Medical Image Segmentation Labeler (Breast Cancer Ultrasound)
ImageSegmentation
Developed an automated workflow for breast lesion segmentation using ultrasound imagery. Preprocessed, augmented, and normalized images and corresponding masks to prepare data for deep learning model training. Trained a U-Net model using PyTorch, evaluated model predictions, and compared output against ground truth for clinical interpretability.
• Performed grayscale conversion, resizing, and normalization of ultrasound images.
• Applied mask alignment to ensure data-label integrity for semantic segmentation.
• Used augmentations to increase data variability and robustness of training.
• Evaluated model metrics with Dice coefficient and IoU for label quality assurance.
Developed an automated workflow for breast lesion segmentation using ultrasound imagery. Preprocessed, augmented, and normalized images and corresponding masks to prepare data for deep learning model training. Trained a U-Net model using PyTorch, evaluated model predictions, and compared output against ground truth for clinical interpretability. • Performed grayscale conversion, resizing, and normalization of ultrasound images. • Applied mask alignment to ensure data-label integrity for semantic segmentation. • Used augmentations to increase data variability and robustness of training. • Evaluated model metrics with Dice coefficient and IoU for label quality assurance.
2025 - Present