A step toward computer-assisted mammography using evolutionary programming and neural networks

Area of Science:

• Medical imaging analysis
• Artificial intelligence in healthcare
• Oncology diagnostics

Background:

• Artificial intelligence (AI) offers potential for improving diagnostic accuracy and cost-effectiveness in medicine.
• AI can serve as a valuable tool for second opinions in clinical settings.
• Early and accurate breast cancer detection is crucial for patient outcomes.

Purpose of the Study:

• To investigate the efficacy of artificial neural networks (ANNs) trained with evolutionary programming for breast cancer detection.
• To evaluate the use of radiographic features and patient age in AI-driven cancer diagnosis.
• To assess the diagnostic performance of simple neural architectures in identifying malignant breast masses.

Main Methods:

• Utilized evolutionary programming to train artificial neural networks (ANNs).
• Employed radiographic features and patient age as input data for the models.
• Analyzed data from 112 biopsy-proven breast masses (63 malignant, 49 benign).

Main Results:

• The trained ANNs demonstrated a significant probability of detecting malignancies.
• Simple neural architectures were sufficient for achieving high detection rates.
• A small percentage of false positives were observed in the diagnostic results.

Conclusions:

• AI techniques, specifically evolutionary programming-trained ANNs, show potential for accurate breast cancer detection.
• The study highlights the feasibility of using AI for improving diagnostic sensitivity and specificity in mammography.
• Further research into refining AI models can minimize false positives and enhance clinical utility.