Improving Performance of Breast Cancer Risk Prediction by Incorporating Optical Density Image Feature Analysis: An...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Oncology

Background:

• Accurate near-term breast cancer risk prediction is crucial for early detection and intervention.
• Traditional mammographic analysis relies on grayscale value (GV) features, which may not capture all relevant information.

Purpose of the Study:

• To enhance the accuracy of near-term breast cancer risk prediction.
• To evaluate a novel mammographic image conversion method combined with a two-stage artificial neural network (ANN) classification scheme.

Main Methods:

• Mammographic images from 168 screening cases were analyzed.
• A new method converted grayscale value (GV)-based images to optical density (OD)-based images.
• A two-stage ANN classification scheme fused features from both GV and OD images.

Main Results:

• The proposed two-stage classification scheme achieved an area under the receiver operating characteristic curve (AUC) of 0.816 ± 0.071.
• This performance was significantly higher than using GV features (AUC = 0.669 ± 0.099) or OD features (AUC = 0.646 ± 0.099) alone (P < .05).

Conclusions:

• Optical density (OD) image conversion provides complementary information to standard grayscale analysis.
• Fusion of image features from both GV and OD mammograms significantly improves near-term breast cancer risk prediction accuracy.