Intraloop neoangiogenesis in an AI-classified scleroderma pattern: recognizing a morphological clue to suspected dermatomyositis.

OBJECTIVES: To describe a diagnostic near miss in AI-assisted nailfold capillaroscopy, illustrating how a correct AI-based classification as an early scleroderma pattern may still lead to diagnostic error if not integrated with additional morphological findings, such as intraloop neoangiogenesis, which should prompt consideration of dermatomyositis. CASE PRESENTATION: A 50-year-old woman with episodic acrocyanosis underwent nailfold videocapillaroscopy (NVC). AI-assisted analysis classified the pattern as "early scleroderma". Expert review identified branching neoangiogenesis within the same capillary loop as a giant capillary. This composite finding is not typical of systemic sclerosis microangiopathy and lies outside the scope of current algorithms, which are trained on scleroderma-based patterns and are not designed to recognize alternative microangiopathic processes. Accordingly, the system did not flag this feature as diagnostically relevant. This prompted immunological testing, which showed anti-Mi-2 positivity on repeat testing, with negative systemic sclerosis-specific autoantibodies, raising suspicion for dermatomyositis. CONCLUSIONS: AI-assisted capillaroscopy provides accurate pattern classification within its training framework but does not replace integrative morphological reasoning. When dominant features such as giant capillaries drive classification, clinicians must assess internal consistency and consider additional findings beyond pattern assignment. Neoangiogenesis is a diagnostically relevant feature that, in the appropriate context, should prompt consideration of dermatomyositis. Diagnostic error reflects not algorithm failure but limits of training scope and over-reliance on pattern-based outputs. Expert interpretation remains essential.