B-0215 Ultrasound elastography in the diagnostic assessment of axillary lymph nodes in women presenting to a breast imaging centre
L. Sim, L. Leong | Thursday, March 7, 14:00 – 15:30 / Room F2
Purpose: To evaluate the performance of elastography in distinguishing benign and metastatic axillary lymph nodes.
Methods and Materials: 67 women with 72 sonographically visible axillary lymph nodes undergoing biopsy at our breast imaging centre were evaluated independently with conventional ultrasound, elastography and combined ultrasound and elastography (CEUS). The elastogram was classified as benign or malignant, based on the strain pattern, the length and area ratios of the lesion seen on elastography versus ultrasound. Validation of radiological diagnosis was by histopathology. The sensitivity, specificity, PPV, NPV and accuracy of each test were compared individually and with CEUS. To obtain a parameter for diagnostic performance, ROC curves were plotted.
Results: Of the 72 axillary lymph nodes biopsied, 33 had metastases and 39 were benign. The sensitivity, specificity and accuracy of conventional ultrasound were 93.9%, 30.8% and 59.7%, respectively. The sensitivity, specificity and accuracy of elastography were 93.9%, 97.4% and 95.8%, respectively, and that of combined ultrasound and elastography were 97 %, 92.3% and 94.4%, respectively. The sensitivities of all 3 tests were similar but the specificity and accuracy obtained by elastography and CEUS were significantly better than conventional ultrasound (P<0.0005). Elastography correctly diagnosed 96 % of histologically benign lymph nodes which were deemed malignant on ultrasound.
Conclusion: The use of elastography alone or combined with ultrasound has a higher specificity and accuracy than conventional ultrasound in evaluating axillary lymph nodes. Given the high specificity of elastography, biopsy could have been avoided in 96 % of cases classified as malignant on ultrasound but benign on elastography.
B-0120 Determining the vulnerable plaque: correlation between 18F-FDG PET and dynamic contrast-enhanced MRI in atherosclerotic plaques of symptomatic patients
M.T.B. Truijman, R.M. Kwee, R.H.M. van Hoof, R.J. van Oostenbrugge, W.H. Mess, J.E. Wildberger, W.H. Backes, J.A. Bucerius, M.E. Kooi | Thursday, March 7, 10:30 – 12:00 / Room N/O
Purpose: Identifying vulnerable atherosclerotic plaques in symptomatic patients with moderate (30-69%) carotid artery stenosis can contribute to clinical decision making. Hallmarks of plaque vulnerability are inflammation and increased neovascularisation. Inflammation can be assessed with 18F-FDG PET, while neovascularisation can be quantified with dynamic contrast-enhanced (DCE) MRI. We aimed to investigate the correlation between inflammation as assessed by18F-FDG PET and neovascularisation as assessed by DCE-MRI.
Methods and Materials: Fifty-eight patients with transient ischaemic attack (TIA) or minor stroke in the carotid territory and ipsilateral carotid plaque causing a moderate stenosis were included. All patients underwent 1.5 T multi-sequence MR imaging. Quantification of neovascularisation was done using a custom-made Matlab program which calculates Ktrans. A 3D PET-CT scan was performed on all patients one hour after injection of 2.75 MBq/kg body weight 18F-FDG. Dedicated fusion software was used to calculate mean blood-normalised 18F-FDG standard uptake values (SUV) of the plaque.
Results: Of the 58 patients, 9 were excluded due to poor image quality of the DCE-MRI. In total, we analysed 49 patients. The mean Ktrans and mean normalised SUV were 0.110 (±0.027) and 1.446 (±0.255), respectively. We found a weak but significant positive correlation between the mean normalised SUV and the meanKtrans (Spearman’s r=0.302, p=0.035).
Conclusion: There is a weak but significant positive correlation between Ktrans, which is a marker for neovascularisation and SUV, which is a marker for inflammation. Future studies are warranted to investigate whether DCE-MRI and/or 18F-FDG PET can be used to predict cerebrovascular events.
B-0688 One-to-one comparison between digital mammography and digital breast tomosynthesis using a fully automated software: breast density underestimation on digital breast tomosynthesis varies in different BI-RADS classes
A. Tagliafico, S. Airaldi, F. Cavagnetto, B. Bingotti, S. Tosto, D. Astengo, M. Calabrese | Sunday, March 10, 10:30 – 12:00 / Room F2
Purpose: To compare breast density on digital mammography (FFDM) and tomosynthesis (DBT) according to different BI-RADS classes (four classes from 1 to 4) with an automated software.
