AI in MF Diagnosis

Evaluation of Artificial Intelligence Algorithms Performance in the Histopathological Diagnosis of Mycosis Fungoides

The aim of this observational study is to evaluate the diagnostic performance of an AI algorithm in the histopathological diagnosis of MF compared to certified dermatopathologists.

Mycosis fungoides (MF) is the most common form of primary cutaneous T-cell lymphoma. Its early histological features may overlap with benign inflammatory dermatoses, making diagnosis challenging.

This observational study aims to evaluate the diagnostic performance of HistoGPT in the histopathological diagnosis of MF compared with certified dermatopathologists.

H&E-stained skin biopsy slides will be digitized using a Leica Aperio GT450 whole-slide scanner at 40× magnification.

The resulting whole-slide images will be analyzed using HistoGPT, an AI-based histopathology platform.

The diagnostic performance of HistoGPT and certified dermatopathologists will be assessed and compared using appropriate diagnostic metrics, including accuracy, sensitivity, specificity, F1 score, and area under the ROC curve.

The findings of this study will help determine whether Artificial intelligence can serve as a diagnostic support tool for the histopathological diagnosis of mycosis fungoides.

Inclusion Criteria:

• Slides will be included in the study if they meet the following criteria:

  • Histopathological slides diagnosed as MF.
  • Slides with adequate staining and preservation allowing clear visualization of histopathological features.

Exclusion Criteria:

• Slides will be excluded if they meet any of the following criteria:

  • Slides with poor staining quality or significant artifacts interfering with histopathological interpretation.
  • Slides that were damaged, faded, or inadequately preserved.
  • Slides with uncertain or inconclusive original diagnoses.
  • Slides that could not be successfully digitized due to technical limitations ex very short or too long slides.