Analysis of Interval Changes on Mammograms for Computer Aided Diagnosis
Abstract
A new classification scheme was developed to classify mammographic masses as malignant and benign by using interval change information. The masses on both the current and the prior mammograms were automatically segmented using an active contour method. From each mass, texture morhological and spiculation features, were extracted. Additionally, difference features were obtained by subtracting the prior from the corresponding current features. The feature space consisted of the current, difference and prior features. Stepwise feature selection and linear discriminant analysis classification were used to select and merge the most useful features. A leave-one-case-out resampling scheme was applied to train and test the classifier using 140 temporal image pairs (85 malignant, 55 benign). An average of 10 features were selected from the 56 training subsets. The classifier achieved an average training A(sub z) of 0.92 and a test A(sub z) of 0.88. For comparison, a classifier was trained and tested using features extracted from the 120 current single images. This classifier achieved an average training A(sub z) of 0.90 and a test A(sub z) of 0.82. The information on the prior image significantly (p = 0.015) improved the accuracy for classification of the masses. A pilot study was performed to evaluate the effects of computer classification on radiologists' estimates of the likelihood of malignancy of masses.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2002
- Accession Number
- ADA404750
Entities
People
- Lubomir Hadjiiski
Organizations
- University of Michigan