Breast cancer is the most common cancer (excluding skin cancer) and the second most frequent cause of cancer death among women in the United States. Dynamic contrast-enhanced MRI (DCE-MRI) screening is recommended as an adjunct for mammography for high-risk women and is increasingly used as a key staging tool for newly diagnosed breast cancer.
Case-based clinical decision support (CDS) methods mimic natural
physicians’ reasoning for diagnosis by reading similar cases that have been previously reviewed and diagnosed. A case-based CDS for breast lesion diagnosis may help physicians to decide whether a queried case is likely to be malignant or benign, thus avoiding misinterpretation and preventing unnecessary biopsies.
Most of the case-based approaches in the published literature focus on breast cancer diagnosis based on mammogram and ultrasound images. Extending the case-based approach to DCE-MRI may improve the sensitivity and specificity. The challenge is to design an algorithm by taking more computer-calculated features into account to capture characteristics of malignant lesions. Additionally, there is also a challenge in obtaining the similarity relationship among cases for a case-based CDS system.
Online access to SPIE eBooks is limited to subscribing institutions.