Products & Tools

Synaptiq's Mediq RT cuts CT contouring time by 92.6% with combined AI tumor and organ contouring

Synaptiq says its Mediq RT platform cut CT contouring time 92.6% including clinician review; head-and-neck cases fell from 97 minutes to 2.7 minutes. A Class IIa CE mark now covers tumor contouring.

Synaptiq Combines AI Tumor and Organ Contouring to Speed Radiotherapy Planning
Synaptiq Combines AI Tumor and Organ Contouring to Speed Radiotherapy Planningjarmoluk / pixabay
By Marcus Bennett5 min read

Why it matters

  • Synaptiq reports an average 92.6% CT contouring time reduction including clinician review, announced November 18.
  • Head-and-neck contouring fell from 97 minutes manually to 2.7 minutes with Mediq RT, a 97.2% decrease.
  • Mediq RT holds a Class IIa CE mark under the EU MDR specifically covering tumor contouring; 92.1% of AI contours were accepted without edits.
  • Synaptiq reports 2.7-minute head-and-neck correction time on the same 80-patient dataset as a 2023 five-vendor study it was not part of, versus 4.4 minutes for the next-fastest system.
  • Founded in 2021 in Cluj-Napoca, Synaptiq serves 11 countries, has processed over 63,000 patients, and plans an FDA 510(k) submission.

Romanian health-tech company Synaptiq says a clinical evaluation of its Mediq RT platform found an average 92.6% reduction in CT contouring time — a figure that includes the minutes clinicians spent reviewing and correcting the AI's output, not raw inference alone. The November 18 announcement brings AI-assisted tumor delineation and organ-at-risk contouring into a single radiotherapy planning workflow, and it arrives with a Class IIa CE mark under the EU Medical Device Regulation specifically covering tumor contouring.

The stakes for radiotherapy departments are concrete. Contouring — outlining the tumor target and the healthy structures clinicians must protect — sits between the scan and the dose calculation, and it is one of the most time-consuming manual steps in treatment preparation. A shorter path from scan to reviewed contours frees specialist time. It does not, by itself, shorten total treatment waiting time, improve tumor control or deliver a survival benefit, since scan acquisition, clinical decisions and dose calculation still surround the contouring task.

What does the 92.6% figure actually measure?

According to Synaptiq, the clinical evaluation assessed as part of its conformity assessment covered head-and-neck, thorax, abdomen and pelvis CT cases. The reported average 92.6% time reduction compares the clinician-reviewed workflow against fully manual contouring — an important distinction, because vendors sometimes quote inference speed without accounting for the human review that regulators require.

The anatomy-specific numbers are the sharpest. Head-and-neck cases, among the most labor-intensive to contour manually, fell from an average of 97 minutes by hand to 2.7 minutes with Mediq RT — a reported 97.2% decrease. Head-and-neck contouring demands precision across many small, adjacent structures, which is why it historically consumes disproportionate specialist time.

Clinical acceptability within the evaluation broke down as follows, according to the company:

  • 92.1% of AI-generated contours were accepted without edits
  • 7.6% needed minor edits
  • Fewer than 0.3% needed major corrections

These are measures of contouring efficiency and clinical acceptability within the evaluation's scope. Synaptiq does not claim outcome benefits beyond that.

How fast is the system, and how does correction work?

Mediq RT's initial contour processing takes approximately 30 seconds per patient, and the platform supports more than 250 organs-at-risk and lymph node regions across anatomical cancer sites. The November 18 announcement puts the model's response to interactive prompts at approximately one to two seconds.

Those three numbers describe different parts of the workflow, and Synaptiq keeps them distinct. Initial processing generates the full set of structures; interactive response time governs how quickly the AI reacts when a clinician steers it; and the 2.7-minute head-and-neck figure includes the human review on top.

The interactive component, called Active Intelligence, gives clinicians a way to refine a result without redrawing every slice. On its website, Synaptiq describes a workflow in which a radiation oncologist outlines one slice of a gross tumor volume, and the AI extends the contour through the remaining volume. Corrections on an individual slice can also guide AI-assisted revisions across the case. Clinicians retain responsibility for approving every structure.

CEO and co-founder Dragoș Dușe framed the product around that division of labor: "With Mediq RT, we're reducing the time clinicians spend on treatment planning by bringing tumour and organ contouring together in one place, while keeping clinicians in control of every decision."

How does Mediq RT compare with other commercial systems?

Synaptiq also reports a separate benchmark comparison on the same 80-patient dataset used in a 2023 study of commercial AI contouring systems. That published study evaluated five vendors across breast, head-and-neck, lung and prostate cases. Synaptiq was not one of the five systems in the original paper — the company ran its own comparison afterward, which readers should weigh accordingly.

In that comparison, Synaptiq reports a head-and-neck correction time of 2.7 minutes, against 4.4 minutes for the next-fastest system. The company's description of the evaluation discusses both geometric agreement and editing effort. The distinction matters. A contour can show strong overall overlap with a reference standard while still requiring a clinically important local correction — which is why editing time, not just overlap metrics, is the number departments care about.

Where does the software sit in the hospital stack?

Synaptiq's published intended-use documentation describes Mediq RT as a tool for trained radiation oncologists and medical physicists working with 3D CT images of adults who have an indication for radiotherapy. Qualified users review, edit and approve the structures before export.

The integration story is deliberately narrow. Mediq RT connects with treatment-planning systems and picture archiving and communication systems through DICOM, the standard used to exchange medical images and related data. A browser-based viewer provides the workspace for reviewing contours, and the platform can run as a cloud service or on hospital premises.

Mediq RT exports reviewed structures into the hospital's subsequent dose-calculation and treatment-design workflow. It does not independently prescribe or deliver radiation — a boundary consistent with its Class IIa classification and with how contouring tools fit into the broader planning chain.

What comes next for Synaptiq?

Founded in Cluj-Napoca in 2021, Synaptiq says it now serves customers in 11 countries and has processed more than 63,000 patients in real-world clinical use. The company's stated roadmap has three tracks: expanding its European presence, developing further clinical evidence, and pursuing U.S. market entry through an FDA 510(k) submission (the U.S. clearance pathway for devices substantially equivalent to existing products).

The 510(k) route signals that Synaptiq intends to compete in the U.S. market where the 2023 benchmark study's five evaluated vendors already operate. The commercial question is whether its reported editing-time advantage holds outside evaluation datasets.

The most useful promise of Mediq RT's approach is the pairing of automation with fast correction. Generating a contour in 30 seconds is valuable only when specialists can assess it, refine it efficiently and return it to the existing planning process. Adoption at the next stage will depend on how consistently that complete workflow — generation, one-to-two-second interactive response, clinician review and export — performs across varied clinical settings rather than on any single benchmark number.

Source: unite.ai

Share this article:

More from Marcus Bennett

Marcus Bennett

Show full bio

Senior reporter covering companies, deals and strategy in artificial intelligence and machine learning at AI In Context.

745 articles

Related articles

  1. TestDynamics Raises $9 Million to Evaluate Medical Imaging AI
  2. AI Estimates Cancer Survival From a Single Patient Selfie
  3. Utah Green-Lights an AI That Prescribes Acne Medication for $5 a Month
  4. Japan Clears Subtle Medical's SubtleHD AI for Faster MRI Scans
  5. Stanford Study Validates Aimbient's Automated OR Timestamps

« Previous articleNext article »