Selection of suspicious ROIs in breast DCE-MRI
ICIAP'11 Proceedings of the 16th international conference on Image analysis and processing: Part I
Image analysis of histological features in molar pregnancies
Expert Systems with Applications: An International Journal
| Hi-index | 0.00 |
This study is designed to segment suspiciousregions using automatic computerized procedures andto classify kinetic patterns using commerciallyavailable three-time-points (3TP) method of computeraideddiagnosis. A novel evaluation method usingperfusion volume fractions is introduced for examiningmeaningful kinetic features in differentiation of benignand malignant breast lesions. Dynamic contrastenhancedMRI was applied to 24 lesions (12 malignant,12 benign). Thresholding for suspicious regions,region growing segmentation, hole-filling and 3Dmorphological erosion and dilation were performedfor extracting final lesion volume. The lesion sphericityand center distance of mass to surface area ratio(CDMSAR) were considered in the process ofautomatic segmentation. The kinetic patterns for eachlesion were classified into six classes by the 3TPmethod. Perfusion volume fraction for each class wascalculated in three partitions of whole, rim and corevolumes of a lesion. Receiver operating characteristiccurve (ROC) analysis was performed using theperfusion volume fractions. When using perfusionvolume fractions divided into rim and core lesionvolume, the classes having more improved accuracyappeared than using perfusion volume fractions withinwhole lesion volume. This result indicates that lesionclassification using local perfusion volume fractions ishelpful in selecting meaningful kinetic patterns fordifferentiation of benign and malignant lesions.