Abstract Archives of the RSNA, 2009
LL-BR4071-D06
Diagnostic Value of CAD-Application to Discriminate Dignitiy of Axillary Lymph Nodes in Dynamic Breast-MRI
Scientific Posters
Presented on November 30, 2009
Presented as part of LL-BR-D: Breast Imaging
Ansgar Malich MD, Presenter: Nothing to Disclose
Recently developed CAD-technology allows an automated quantification of contrast enhancement of lymph nodes at dynamic breast-MRI. MR-based analysis of lymph node metastases is still a diagnostic challenge despite its outstanding importance for prognosis of breast cancer. Study aimed to analyze dynamic pattern of lymph nodes using CAD, differentiating metastatic and non metastatic nodes.
200 breast-MRI containing a malignant lesion (IDC:n=140; ILC: n=17; ITC: n=12, IPC: n=2, DCIS: n=22, other malignancies: n=7) were analyzed including the axillary edge and nodes (histologically proven). 0.1 mmol Gd-DTPA/kg bw was applied, 7 dynamic T1w series (1min. time interval) were obtained, peak contrast uptake, wash in ratio, wash out ratio, dynamic enhancement of the entire lesion/node were registered and matched. In 64 cases metastatic nodes were histologically proven.
33/136 non metastatic and 2/64 metastatic nodes were not visible T2-based. 25/103 visible non metastatic and 1/62 visible metastatic nodes were without a quantifiable contrast enhancement. Peak contrast uptake differed between metastatic and non metastatic nodes suggesting a stronger and more continued uptake of metastatic nodes (423%/191%). Wash out occurred similarily in metastatic and non-metastatic nodes (74.6%vs.72.3%).
According to ROC-curve-analysis, using a cut off value of 240% (peak contrast uptake) reveals a sensitivity of 75.0% and a specificity of 82.4%. Strong initial wash in >35% (cut off 2) reveals a sensitivity of 75%, and a specificity of 83%. N1a-metastases are not characterized by an altered enhancement. Excluding this group, sensitivity is >90% using both cut off parameters. Correlation peak uptake metastatic lymph nodes – tumor was not significant (r=-0.08).
Most dynamic pattern including wash out are of less importance in the discrimination of lymph nodes. The maximum contrast uptake and the rapid initial wash in ratio, however, differ significantly between the several nodes dignitiy-dependend. Very early metastatic nodes (N1a) can not be differentiated using enhancement parameters.
It is possible to reliably differentiate lymph node metastases from non metastatic nodes due to a stronger initial wash in ratio and a higher peak contrast uptake using CAD at MRI-images.
Malich, A,
Diagnostic Value of CAD-Application to Discriminate Dignitiy of Axillary Lymph Nodes in Dynamic Breast-MRI. Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL.
http://archive.rsna.org/2009/8012769.html