|
|
 |
| |
|
|
|
MATLAB-Based Multi-Marker Data Analysis System for Early Detection of Ovarian Cancer |
|
|
|
PP: 697S-703S |
|
|
|
Author(s) |
|
|
|
Yu-Seop Kim et al,
|
|
|
|
Abstract |
|
|
| Recent research has focused on testing multiple biomarkers for the early detection of cancer to overcome the intrinsic limitations of individual markers. An analysis of multiple biomarker data requires data-mining tools and a statistical suite to investigate individual and combined characteristics. The existing general purpose systems lack compact functionality and an intuitive graphical user interface that allows the user to easily identify characteristics that can distinguish among target disease states. This paper presents an analytical system with statistical and data-mining functions for the multi-marker detection of ovarian cancer. The system was written in MATLAB, because this platform provided a high degree of functionality and facilitated the flow of information among the appropriate analyses. MATLAB also provided several visualization choices and an intuitive graphical user interface. The proposed system was tested with serum level data measured with a high-throughput, multiplex, bead-based, immunoassay platform (Luminex). The system was designed to offer the user the choice of manually selecting marker combinations or automatically selecting marker combinations. It was constructed of modular analyses that evaluate selected markers for classification efficiency based on the best performance. The results can be visualized with dot and box plots for individual markers and 2D and 3D scatter plots of the combined characteristics of multiple markers. This system offers a user-friendly method for evaluating multiple markers and facilitating clinical diagnoses. |
|
|
|
|
|
|
|