Methods and Materials: IRB approval and written informed consent were obtained. Digital breast tomosynthesis and digital mammography were obtained in the same patient. A total of 160 consecutive patients (mean age years: 50±14; mean BMI: 22 ± 3) were included. One-to-one comparison between FFDM and DBT was made with a fully automated software previously validated. Statistical analysis was performed with two-tailed t-test for paired data using statistical software.
Results: In BI-RADS class 1, digital mammography overestimated breast density of a 16 %. In BI-RADS class 2, digital mammography overestimated breast density of a 11.9%. In BI-RADS class 3, digital mammography overestimated breast density of a 3.5%. In BI-RADS class 4, digital mammography overestimated breast density of a 18.1%. The differences resulted highly statistically significant (p<0.0001). There was a good correlation between BI-RADS categories and the density evaluated with digital mammography and digital breast tomosynthesis (r=0.56, p <0.01 and r=0.48 p<0.01).
Conclusion: Breast density values were underestimated by DBT in comparison to FFDM with a non-linear relationship in the different BI-RADS classes. This data should influence clinical and research studies dealing with breast density as a qualitative biomarker.
B-0984 Hepatic parenchymal and vascular contrast improvement in super-delayed phase images of Gd-EOB-DTPA-enhanced MRI
S. Kobayashi, O. Matsui, T. Gabata, W. Koda, T. Minami, K. Kozaka, A. Kitao | Monday, March 11, 14:00 – 15:30 / Room I/K
Purpose: To elucidate the parenchymal and vascular contrast improvement effect of super-delayed phase (SDP) images of Gd-EOB-DTPA (EOB)-enhanced MRI in poor hepatobiliary phase (HBP) image cases special focus on Child-Pugh (CP) classification.
Methods and Materials: 76 cases, who have examined EOB-enhanced MRI for closer examination of hepatic lesions, and taken SDP images approximately 90 minutes after iv administration of EOB because of poor HBP image are subjected to this study. 20 hepatobiliary disease cases who had also taken SDP images which show normal HBP images were used as control. Hepatic vascular/parenchymal enhancement ratios (ER) were defined as signal intensity (SI) of intrahepatic vessel / SI of liver. ER of HBP and SDP were calculated and compared between each CP class liver damage groups. Chi square test was used for statistics and p<0.05 was considered statistical significant.
Results: In poor HBP cases (n=76), ER of HBP and SDP were 0.88±0.16 and 0.64±0.16. In control cases (n=20), ER of HBP and SDP were 0.54±0.08 and 0.39±0.06. ER of HBP and SDP in CP-A poor HBP (n=27), CP-B poor HBP (n=47), CP-C poor HBP (n=2) were 0.83±0.14 and 0.60±0.13, 0.90±0.16 and 0.65±0.16, 1.03±0.16 and 0.99±0.19, respectively (all combinations except CP-C showed significance difference).
Conclusion: In most of the poor HBP image cases, SDP image improve parenchymal and vascular contrast except CP-C liver damage cases.
B-0789 CT colonography: accurate registration of prone and supine endoluminal surfaces of the colon
T.E. Hampshire, H.R. Roth, E. Helbren, A. Plumb, D. Boone, G. Slabaugh, S. Halligan, D.J. Hawkes | Monday, March 11, 10:30 – 12:00 / Room E2
Purpose: Computed tomographic (CT) colonography is a technique for detecting bowel cancer or potentially precancerous polyps. Because retained fluid and stool can mimic pathology, CT data are acquired with the patient in both prone and supine positions. Radiologists then match endoluminal locations between the two acquisitions to determine whether pathology is real. This process is hindered by the fact that the colon can undergo large deformations that often occur during repositioning of the patient. Automated registration between datasets could potentially improve efficiency and diagnostic accuracy.
Methods and Materials: We have developed software to establish correspondence between prone and supine endoluminal surfaces. An initialisation step generates image patches at the positions of haustral folds using depth map renderings and is optimised by virtual camera registration. Additional neighbourhood information is then included in a Markov Random Field model to establish landmark-based correspondences. Subsequently, the complexity of the registration task is reduced by mapping both prone and supine surfaces onto a cylindrical domain in which correspondence is established using non-rigid image registration.
Results: The registration was applied to 17 CTC cases including cases exhibiting luminal collapse, achieving fold matching accuracy of 96 %. Providing an accurate initialisation, the method significantly improved the cylindrical registration (p<0.001), achieving a mean error of 6.0mm measured at 1743 reference points.
Conclusion: The proposed method can successfully establish correspondence between prone-supine locations on the endoluminal surface derived from CT colonography. The ability to rapidly and automatically match polyps between acquisitions will facilitate CT colonography interpretation.
