AutoIHC-Analyzer: computer-assisted microscopy for automated membrane extraction/scoring in HER2 molecular markers

Area of Science:

• Biomedical image analysis
• Computational pathology
• Cancer diagnostics

Background:

• Human epidermal growth factor receptor 2 (HER2) is a crucial biomarker for breast cancer prognosis and therapeutic decisions.
• Current manual HER2 scoring via Immunohistochemistry (IHC) is time-consuming, subjective, and prone to interobserver variability.
• The increasing patient load necessitates efficient and objective methods for HER2 assessment.

Purpose of the Study:

• To develop an automated IHC scoring system, AutoIHC-Analyzer, for rapid and accurate HER2 cell membrane extraction and molecular expression assessment.
• To address the limitations of manual HER2 scoring, including time, tedium, and interobserver variability.

Main Methods:

• Automated cell and membrane region extraction using image processing techniques.
• Quantification of complete versus broken cell membranes.
• Classification of HER2 expression levels (0/1+, 2+, 3+) using a Support Vector Machine (SVM) classifier trained on six features.

Main Results:

• The AutoIHC-Analyzer achieved 88.3% accuracy in training samples.
• Significant correlation was observed between AutoIHC-Analyzer scores and expert pathologist scores (r = 0.9448, p < 0.001).
• The system demonstrated comparable performance to the ImmunoMembrane software (r = 0.8521, p < 0.001).

Conclusions:

• AutoIHC-Analyzer provides a promising automated solution for quantitative HER2 scoring in breast cancer.
• The system can significantly improve the efficiency and objectivity of HER2 assessment in clinical practice.
• Automated HER2 scoring has the potential to aid pathologists in therapeutic decision-making.