A 76-year-old woman arrived for a standard tuberculosis screening with no respiratory symptoms. Among the large volumes of chest X-rays processed that day, an AI-prioritized worklist flagged her case for closer review. It was a finding that might otherwise have gone unnoticed.

When Volume Becomes a Clinical Risk

Tuberculosis screening programs in Thailand’s public hospitals operate at scale. On a typical screening day at Lampang Hospital, radiologists review large volumes of chest X-rays from largely asymptomatic patients who came in not because they felt unwell, but because routine screening protocols require it.

Reviewing large volumes of images in a single session, most of which are unremarkable, places a sustained attention burden on radiologists. When the clinical expectation is a normal result, and patients present without symptoms, subtle or early-stage abnormalities can easily fall below the threshold of urgency. The cases that most need attention are not always the ones that naturally stand out in a crowded worklist.

This is where AI-driven worklist prioritization becomes clinically meaningful.

A Finding Hidden Across Two Studies

In February 2024, a 76-year-old woman with a chronic smoking history presented at Lampang Hospital for a pre-operative chest X-ray ahead of cataract surgery. The imaging showed mild cardiomegaly, but no definite lung nodules or masses were identified.

Eight months later, in October 2024, the same patient returned for a routine pulmonary TB screening. She remained asymptomatic. Without an AI-assisted worklist, her case would have entered the standard reading queue alongside every other study from that session.

Inspectra CXR analyzed her chest X-ray and identified a 2.8 cm nodule in the left upper lobe, near the left hilar region. The AI output included a heatmap visualization with a 76% confidence score for a nodule and 61% for a mass. Based on these findings, the system automatically elevated the case to the top of the radiologist’s worklist dashboard.

The radiologist reviewed the flagged case and proceeded with further investigation. A CT chest scan using a lung cancer protocol confirmed a mass in the anterior segment of the left upper lobe, along with multiple liver nodules consistent with lung cancer and liver metastasis. A biopsy performed three weeks later confirmed small cell carcinoma.

How AI Worklist Prioritization Supports Radiologists

In a TB screening program, the majority of cases reviewed on any given day are normal or near-normal. This is expected. But it also means radiologists spend a significant portion of their time confirming unremarkable results, while the case that genuinely requires urgent attention may not be immediately visible in the queue.

AI worklist prioritization addresses this directly. Suspicious findings are surfaced automatically and appear at the top of the reading queue with supporting visual evidence. Radiologists can direct their attention to high-priority cases first, while still reviewing the full worklist with appropriate efficiency.

The 2024 pilot program in Health Region 1, which included Lampang Hospital, applied this model to lung cancer screening triage. Using AI CXR analysis to identify patients who warranted referral for low-dose CT, the program supported faster identification of high-risk individuals in communities where access to specialist imaging would otherwise involve significant delays.

Key Takeaways for Radiology and Screening Programs

• A feeling of being well does not always mean nothing is wrong. This patient had no respiratory complaints when she came in for TB screening. The finding was only visible on imaging, not from any symptom she reported. In screening programs where most patients feel perfectly healthy, consistent AI-assisted review helps ensure that abnormalities are not overlooked simply because no one was looking for them.
• The radiologist remained in full control of every clinical decision in this case. What the AI contributed was a clear and timely signal, elevating this case to the top of the worklist and presenting supporting evidence through the confidence score and heatmap visualization. The decision to pursue further investigation was made by the physician, informed by that signal.
• High patient volume is itself a clinical challenge. When a radiologist works through a large volume of chest X-rays in a single session, the volume of normal results can make it harder to act on the ones that are not. Workflow design matters as much as clinical skill, and tools that surface the right case at the right moment make a genuine difference in patient outcomes.

Conclusion

This case illustrates an important reality of modern screening programs: the challenge is not simply detecting abnormalities, but ensuring they receive attention at the right time.

The patient presented without respiratory symptoms, and the finding was identified through routine imaging rather than clinical suspicion. In a high-volume screening environment, cases like this can be difficult to distinguish from the thousands of normal studies that pass through radiology workflows each year. As screening programs expand, the risk is not only missed detection but also missed attention.

This is where AI can provide meaningful value. By automatically flagging suspicious findings and prioritizing cases for review. The technology does not make diagnoses, determine treatment plans, or replace medical expertise. Those decisions remain the responsibility of qualified clinicians.

Instead, AI functions as a decision support tool that helps radiologists focus their attention where it may be needed most. In this case, the final diagnosis depended on radiologist’s interpretation, CT imaging, and pathological confirmation. AI contributed by identifying a potential abnormality early in the workflow and ensuring it received timely review.

Learn more about Inspectra CXR: https://perceptra.tech/inspectra-cxr/