MEDICAL IMAGING / ANNOTATION PLATFORM
Connect DICOM intake, pre-annotation, manual revision, collaborative review and export. Prepare reviewable annotations for research, medical AI and data delivery teams.
Start with the task. No patient-image upload required.
Illustrative workflow, not a product screenshot or measured result. Verify the actual interface in a demo.
01 / START WITH THE TASK
Configure the workflow around your target structures, sequences and outputs.
01 / TASK DESIGN
Use rectangles, ellipses, lassos, free contours and MASK brushes. Agree the target structures, labels and boundary rules first.
Verify contour edits, negative and uncertain labels, and alignment with source images.
Book a task-focused demo ↗02 / TASK DESIGN
Combine suitable AI drafts, interpolation and manual edits. Check annotations with MPR and 3D views.
Verify cross-slice continuity, voxel spacing, orientation and mask exports.
Book a task-focused demo ↗03 / TASK DESIGN
Organize research images by dataset and subject; associate clinical variables at Patient, Study and VOI levels.
Verify inclusion criteria, variable mappings, case consistency and Excel exports.
Book a task-focused demo ↗Rectangles, ellipses, free contours, lassos and MASK brushes support contour edits, interpolation and threshold interactions. Label naming and colors help organize targets.
Window and level, sequence browsing, suitable MPR, 3D target views and measurements. PET-CT fusion depends on matched data conditions.
Start from suitable pre-segmentation outputs, then inspect, revise and review. Confirm model configuration and applicability in the demonstration.
Member roles, annotation sharing, single or multiple blinded review and status tracking. Personal and collaborative datasets can associate Patient, Study and VOI variables, including Excel imports.
Radiomics, feature selection and machine-learning analysis can be assessed as optional modules. WSI, video, text, advanced arbitration and version-diff workflows require separate scoping.
02 / MODEL COVERAGE
186 is a model count, not a disease count or case inventory. Check applicability, licensing and manual revision needs using your target sequences.
The published catalog lists 128 CT, 56 MR, one US and one MG entry. The US and MG entries require project-specific assessment. PET is part of the imaging workflow; this count does not establish a preset PET model.
Browse all 186 entries (Chinese) ↗Catalog entries require task-specific verification. Drafts need manual revision and review; no sensitivity or Dice result is promised.
03 / REVIEWABLE DELIVERY
Agree target structures, labels, boundaries and treatment of uncertain samples.
Agree blinded or cross-review, sampling coverage and responsibility for disagreements.
Check labels, dimensions, orientation and voxel spacing with a sample in the downstream tool.
Dice / IoU calculations, automated sampling and multi-level arbitration require separate scope agreement; they are not all assumed to be built-in.
04 / INPUT & OUTPUT
Local batch upload uses a companion client. PACS retrieval needs integration. Verify devices, sequences and transfer methods using samples.
Assess WSI, video and other non-DICOM inputs separately.
NIfTI export does not imply NIfTI import. DICOM export does not imply DICOM SEG / RTSTRUCT annotation exchange; confirm these separately.
05 / PLAN THE DEPLOYMENT
Assess on-premises, private-environment and integration requirements before defining delivery. Itemize software, optional modules and implementation services.
Servers, storage, GPU, network, concurrency and model configuration.
Member roles, source permissions, de-identification scope, logs and backup requirements.
PACS, migration, training, maintenance, upgrades and data exit arrangements.
06 / CHOOSE YOUR NEXT STEP
Choose the right route so we can scope the right engagement.
Verify the interface, collaboration and outputs; discuss the platform and deployment.
Book a platform demo ↗02 / SERVICEAgree the task brief, protocol, pilot annotation, review and batch delivery.
Discuss annotation services ↗03 / BIOLOGICAL AIScope cell images, omics and experimental data separately from medical imaging models.
Explore biological projects ↗QUESTIONS / ANSWERED
This page covers platform software and deployment options. Choose annotation service if you want a team to perform the work; task definitions, personnel and acceptance are agreed separately.
No. Suitable models provide drafts. Experts must check and revise them. Model availability, licensing and applicability are confirmed for the target anatomy and sequence; no sensitivity or Dice score is promised.
Private or on-premises requirements can be assessed. Network, compute, storage, accounts, security requirements and interfaces are scoped before a deployment proposal.
The main input path is DICOM. Images export as DICOM or nii.gz, annotations as TXT, NRRD or nii.gz, and clinical variables as Excel. NIfTI export does not establish NIfTI import. DICOM SEG and RTSTRUCT exchange require separate confirmation.
No. Start with your task and contact information. Any later sample exchange requires agreed permissions, de-identification, transport and retention arrangements.
Scope software, optional modules, model licensing, integrations, training and maintenance separately. Use an authorized sample to verify import, annotation, review and export before agreeing acceptance.
The standard workflow here is medical imaging. Cell microscopy, omics, WSI, video, text and other biological tasks require separate scoping; medical preset models do not establish support for those tasks.
LET’S SCOPE YOUR PROJECT
Tell us the modality, target structures and environment. We will discuss the demonstration, fit checks and next materials.
+86 15010400907 ↗date@lumesage.com
Scan to discuss your project
Mention “imaging annotation”
No patient images are needed for initial contact. Agree permissions, de-identification and transfer before sample exchange.