Classification of Dynamic Breast MR Image Data - Biomedical Signal Analysis: Contemporary Methods and Applications - page 277

Trough use of a Kohonen neural network, areas with similar signal time
courses in mammographic image series were detected, making possible
a clear detection of carcinoma [85].
In Summary, the major disadvantages associated with standard
techniques in breast MRI are (1) requirement of a fixed MR imaging
protocol, (2) lack of increase in sensitivity and/or specificity, (3) inability
to capture the lesion structure, and (4) training limitations due to an
inhomogeneous lesion data pool.
To overcome the above-mentioned problems, a minimal free energy
vector quantization neural network is employed that focuses strictly
on the observed complete MRI signal time series and enables a self-
organized, data-driven segmentation of dynamic contrast-enhanced
breast MRI time series with regard to fine-grained differences of sig-
nal amplitude, and dynamics, such as focal enhancement in patients
with indeterminate breast lesions. This method is developed, tested,
and evaluated for functional and structural segmentation, visualization,
and classification of dynamic contrast-enhanced breast MRI data. Thus,
it is a contribution toward the construction and evaluation of a flexible
and reusable software system for CAD in breast MRI.
The results show that new method reveals regional properties of
contrast-agent uptake characterized by subtle differences of signal am-
plitude and dynamics. As a result, one obtains both a set of prototypical
time series and a corresponding set of cluster assignment maps which
further provide a segmentation with regard to identification and regional
subclassification of pathological breast tissue lesions. The inspection of
these clustering results is a unique practical tool for radiologists, en-
abling a fast scan of the data set for regional differences or abnormali-
ties of contrast-agent uptake. The proposed technique contributes to the
diagnosis of indeterminate breast lesions by noninvasive imaging.

10.1 Materials and Methods

Patients

A total of 13 patients, all female and ranging in age from 48 to 61, with
solid breast tumors, were examined. All patients had histopathologically
confirmed diagnosis from needle aspiration/excision biopsy and surgical
removal. Breast cancer was diagnosed in 8 of the 13 cases.