Quantitative grading of rat esophageal carcinogenesis using computer-assisted image tile analysis

Area of Science:

• Oncology
• Biotechnology
• Medical Imaging

Background:

• Esophageal intraepithelial neoplasia (dysplasia) is a precursor to cancer.
• N-nitrosomethybenzylamine (NMBA) is used to induce esophageal carcinogenesis in rat models.
• Quantitative image analysis offers objective methods for evaluating neoplastic changes.

Purpose of the Study:

• To grade esophageal intraepithelial neoplasia using computer-assisted quantitative image tile analysis.
• To validate grading parameters for screening chemopreventive agents in a rat model.
• To assess the sensitivity and objectivity of the developed grading method.

Main Methods:

• Computer divides neoplastic epithelium images into small tiles for analysis.
• Quantitatively measures four tissue features within each tile.
• Calculates a tile grade using weighted feature measurements and Z-scores.
• Determines weights via Fisher linear discriminant analysis comparing neoplastic and normal epithelium.

Main Results:

• Two grading parameters, mean tile grade (MTG) and %TG>4SD, were validated.
• %TG>4SD correlated with tumor incidence and multiplicity after NMBA treatment.
• The chemopreventive agent phenethylisothiocyanate reduced MTG, tumor incidence, and multiplicity.

Conclusions:

• Computer-assisted quantitative image tile analysis provides sensitive and objective grading of esophageal neoplasia.
• MTG and %TG>4SD are effective parameters for evaluating chemopreventive agents in the NMBA-induced rat model.
• This method allows for precise comparison of neoplastic changes before and after treatment.