An adaptive tissue characterization network for model-free visualization of dynamic contrast-enhanced magnetic resonance image data

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has become an important source of information to aid cancer diagnosis. Nevertheless, due to the multi-temporal nature of the three-dimensional volume data obtained from DCE-MRI, evaluation of the image data is a challenging task and tools are required to support the human expert. We investigate an approach for automatic localization and characterization of suspicious lesions in DCE-MRI data. It applies an artificial neural network (ANN) architecture which combines unsupervised and supervised techniques for voxel-by-voxel classification of temporal kinetic signals. The algorithm is easy to implement, allows for fast training and application even for huge data sets and can be directly used to augment the display of DCE-MRI data. To demonstrate that the system provides a reasonable assessment of kinetic signals, the outcome is compared with the results obtained from the model-based three-time-points (3TP) technique which represents a clinical standard protocol for analysing breast cancer lesions. The evaluation based on the DCE-MRI data of 12 cases indicates that, although the ANN is trained with imprecisely labeled data, the approach leads to an outcome conforming with 3TP without presupposing an explicit model of the underlying physiological process.

I. Introduction

In recent years, multivariate imaging techniques have become an important source of information to aid diagnosis in many medical fields. One example is the dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) technique which has been successfully applied for, e.g., diagnosis of cancer in the breast [1], [2], prostate [3], [4], and liver [5]. After administration of a gadolinium-based contrast agent (CA), a temporal sequence of three-dimensional (3-D) MRI volumes is recorded, e.g., from the female breast. The patients are placed in a prone position to minimize motion artifacts. In this study, only data sets are used that do not require additional registration, i.e., show a sufficient anatomical alignment over time. Thus, each position of the data volume can be associated with a -dimensional vector describing the temporal kinetic pattern of the local CA concentration.