Manual Versus AI X-Ray Annotation: A Workflow Checklist

1. Confirm the study before choosing a method

Check the examination you intend to review, the available views, and the region included in the image. Confirm orientation and whether the image can support the intended measurement. Keep any preparation required by the clinic's imaging workflow separate from the AI function. Uploading an image successfully is not the same as verifying that a particular analysis is suitable for it.

This checklist starts after a clinically appropriate examination has been obtained. It does not recommend taking an X-ray solely to use software. It also does not establish compatibility with every imaging system. For transfer questions, confirm the supported integration and configuration or the manual upload process with the product team.

2. Identify the task, not just the tool name

Write down the question you need a measurement or annotation to address. Then choose the documented tool or AI model. A manual Cobb Angle control is a method operated by the reviewer; its presence does not mean every Cobb measurement is automatically generated. An AI model's supported outputs do not convert the remainder of the manual inventory into automated functions.

Use the Cobb Angle Measurement Guide when reviewing that manual method. Confirm the relevant definitions and reference points before interpreting a numerical result. Do not substitute a similarly named control merely because it produces an angle in the same units.

3. Select the supported analysis

AP Lumbar Analysis, AP Pelvis Analysis, and Full Lateral Spine Analysis have separate documented descriptions. Check the selected model against the image and intended task. Do not assign one model's measurements to another view. The Gonstead menu groups access to the three AI models and manual tools; it is not an additional AI model.

In a demonstration, ask the team to identify which action launches the model and which controls are manual. Distinguish a saved annotation, an AI result, and an observation entered during report editing. They can appear in the same workspace without having the same origin or meaning.

4. Inspect the underlying image and reference points

Review the result alongside the image rather than reading only a number or label. Check whether displayed reference points and lines match the anatomy and method you intended to evaluate. If an image area is obscured or the displayed result is not suitable for your task, make that limitation explicit instead of silently accepting the output.

Review is a deliberate part of either workflow. Before moving to reporting, check the displayed method, the region assessed, and any limitations you identified. Keep an additional manual observation distinguishable from the selected model's output so that the final record reflects what was actually reviewed.

5. Address corrections without assuming an edit function

When you need an additional measurement or annotation, use the appropriate documented manual tool. Confirm which outputs can be adjusted, rerun, supplemented, or excluded in the current interface. The tool catalog alone does not establish that every AI-generated landmark or measurement is editable. Ask about the exact output you need to change.

Keep the distinction between a software result and your own added observation clear in the working record. Check relevant values again after any supported change. Avoid assuming that editing a report paragraph also changes the underlying image annotation, or that a corrected annotation automatically updates every exported document.

6. Check comparisons and report content

The documented reporting workflow supports current findings in Initial Report and comparisons in Progress Report when comparable previous measurements are available. Confirm the examinations and measurements actually used in a comparison. The availability of an older image alone is not a guarantee that a valid automated comparison exists.

Review supported report sections, patient and provider details, annotations, and observations before final approval. Manage the intended signature image deliberately. Export the completed PDF only after the review steps required by your workflow. Do not describe the signature image as independently verified digital signing or promise integration with an external note system.

7. Record the workflow you actually used

For an internal review, note the selected model or manual method, the outputs inspected, and any limitation that affected the record. If evaluating efficiency, observe the whole workflow rather than importing a processing-time claim into a clinical visit estimate. The purpose is to understand your process, not to promise a saving or a patient outcome.

Read the AI software evaluation guide for a separate purchasing and demonstration framework. Contact ChiroPax for confirmation of a particular edit or export behavior before including it in a clinic procedure.