Services · Medical and dental imaging software

Imaging software from the scanner to the clinical decision.

Imaging manufacturers and imaging software companies need image pipelines, viewers, analysis and AI that run reliably next to PACS, across modalities and vendors, and pass an audit. The engineers of Silurus Software build and maintain imaging software for every step from acquisition to report, under IEC 62304 inside your ISO 13485 quality management system, with our regulatory specialists and FDA consultants on the file for the EU MDR and the FDA.

Book a 30 minute call Write to us

Modalities: CT · MR · X-ray · US · PET · CBCT
Standards: DICOM · DICOMweb · HL7 · FHIR
Process: IEC 62304 · ISO 13485 · ISO 14971
Route: EU MDR · FDA

The chain

Every step from acquisition to report

ModalityCT, MR, X-ray, US, PET, CBCT

PACS and RISStorage, worklists, DICOMweb

IngestionValidation, sorting, anonymisation

Processing and AISegmentation, registration, inference

Viewer2D, MPR, 3D, fusion, measurement

Report and planSR, SEG, RTSTRUCT, export

What we cover

Imaging software, step by step

DICOM import from many modalities

We build the import that checks each incoming DICOM study, sorts it into series and converts it to one internal format. For each modality and vendor agreed for the product we verify that image orientation, pixel spacing and calibration are read correctly, before the image reaches an algorithm or a viewer.

DICOM connectivity and hospital integration

We build storage, query and retrieve, worklists and DICOMweb, connect results to RIS and record systems over HL7 and FHIR, and keep the DICOM conformance statement current.

Interoperability →

Viewers and visualisation

We build desktop and zero footprint web viewers with multiplanar and 3D rendering, fusion, measurement, contour editing and hanging protocols, with the usability work IEC 62366-1 asks for.

Image processing and quantification

We build segmentation, registration, filtering and quantification as production code with GPU acceleration where needed, verified per release against reference data.

AI in imaging products

We build the inference service, return results as DICOM objects the workstation displays, and verify and monitor the model per scanner, site and patient group.

AI model to medical device →

Treatment and surgical planning

We build contouring and planning software, from the 3D model to the exported plan, with RT objects and surface formats other systems can read.

Cloud, research and scale

We build cloud platforms connected to hospital PACS, with anonymisation and pseudonymisation, isolated algorithm services and scaling with demand.

Long running imaging platforms

Our engineers join platform teams for years, on features, performance, security work under IEC 81001-5-1 with the SBOM (software bill of materials, the list of every software component in the build) and threat model (a structured analysis of how the device could be attacked), field defects and platform updates.

Team extension →

Segmentation

Segmentation of CT, MR and CBCT images

Segmentation in a medical product is production code with a verification record. Classical methods such as thresholding, region growing and atlas based segmentation are written in C++ with ITK, deep learning models of the U-Net family run as a versioned inference service. Clinicians correct results with brush, contour editing and propagation across slices. Label maps leave the product as DICOM SEG, contours as RTSTRUCT and measurements as a DICOM structured report, and each release is measured against reference data with Dice coefficient and surface distance (mean and 95th percentile Hausdorff). The acceptance criteria are set from the clinical risk of each structure.

Registration

Registration of images and surfaces

Registration brings two images, or an image and a surface, into one coordinate system. Rigid registration fits when the anatomy keeps its shape, for example CT to MR of the head or a follow up to its baseline. Affine registration adds scaling and shear. Deformable methods model breathing and anatomical change and can carry contours from one image to another. Surface registration with iterative closest point aligns meshes such as intraoral scans with surfaces extracted from CT or CBCT. Each method is verified on reference cases with target registration error or surface distance in millimetres. Deformable results are also checked for folding with the Jacobian determinant.

Dental

Dental CBCT and intraoral scan software

Dental planning software combines CBCT volumes with intraoral STL scans. CBCT grey values differ between devices and fields of view, metal from crowns and implants disturbs segmentation, and scanners deliver meshes with different units, holes and borders. STL carries no unit, so unit and orientation are checked per scanner. Teeth are segmented and numbered, the jaws and the mandibular canal are segmented, the crowns are aligned with the scan by landmarks and refined with iterative closest point, and a panoramic view is reformatted along the dental arch. Meshes, labels and the transform go to the planning front end through a versioned API, and the surgical guide is exported for design.

In practice

Situations we are asked to solve

Situation

A core imaging engine for a new product

An imaging product needs its engine: DICOM ingestion, 3D segmentation, registration and visualisation, in C++ and Python, as a regulated medical device called by a web application.

How we approach it

We build the engine as a headless service with a versioned API, or as a library linked into the product, on open source toolkits such as DCMTK, ITK and VTK, which go into the SOUP list (software of unknown provenance, meaning third party code used in the device). The DICOM import checks and converts the studies. Segmentation and registration use your algorithms or models, or ones we develop with you on data you provide, and run as verified production code. Results return as DICOM SEG, SR or RTSTRUCT objects or as meshes and transforms through the API.

The regulated boundary is agreed at the start, so each component carries the IEC 62304 records it needs.

Situation

Dental planning from CBCT and intraoral scans

A dental implant planning platform needs jaw and tooth segmentation from CBCT, import of intraoral STL scans and their registration to the CBCT volume.

How we approach it

The work starts with a data phase on your own cases: CBCT grey values per device, metal artefacts, mesh quality per scanner. Teeth are segmented and labelled, the crowns are aligned with the scan by landmarks and refined with iterative closest point, and each case reports its surface distance in millimetres against an agreed threshold. Registered meshes, labels and the transform are returned through the API to the planning front end.

Situation

Contouring and radiotherapy planning tools

An oncology software company builds a contouring and planning module that has to work with existing radiotherapy systems.