B-0680 Texture analysis of malignant breast tumours: is a differentiation of ductal carcinoma in situ, invasive ductal and invasive lobular breast cancer possible?
T. Knogler, K. Pinker-Domenig, N. Perry, S. Milner, K. Mokbel, M.E. Mayerhoefer | Sunday, March 10, 10:30 – 12:00 / Room F2
Purpose: To evaluate the ability of texture features (TF), to differentiate between ductal carcinoma in situ (DCIS), invasive ductal carcinoma (IDC) and invasive lobular carcinoma (ILC) of the breast on full-field digital mammograms (FFDM).
Methods and Materials: 110 screen detected and histopathologically verified breast cancers (27 DCIS, 73 IDC, 10 ILC) imaged with FFDM in standard views were included in this study. For each lesion, a region of interest (ROI) was manually defined, which covered the lesion as well as a rim (1cm width) of normal-appearing breast tissue around the lesion in the view, where the lesion was depicted in largest diameter. TF derived from the grey-level histogram, co-occurrence matrix (COC), run-length matrix (RLM), absolute gradient (AG), autoregressive model (ARM) and wavelet transform were calculated for the ROIs. Fisher coefficients were calculated to determine which TF were best-suited for distinguishing between DCIS, IDC and ILC. Lesion classification was performed using linear discriminant analysis in conjunction with a k-nearest neighbour classifier, based on the combination of the 10 TF with the highest Fisher coefficients. Classification accuracy was used as the primary outcome measure.
Results: The accuracy of texture-based lesion classification was 84.33% (70 of 83 lesions) for IDC vs. ILC, 81.1% (30 of 37 lesions) for ILC vs. DCIS, but only of 70 % (70 of 100 lesions) for IDC vs. DCIS.
Conclusion: TF derived from FFDM may be of value for differentiating between ILC and IDC, and ILC and DCIS, but of limited value for differentiating between IDC and DCIS.
B-0959 Breast cancer prediction modelling based on common mammographic findings in screening
J. Timmers, A.L.M. Verbeek, R.M. Pijnappel, J. in ‘t Hout, M.J.M. Broeders, G.J. den Heeten | Monday, March 11, 14:00 – 15:30 / Room F2
Purpose: To develop a prediction model for breast cancer (nomogram) based on common mammographic findings on screening mammograms. The model is designed to reduce interobserver variation in assigning BI-RADS in the Dutch breast cancer screening programme.
Methods and Materials: We retrospectively reviewed 352 positive (digital) screening mammograms of women participating in the Nijmegen region of the Dutch screening programme (December 2006-November 2008). The following mammographic findings were assessed by consensus reading of 3 expert radiologists: masses and features of masses, calcifications, parenchymal deformity, asymmetric density and mammographic density and BI-RADS. Data on age, diagnostic work-up, final diagnosis and surgical procedures were collected from patient records. Multivariable logistic regression analyses were used to build our breast cancer prediction model, presented as a nomogram.
Results: Breast cancer was diagnosed in 108 cases (31%). The highest positive predictive value (PPV) was reported for spiculated masses (96%) and the lowest for well-defined masses (9%). Characteristics included in the nomogram based on statistical significance and clinical relevance are: age, mass, calcifications, parenchymal deformity and asymmetric density.
Conclusion: With our nomogram we developed a tool to assist screening radiologists in determining the chance of malignancy based on mammographic findings. We propose cut-off values for assigning BI-RADS categories in the Dutch screening setting based on our nomogram which will need to be validated in future research. These values can easily be adapted for use in other screening programmes.
B-0158 Initial clinical results of simultaneous PET/MRI in comparison with PET/CT in patients with head and neck cancer
P. Stumpp, K. Kubiessa, S. Purz, M. Gawlitza, A. Kühn, K.G. Steinhoff, A. Boehm, R. Kluge, T. Kahn | Thursday, March 7, 14:00 – 15:30 / Room C
Purpose: Describing diagnostic capability of simultaneous PET/MRI in comparison with PET/CT and their single components CT, MRI and PET in an initial prospective study based on 17 patients with head-and-neck cancer.
Methods and Materials: 17 patients with head-and-neck cancer received an 18F-FDG-PET/CT for staging or follow-up and an additional simultaneous PET/MRI scan with a whole body imaging part and a dedicated examination of the neck. Sole MRI, CT and PET components of the multimodal acquisitions plus the PET/MRI and PET/CT examinations were evaluated independently, blinded and in a randomised order by two readers. Results were compared with the reference standard, and sensitivity, specificity, positive (PPV) and negative predictive value (NPV) were calculated.
