Detection and classification the breast tumors using mask R-CNN on sonograms

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Breast cancer poses a significant global health challenge, particularly for women.
• Early detection through screening is crucial for reducing mortality rates.
• Current screening methods like ultrasound combined with biopsy are invasive and may yield incomplete information.

Purpose of the Study:

• To develop an automated deep learning model for breast lesion detection, segmentation, and classification using ultrasound images.
• To provide a comprehensive and noninvasive alternative to traditional biopsy methods.

Main Methods:

• A deep learning approach utilizing Mask Regions with Convolutional Neural Networks (Mask R-CNN) was employed.
• The model was trained and evaluated on ultrasound images for lesion identification and characterization.

Main Results:

• The model achieved a mean average precision of 0.75 for lesion detection and segmentation.
• The classification accuracy for distinguishing between benign and malignant lesions was 85%.

Conclusions:

• The proposed deep learning method offers a noninvasive and comprehensive approach for breast lesion analysis.
• This technology has the potential to enhance early breast cancer diagnosis and improve patient outcomes.