1168 ones, a strength correlation according to quadratic regression between the NASCET and ECST methods was observed (Spearman rho coefficient, 0.948; P < 0.0001). An inverse correlation according to linear regression was observed between the NASCET and the direct millimeter measurement (Spearman rho coefficient, -0.972; P < 0.0001); the CSI shows a quadratic regression with the NASCET, a linear regression with the ECST, and an inverse linear regression with the direct millimeter measurement (Spearman rho coefficient, 0.946, 0.932, and -0.939 respectively). The cutoff values for 50% and 70% NASCET stenosis were 2.36 and 1.51, respectively. CONCLUSIONS: Our study results indicate that the direct millimeter measurement of stenosis, by using appropriate equations, can reliably predict NASCET-, ECST-, and CSI-type percent stenoses. The use of direct millimeter measurement may remove the pitfalls and the discrepancies deriving from the use of different ratio-percent methods.

Agreement and reproducibility in identification of endometriosis using magnetic resonance imaging.

BACKGROUND: Magnetic resonance imaging (MRI) is a promising technique in the study of endometriosis, allowing a complete mapping of lesions before surgery. However, the value of MRI in the diagnosis of endometriosis in the bladder, in superficial peritoneal lesions, and in ovarian foci and uterosacral ligaments is still under debate. PURPOSE: To assess inter- and intra-observer agreement in the evaluation of endometriosis in different anatomical locations using MRI. MATERIAL AND METHODS: From June 2006 to February 2008, 83 female patients (mean age 39, range 19-49 years) who had undergone MRI examination for suspected endometriosis were evaluated by two radiologists. MRI at 1.5 Tesla was performed with SE and TSE sequences, T1- and T2-weighted with and without fat suppression. Each examination was completed with gadolinium administration. Each dataset was independently evaluated by the radiologists for the presence or absence of endometriosis. The location (ovaries, uterosacral ligaments (USLs), pouch of Douglas, vagina, rectosigmoid, rectovaginal septum, and bladder) of suspected lesions was recorded. Cohen kappa statistical analysis was performed to calculate agreement between measurements. After 2 months the data were analyzed again by the two observers to assess intra-observer agreement. RESULTS: Of the 83 MRI examinations performed, 12 patients demonstrated no evidence of endometriosis. In the remaining 71 studies, 157 endometriotic lesions ranging in size from 0.4 to 6.2 cm were detected. Of the 157 lesions, 53 (33.75% incidence) were smaller than 1 cm. In the ovaries, the inter-observer agreement was 92.77% and the kappa value was 0.802 (95% CI, 0.695-0.91). In the bladder the inter-observer agreement was 96.39% and the kappa value was 0.553 (95% CI 0.056-1). In the USLs the inter-observer agreement was 90.96% and the kappa value 0.583 (95% CI, 0.381-0.784). In the rectovaginal septum the inter-observer agreement was 94.58% and the kappa value 0.739 (95% CI, 0.572-0.905). In the rectovaginal pouch the inter-observer agreement was 88.55% and the kappa value 0.608 (95% CI, 0.443-0.774). In the vaginal fornix the inter-observer agreement was 94.58% and the kappa value 0.726 (95% CI, 0.552-0.901). In the rectosigmoid the inter-observer agreement was 89.76% and the kappa value 0.589 (95% CI, 0.389-0.768). CONCLUSION: The results of our study indicate that MRI has a high inter- and intra-observer agreement in the identification of endometriosis located in the ovary, rectosigmoid, and rectovaginal septum, whereas the agreement is suboptimal for the identification of endometriosis located in the USLs.