How we approach it

Contours, plans and doses move as DICOM-RT objects (RTSTRUCT, RTPLAN, RTDOSE). Contouring tools, automatic organ at risk segmentation and plan evaluation are built under IEC 62304 and IEC 62083, with the safety class (A, B or C, by the harm a software failure could cause) the planning function requires.

Situation

Replacing the core of a diagnostic workstation

A diagnostic imaging workstation gets a new core: DICOM handling, 2D and 3D visualisation and GPU accelerated processing, while the product stays on the market.

How we approach it

The new core is built behind stable interfaces and verified against the old one with recorded studies, so measurements and images can be compared before the switch. Whether the change has to be notified to the notified body (the independent organisation that certifies a device under the EU MDR) or needs a new 510(k) is assessed with our regulatory specialists before the switch.

What you receive

Imaging software with its evidence

D-01

Imaging software

Engine, viewer or service, as a product, a module or a library in the host application.

D-02

DICOM interfaces

Connections to modalities, PACS and RIS, and the DICOM conformance statement.

D-03

Verification

Accuracy and performance tests per release, against reference data and per scanner.

D-04

IEC 62304 file

Software requirements, architecture, SOUP records and traceability under IEC 62304, the software risk analysis as input to your ISO 14971 file, and the usability file under IEC 62366-1.

EU and US

Imaging software for both markets

European Union

EU MDR and CE marking

The software part of the technical documentation, written with our regulatory specialists.

United States

FDA 510(k) and De Novo

The software documentation of the submission, prepared with our FDA consultants.

Technologies and standards

What we work with

Technologies and tools

C++PythonC#TypeScriptCUDATensorRTDCMTKfo-dicomOrthancOHIFCornerstone3DITKVTK3D SlicerMONAIPyTorchTensorFlowONNX RuntimeDICOMwebDocker

Standards, regulations and practices

IEC 62304ISO 13485ISO 14971IEC 62366-1EU MDRFDA 510(k)

Silurus Software

Silurus Software, founded in 2017, offices in Budapest, Hungary, and Singapore
More than 40 engineers, EU MDR regulatory specialists and FDA consultants
Work inside the client's ISO 13485 quality management system, with the split of tasks written down before the work starts. Where the client has no documented software process yet, we bring IEC 62304 templates for the client to adopt into its own system
Multi-year engineering partner of a European MedTech manufacturer
Founder and CTO Péter Török, in medical technology software since 1999, formerly software development director at Siemens Healthcare (syngo.via) and head of R&D at Mediso Medical Imaging Systems
ISO 9001:2015 (NQA, certificate 171041, valid until March 2027) and ISO/IEC 27001:2022 (NQA, certificate 728872, valid until August 2029)
Awarded AAA, Highest Creditworthiness, by Dun & Bradstreet in 2025
Research partner of ELTE Eötvös Loránd University, Budapest
Industrial partner of the University of Bologna, the University of Pisa, the University of Amsterdam and Masaryk University, Brno, in the EU Horizon Europe project LOGOS from 2027
Female Business Leader of the Year 2025 and The Pioneer Award 2026, awarded to CEO Dr. Linda Szántó

How we work →Quality and standards →Sample deliverables →Case studies →

Related work and pages.

Client names are withheld unless the client has agreed to be named.

Case study

Engineering capacity for a mission critical imaging platform

Eight engineers inside a European manufacturer's imaging platform teams, IEC 62304, quarterly releases each through full V&V, over several years.

Case study

Cloud platform for radiotherapy contouring algorithms

A European manufacturer's cloud platform on which research hospitals run their own contouring algorithms, connected to PACS over DICOM.

AI model to medical device Interoperability Team extension Medical imaging OEMs Radiotherapy OEMs

FAQ

Frequently asked questions

Which modalities do you work with?

CT, MR, X-ray and fluoroscopy, mammography, ultrasound, PET, SPECT and dental CBCT and intraoral scans. The pipeline is built to handle the specifics of each.

Can you build an imaging engine that other applications call?

Yes, as a service with a versioned API or as a C++ or Python library, on premise or in the cloud. It is built on open source toolkits such as DCMTK, ITK and VTK, listed as SOUP, and the segmentation and registration use your algorithms or models, or ones we develop with you on data you provide.

Which toolkits and DICOM objects do you work with?

DCMTK, ITK and VTK, DICOMweb, OHIF and Cornerstone3D, and the result objects SEG, RTSTRUCT, SR and GSPS. Every third party library is recorded as SOUP and in the SBOM.

Can your engineers join our platform team?

Yes. Our engineers work inside manufacturers' teams under their QMS, from features to field defects.

How do you verify a segmentation or a measurement?

Against reference data with agreed accuracy measures, per release and per scanner, recorded in the IEC 62304 verification.

Do you work with DICOM-RT for radiotherapy?

Yes. RTSTRUCT, RTPLAN and RTDOSE objects, contouring tools and plan evaluation, connected to treatment planning and record and verify systems over DICOM.

Can the engine return meshes and transforms through an API?

Yes. Segmentations, surface meshes in STL or OBJ and the registration transform are returned through a versioned API, next to DICOM result objects.

How does a dental imaging project start?

With a data phase on your own anonymised CBCT and intraoral cases, ending in a report on accuracy, risks and a costed plan for the product phase.

Do you work with 3D Slicer, CUDA and TensorRT?

Yes. 3D Slicer for research prototypes and tools, CUDA for GPU accelerated processing, and TensorRT or ONNX Runtime for model inference, next to ITK, VTK and DCMTK.

Talk to us

Tell us about the imaging product.

The modality, the product and what should be built. We reply with the first step.

Book a 30 minute technical call