Artificial Intelligence in Thyroid Cytopathology: Diagnostic and Technical Insights

Abstract

Fine-needle aspiration cytology (FNAC) is the cornerstone of thyroid nodule evaluation, standardized by the Bethesda System. However, indeterminate categories (Bethesda III-IV) remain a major challenge, often leading to unnecessary surgery or delayed molecular testing. Deep learning (DL) has recently emerged as a promising adjunct in thyroid cytopathology, with applications spanning triage support, Bethesda category classification, and integration with molecular data. Yet, routine adoption is limited by preanalytical variability (staining, slide preparation, Z-stack acquisition, scanner heterogeneity), annotation bias, and domain shift, which reduce generalizability across centers. Most studies remain retrospective and single-institution, with limited external validation. This article provides a technical overview of DL in thyroid cytology, emphasizing preanalytical sources of variability, architectural choices, and potential clinical applications. We argue that standardized datasets, multicenter prospective trials, and robust explainability frameworks are essential prerequisites for safe clinical deployment. Looking forward, DL systems are most likely to enter practice as diagnostic co-pilots, Bethesda classifiers, and multimodal risk-stratification tools. With rigorous validation and ethical oversight, these technologies may augment cytopathologists, reduce interobserver variability, and help transform thyroid cytology into a more standardized and data-driven discipline.

Keywords: Bethesda system; artificial intelligence; convolutional neural networks; deep learning; explainable AI; molecular prediction; multimodal models; multiple instance learning; thyroid cytology.

Figure 1

Deep learning workflow in thyroid cytology. The diagram outlines the end-to-end pipeline across four phases. Pre-analytical (laboratory): cytological sample preparation with common stains (Diff-Quik, Papanicolaou, H&E). Pre-analytical (digital): slide digitization at 20×/40× magnification, with or without Z-stack acquisition (improved focal coverage vs increased time, file size, and computational load). Analytical: preprocessing (artifact removal, color normalization, white-background rejection), followed by model training/inference using supervised (ROI-based) or weakly supervised (slide-level/MIL/attention) strategies. Evaluation and post-analytical: performance assessment with standard metrics (accuracy, precision, recall, F1-score, AUC) and integration of explainability tools (e.g., Grad-CAM, SHAP) to support interpretability. Clinical outputs: triage of indeterminate nodules, diagnostic decision support (e.g., Bethesda categorization), and integrated reporting. Detailed assumptions and trade-offs for each block are discussed in Section 2, Section 3 and Section 4.