1. Define the task and eligible examinations

ClearRead CT K161201 illustrates why exact labeling matters: its pulmonary-nodule reading aid concerns an asymptomatic population, requires both lungs in view and accompanies the original CT series. This example does not establish another version's scope or your site's eligibility.

  • Are you evaluating nodule detection, another chest CT finding, or a change to reading workflow? State one primary job.
  • Which exact submission, software version, population, acquisition protocol and input series match that job? Record exclusions before counting eligible examinations.
  • Can the supplier reconcile the proposed licensed version and modules with the decision materials and current labeling?

Source context: FDA K161201: ClearRead CT

2. Walk through the reader and IT handoffs

Use the labeled requirement to retain the original CT as the starting point for a workflow walkthrough. A DICOM conformance statement helps assess interfaces; it does not replace a test of your installed route and viewer.

  • Where do original and derived series appear, and who checks patient, examination and series matching before interpretation?
  • What happens when a study is incomplete, unsupported, delayed, duplicated or reprocessed? Identify the owner and the visible failure state.
  • What installed PACS or viewer build, routing configuration and output objects will the local acceptance test cover?

Source context: FDA K161201: ClearRead CT · DICOM PS3.2: Scope and Field of Application

3. Agree on evidence and a stop rule

FDA public materials describe a particular authorized device. Use them to frame questions about the tested task, then define your own evaluation with clinical oversight. The directory does not supply a validated local endpoint or an expected improvement.

  • Which source reports the exact tested population, comparator, endpoint and software version? Keep those together when discussing a result.
  • Who will assess missed findings, extra findings, review time and workflow failures in the proposed local evaluation? Set the sample and acceptance rule before starting.
  • What would stop the evaluation, and how would the team restore the prior workflow? Add ownership and unresolved questions to the committee packet.

Source context: FDA list of artificial intelligence enabled medical devices

4. Work a fictional evaluation

A fictional hospital team asks how to evaluate pulmonary-nodule reading support for asymptomatic chest CT examinations. No volume, patient data or expected performance is assumed. Synthetic buyer inputs only. No product winner, clinical approval, offer or local acceptance is implied. Use the worked links below.

  1. Define the task. Keep pulmonary-nodule support distinct from another chest finding. Use K161201 as an exact labeling example, not a recommendation or current offer.
  2. Inspect identity and eligibility. Open its FDA identity, label constraints and evidence-context card. Write the relevant input/population exclusions before estimating an eligible cohort.
  3. Compare only a matching job. A comparison against a calcium-assessment record would expose a task mismatch, not identify a winner. Keep an unavailable comparator unavailable.
  4. Keep questions open. Ask who verifies original/derived image matching, routing, failures and the installed viewer. A paper count or a platform listing does not close local evidence or integration questions.
  5. Prepare a private committee draft. Open the exact-record checklist and packet, record unresolved evidence/integration questions in the private action register, then inspect the qualified draft before downloading. No clinical or purchasing approval is implied.

Source context: FDA K161201: ClearRead CT · ACR ARCH-AI: governance and implementation context · ACR Assess-AI: clinical performance-monitoring context

Take the next step

Inspect the exact example records

Examples illustrate the questions above. They are not a ranking, equivalent alternatives, confirmed offers or the complete market.

Explore a private interpretation-capacity and cost scenario · Read a fictional committee packet · Check system compatibility scope

The worksheets and scenario use your assumptions. Save or attach private inputs explicitly; public evidence does not turn those inputs into verified results. The public sample uses fictional facility values.

Sources and scope

Sources anchor the stated regulatory or interoperability context. The questions are original editorial prompts; a citation does not validate a financial assumption or establish local fit. Source dates below apply to this guide, not the whole directory.

  1. FDA K161201: ClearRead CT

    PDF pages 3-4: indications for use and device description. Checked .

    Exact K161201 labeling context only; no current offer or local acceptance is established.

  2. DICOM PS3.2: Scope and Field of Application

    Chapter 1; page identifies edition 2026d at review. Checked .

    Conformance statement context. The current URL can change; it does not qualify a particular integration.

  3. FDA list of artificial intelligence enabled medical devices

    About this list and public decision summaries. Checked .

    Regulatory source context. The list is not comprehensive and public summaries do not include all submission information.

  4. ACR ARCH-AI: governance and implementation context

    Requirements At a Glance: governance team, inventory, testing and workflow integration. Checked .

    General governance resource only. Linking does not certify this directory, a device or a local site; no affiliation or endorsement.

  5. ACR Assess-AI: clinical performance-monitoring context

    Real World Data and monitoring: model inputs, versioning and concordance. Checked .

    General post-deployment monitoring resource only. No participation, coverage for an example device, certification, affiliation or endorsement is claimed.

Dataset and maintenance dates · Methods · Submit a source-backed correction