ARTIFICIAL INTELLIGENCE–ASSISTED DIGITAL HISTOPATHOLOGICAL DETECTION OF CANCER STEM CELL NICHES IN ORAL SQUAMOUS CELL CARCINOMA

Abstract

Background: Cancer stem cells play an important role in tumor progression, invasion, metastasis, and treatment resistance in oral squamous cell carcinoma. Artificial intelligence–assisted digital histopathology offers a promising approach for automated detection of complex tumor microenvironmental patterns, including cancer stem cell niches.

Objective: To evaluate the diagnostic performance and clinicopathological correlation of artificial intelligence–assisted digital histopathological detection of cancer stem cell niches in oral squamous cell carcinoma

Methods: This cross-sectional analytical study was conducted at AI-assisted digital histopathology labs across Karachi, from March 2023 to September 2025, including 190 patients with histopathologically confirmed oral squamous cell carcinoma.

Results: The mean age was 54.2 ± 11.8 years, with 69.5% males. AI-assisted analysis demonstrated sensitivity of 89.4%, specificity of 84.7%, positive predictive value of 82.6%, negative predictive value of 90.8%, and overall diagnostic accuracy of 87.1%, with strong agreement with expert histopathologists (kappa = 0.79). AI-detected CSC niches were identified in 104 (54.7%) patients and were significantly associated with larger tumor size (4.4 ± 1.4 vs. 3.1 ± 1.2 cm; p<0.001), lymph node metastasis (60.6% vs. 26.7%), poorly differentiated histology (36.5% vs. 12.8%), perineural invasion (40.4% vs. 18.6%), and advanced pathological stage (66.3% vs. 33.7%). Tumor size >4 cm was the strongest independent predictor of CSC niche positivity (aOR 4.36; p<0.001).

Conclusion: Artificial intelligence–assisted digital histopathology demonstrates strong diagnostic performance for cancer stem cell niche detection and shows significant correlation with aggressive clinicopathological features in oral squamous cell carcinoma.