Statistical Model for Diagnosing Breast Lesions

 

Introduction

 

Breast magnetic resonance imaging (MRI) has developed into a routine clinical practice for breast tumor detection and diagnosis. The American Cancer Society advises women at high risk for breast cancer to undergo a breast MRI rather than mammography. The current method used is a dynamic contrast-enhanced MRI. While breast MRI is more sensitive for lesion detection, it is still difficult to differentiate between benign and malignant lesions, resulting in a high incidence of false-positives. There is a clear need for solutions to improve differentiation between benign and malignant lesions.

 

Description of Technology

 

MSU’s invention is a novel, kinetic, feature-based statistical model for differentiating benign from malignant contrast-enhancing breast lesions. This method involves automatic determination of the boundary of a manually selected dynamic contrast enhancing lesion using the differential signal intensity of the lesion from surrounding tissue. After identifying an objective region of interest for the lesion, the model is then applied to quantitatively analyze the kinetic behavior of post-contrast signal intensity time courses voxel by voxel.

 

Preliminary clinical studies show that the WO volume fraction (wash-out cluster volume to whole-lesion volume) can be used as a biomarker for differentiating benign from malignant contrast-enhancing lesions due to a higher WO volume fraction found in malignant lesions. The use of breast MRI as a breast cancer screening tool is growing and will likely continue to become increasingly common. MSU’s invention addresses the most pervasive problem associated with breast MRI ‒ the low specificity and high incidence of false positives.

 

Key Benefits

  • Automatic determination of lesion boundary: The technology produces an objective lesion region of interest (ROI).
  • Improved suspicious breast lesion characterization: The positive predictive value of biopsies is potentially increased by 87 percent.
  • Reduction of unnecessary biopsies: The number of unwarranted biopsies is potentially reduced by 72 percent.
  • Ease of implementation: The invention should be relatively easy to implement into existing MRI software.

 

Applications

 

The technology would be used to enhance magnetic resonance imaging of breast lesions for use in improving suspicious breast lesion characterization in breast cancer diagnosis.

 

Patent Status

 

Multiple patents pending

 

Inventors

 

Jie Huang

 

Tech ID

 

TEC2009-0002

 

Patent Information:

Category(s):

For Information, Contact:

Randy Ramharack
Technology Manager
Michigan State University
ramharac@msu.edu
Inventors:
Jie Huang