Results: 23 malignant tumours and 55 benign changes were found with the reference standard. For PET/CT the two reader groups showed a sensitivity of 78.3% and 87 %, a specificity of 85.5% and 89.1%, a PPV of 71.4% and 75 % and a NPV of 90.7% and 94 %. For PET/MRI sensitivity was 78.3% and 82.6%, specificity 81.8% and 94.5%, PPV 65.5% and 85.7% and NPV 91.2% and 93.8%. Evaluation of the single PET part from PET/CT revealed highest sensitivity of 95.7%, whereas evaluation of the sole MRI component from PET/MRI showed best specificity of 96.4%. There was a high interrater agreement in all modalities (Cohen’s kappa coefficient: 0.61 – 0.82).
Conclusion: PET/MRI of patients with head and neck cancer yielded good diagnostic capability, similar to PET/CT. Further studies on larger cohorts to prove these first results seem justified.
B-0012 Language impairment and reduced structural connectivity in Rolandic epilepsy
R. Besseling, J. Jansen, W.H. Backes | Thursday, March 7, 10:30 – 12:00 / Room C
Purpose: Rolandic epilepsy (RE) is a childhood epilepsy with mild seizure semiology and epileptic discharges originating from the sensorimotor (Rolandic) area. A serious co-morbidity in RE is language impairment, for which we aim to find a correlate using structural connectivity.
Methods and Materials: Diffusion MRI was performed at 3.0 Tesla and 2×2×2 mm voxel size with b=1200 s/mm2 and 66 diffusion gradient directions in 23 children with RE with proven language deficits and 23 matched controls. Constrained spherical deconvolution provided voxel-wise tract orientations and whole-brain tractography (5M streamlines). For each of the 4 Rolandic areas (bilateral pre- and postcentral gyri), the streamlines to any of the 70 regions, obtained by automatic cortical parcellation (Freesurfer), were selected from the whole-brain tractogram. For reliable connections (top 20 % of number of streamlines), connectivity was quantified by tract fractional anisotropy (FA), compared between groups, and correlated to language scores.
Results: In the left hemisphere, reduced tract FA was found for connections between the Rolandic areas and inferior frontal gyrus (Broca’s area) and supramarginal gyrus (Wernicke’s area), but not for contralateral homologue regions. For the aberrant connection between the left postcentral and inferior frontal gyrus, tract FA decreased with languages score in the RE group (p=0.04).
Conclusion: In RE structural connectivity between Rolandic and language areas is disturbed and correlates with the language dysfunction. This provides a neuronal link between the location of the epileptic discharges and the language impairment. This research was supported by the Dutch Epilepsy Foundation.
B-0786 Evaluation of a new method for the assessment of anterior acetabular coverage and hip joint space narrowing
R. Ferré, E. Gibon, A. Feydy, H. Guerini, R. Campana, N. Zee, C. Bourdet, M. Hammadouche, J.-l. Drapé | Monday, March 11, 10:30 – 12:00 / Room E1
Purpose: The Lequesne’s false profile (LFP) view is commonly used to evaluate hip joint space narrowing (JSN) and anterior acetabular coverage using the vertical-center-anterior margin (VCA) angle. A novel low-dose biplanar slot scanner (SS) allows simultaneous acquisitions of weight-bearing oblique views of both hip joints. The aim was to compare LFP views versus biplanar oblique views obtained by SS.
Methods and Materials: LFP views were obtained on 56 hips on a computed radiography system. On the same hips, simultaneous oblique views of both hips were acquired in the SS, with the patient’s pelvis positioned at 45° from each acquisition plane. Two independent observers measured VCA angle and JSN on each acquisition. JSN was evaluated through the joint space/femoral head diameter ratio. Measurements from both techniques were compared using the student t-test and the Pearson’s correlation coefficient. Interobserver agreement of VCA angle and JSN assessments were calculated with the intraclass correlation coefficient (ICC).
Results: VCA angle was 33.5° (SD = 6.2°) with SS and 35° (5.6°) with LFP views (p<0.05). Pearson correlation coefficient between two techniques was 0.78 (p<0.01). JSN was 0,114 (SD = 0.03°) with SS and 0.108 (SD = 0.03°) with LFP views (p<0.05). Pearson correlation coefficient was 0.85 (p<0.01). ICCs for VCA angle were 0.91 with SS and 0.69 with LFP views. ICCs for JSN evaluation were 0.76 for both SS and LFP views.
Conclusion: SS is a reliable, easy, and low-dose evaluation of JSN and VCA angle despite a slight angular positioning compared with LFP.