All work Healthcare AI · Medical imaging

3D Segmentation of CT/MRI Volumes

Automated deep-learning segmentation of volumetric CT and MRI scans for medical-imaging products.

RoleMachine Learning Specialist
TypeApplied R&D
DomainMedical imaging
Status2021–2022
Context

Built at Future Processing Healthcare (GrayLight Imaging), a company developing medical-imaging software.

Problem

Manual delineation of structures in 3D CT/MRI volumes is slow and operator-dependent.

Objective

Automate accurate 3D segmentation of anatomical structures in volumetric scans.

Stakeholders

Radiologists and medical-imaging software teams.

My responsibility

Designed, trained and evaluated the deep-learning segmentation models.

Approach

Volumetric deep-learning models trained on annotated CT/MRI data with a 3D imaging pipeline.

Solution

Automated 3D segmentation models integrated into the imaging workflow.

Outcome

Reusable segmentation capability for CT/MRI volumes.

Current status

Delivered as part of the role (2021–2022).

Technical deep dive Implementation detail
DataVolumetric CT/MRI scans.
Model3D deep-learning segmentation networks.
Stack
PyTorchTensorFlowITKPyVistaDICOM