Applying an artificial intelligence deep learning approach to routine dermatopathological diagnosis of basal cell carcinoma

Abstract

SummaryBackgroundInstitutes of dermatopathology are faced with considerable challenges including a continuously rising numbers of submitted specimens and a shortage of specialized health care practitioners. Basal cell carcinoma (BCC) is one of the most common tumors in the fair‐skinned western population and represents a major part of samples submitted for histological evaluation. Digitalizing glass slides has enabled the application of artificial intelligence (AI)‐based procedures. To date, these methods have found only limited application in routine diagnostics. The aim of this study was to establish an AI‐based model for automated BCC detection.Patients and MethodsIn three dermatopathological centers, daily routine practice BCC cases were digitalized. The diagnosis was made both conventionally by analog microscope and digitally through an AI‐supported algorithm based on a U‐Net architecture neural network.ResultsIn routine practice, the model achieved a sensitivity of 98.23% (center 1) and a specificity of 98.51%. The model generalized successfully without additional training to samples from the other centers, achieving similarly high accuracies in BCC detection (sensitivities of 97.67% and 98.57% and specificities of 96.77% and 98.73% in centers 2 and 3, respectively). In addition, automated AI‐based basal cell carcinoma subtyping and tumor thickness measurement were established.ConclusionsAI‐based methods can detect BCC with high accuracy in a routine clinical setting and significantly support dermatopathological work.

Artificial intelligence and skin cancer

“…Duschner et al applied AI to automated diagnosis of BCCs, and demonstrated both sensitivity and specificity of over 98%. Notably, the model demonstrated successful generalization to samples from other centers with similar sensitivity and specificity ( 54 ).…”

Digitalisierung und Einsatz von Künstlicher Intelligenz in der Dermatopathologie

“…Eine gute Übersicht über diese Studien findet sich in der Arbeit von Duschner et al. in dieser Ausgabe 6 . Eine wichtige Limitierung dieser PoC‐Studien ist allerdings, dass sie meist monozentrisch erfolgten und die Genauigkeit des Modells anhand eines Testdatensatzes errechnet wird, der zuvor vom Gesamtdatensatz separiert wurde.…”

“…wurde die Genauigkeit eines zuvor in einer PoC‐Studie trainierten Modells, erstmalig an einem nicht‐selektierten Datensatz aus der Routinediagnostik untersucht. Das KI‐System erzielte dabei eine hohe Sensitivität von 98.23 % und eine Spezifität von 98.51 %, vergleichbar mit den Werten aus der PoC‐Studie 6,7 . Besonders erwähnenswert ist, dass das Modell ohne weiteres Training eine ähnliche Genauigkeit bei der Analyse von WSI aus zwei anderen Laboren erzielte.…”