Medical image segmentation — radiology AI
Performed pixel-level segmentation of CT, MRI, and X-ray scans for a U.S.-based radiology AI company developing diagnostic imaging models for early disease detection. Annotated tumors, organ boundaries, lesions, and anatomical landmarks across chest, abdominal, and neurological datasets using OHIF Viewer and ITK-SNAP. Followed detailed clinical annotation guidelines developed in collaboration with board-certified radiologists, ensuring medically accurate boundary delineation and consistent labeling across multi-annotator teams. Handled DICOM file formats exclusively and maintained strict HIPAA-compliant data handling throughout the engagement. Later transitioned into a QA reviewer role, auditing annotations produced by junior labelers and calibrating inter-annotator agreement scores to maintain dataset quality above a 96% threshold. Delivered annotated datasets in structured JSONL format ready for model ingestion, contributing directly to a diagnostic AI pipeline targeting radiology workflow automation.