Research

Dentistry that does research.

Peer-reviewed publications on AI-assisted diagnostics in endodontics, together with Prof. Dr. Falk Schwendicke, one of the world's most-cited researchers on artificial intelligence in dentistry.

Why a dental practice does research

Most clinical recommendations are only as good as the data they rest on. In endodontics and AI diagnostics, that data is moving faster than the textbooks. To know what works, not what merely sounds good, we have to read the studies, contextualise them, and ideally co-author them. The work below was produced during the ongoing doctoral project in Prof. Dr. Falk Schwendicke's research group and is indexed in PubMed or Scopus.

Publications

Journal of Dentistry · 2024Systematic review & meta-analysis

Artificial intelligence for detecting periapical radiolucencies: a systematic review and meta-analysis

Pul U, Schwendicke F.

This systematic review pools all studies published up to November 2023 on the diagnostic accuracy of AI for detecting periapical radiolucencies, inflammatory changes at the root apex. Such findings are commonly overlooked in clinical practice and are the primary indication for root canal treatment. The meta-analysis shows that modern AI systems already reach detection rates comparable to experienced clinicians, making them well suited as a second diagnostic layer.

What this means for you: Why it matters for patients: when a lesion is detected early, the tooth can more often be preserved, without an implant. This validates our practice approach: your own tooth before replacement.

Full text (DOI) →DOI: 10.1016/j.jdent.2024.105104PMID: 38851523

Journal of Dentistry · 2025Randomized controlled trial

Impact of artificial intelligence assistance on diagnosing periapical radiolucencies: a randomized controlled trial

Pul U, Tichy A, Pitchika V, Schwendicke F.

In this randomized controlled trial we examined what an AI second read actually does in practice, on diagnostic accuracy, clinician confidence, and the subsequent treatment decision. Clinicians were randomized into two groups: reading alone, or with AI assistance. The result confirms what the prior review suggested: with AI assistance, detection rates for periapical radiolucencies rise measurably, especially among less experienced clinicians.

What this means for you: Why it matters for patients: this work shows not only whether AI works, but under what conditions it concretely helps patients. That is exactly why we use AI-assisted diagnostics in the practice, as a second safety net, not as a replacement for clinical judgement.

Full text (DOI) →DOI: 10.1016/j.jdent.2025.105868

International Endodontic Journal · 2026Health-economic modeling study

Cost-effectiveness of AI-assisted detection of apical periodontitis on panoramic radiographs

Benz L, Pul U, Brock T, Schwendicke F, Walter E.

In this health-economic modeling study we quantify the effect of an AI second opinion on the costs and outcomes of treating apical periodontitis, including the downstream clinical decision-making. The AI helps avoid missed findings, which reduces unnecessary extractions and downstream costs (implant, bone graft, suprastructure). The benefit is most pronounced among less experienced clinicians; adapting AI systems to individual examiners or experience levels could further enhance the clinical and economic impact.

What this means for you: Why it matters for patients: correct diagnostics avoids expensive downstream chains. A properly executed root canal is on average cheaper than the implant chain (extraction, bone graft, implant, crown) that follows when a tooth is wrongly considered lost.

Full text (DOI) →DOI: 10.1111/iej.70142PMID: 41826269

Research meets practice.

In the practice we use the same AI-assisted tools we examine in our publications, as a second opinion alongside the clinical finding, never as a replacement.