Skip to content

Assessment performance of large language models (LLMs) for standardized, semantically grounded evaluation of flap perfusion using polarized dermoscopic images

Assessment performance of large language models (LLMs) for standardized, semantically grounded evaluation of flap perfusion using polarized dermoscopic images - LLM-FLAP-DERM

Status
Active, not recruiting
Phases
Unknown
Study type
Interventional
Source
DRKS
Registry ID
DRKS00039191
Enrollment
50
Registered
2026-02-02
Start date
2026-03-01
Completion date
Unknown
Last updated
2026-03-02

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Postoperative perfusion monitoring of microvascular free flaps in head and neck reconstruction (early detection of perfusion impairment/free-flap complications after free tissue transfer).

Interventions

Group 1: Dermoscopy images of four flap regions (central, intermediate, peripheral, and one clinician-defined “critical” region) at predefined postoperative time points
perfusion assessment using an LLM-based standardized scoring system. Group 2: As in Arm A, plus acquisition of a reference region (contralateral skin at/near the donor site) at the same time points
LLM-based scoring incorporating the reference region.

Sponsors

Universitätsklinik und Poliklinik fürMund-, Kiefer- und Plastische Gesichtschirurgie
Lead Sponsor

Eligibility

Sex/Gender
All
Age
18 Years to No maximum

Inclusion criteria

Inclusion criteria: - Patients undergoing microvascular free flap reconstruction (myocutaneous/fasciocutaneous/osteomyocutaneous) with a skin paddle in the head and neck region. - Planned postoperative perfusion monitoring at the study site with feasibility to obtain the scheduled dermoscopic image acquisitions at predefined time points. -Written informed consent for study participation and pseudonymized image data acquisition/processing.

Exclusion criteria

Exclusion criteria: - No informed consent or withdrawal of consent / refusal to participate. - Inability to provide informed consent / lack of legal capacity (without permissible proxy consent per ethics approval). - Objection to scientific use of collected data/images. - Severe microangiopathy or relevant skin disease at/near the donor site (reference region) substantially limiting comparability.

Design outcomes

Primary

MeasureTime frame
Diagnostic/prognostic performance (ROC AUC) of the LLM total score (0–10) to predict clinically relevant perfusion impairment/complications of the microvascular free flap up to postoperative day 7.

Secondary

MeasureTime frame
- Agreement between the LLM score and the clinician perfusion score (e.g., ICC; optionally ? for categorical ratings). - Predictive value of individual flap regions (central/intermediate/peripheral/critical) for subsequent perfusion impairment/complications. - Predictive value of individual score components (e.g., color, hemoglobin component, capillary refill, temperature, healing status) for subsequent perfusion impairment/complications. - Added value of a reference region: comparison of predictive performance of LLM scoring with vs. without reference skin (e.g., ?AUC / model comparison).

Countries

Germany

Contacts

Public ContactFelix Tilsen

Universitätsklinik und Poliklinik fürMund-, Kiefer- und Plastische Gesichtschirurgie

felix.tilsen@uk-halle.de+493455575244

Outcome results

None listed

Source: DRKS (via WHO ICTRP) · Data processed: Mar 14, 2026