# Radiomic Analysis as a Powerful Tool for Cytological Images of Benign Thyroid Nodules Treated by Thermal Radiofrequency Ablation

## TL;DR

This study explores how radiomic and cytological analysis of thyroid nodule images can predict their response to thermal radiofrequency ablation treatment.

## Contribution

The study introduces a combined radiomic, chromatic, and morphological analysis approach for predicting thyroid nodule response to RFA.

## Key findings

- Chromatic analysis successfully identified most separated nuclei with only 5% remaining unrecognized.
- Radiomic analysis showed strong links between nuclear shape and texture features.
- PCA highlighted the importance of texture and first-order features in cytological heterogeneity.

## Abstract

(1) Background: Thermal radiofrequency ablation (RFA) is an innovative treatment for benign thyroid nodules. This study aims to identify morphological and texture-based cytological parameters through radiomic and cytological analysis of fine-needle aspiration cytology (FNAC) images to support the prediction of the nodules’ response to RFA. (2) Methods: The study, conducted in collaboration with EOC—Ente Ospedaliero Cantonale (Lugano, Switzerland), analyzed FNAC images from three patients with benign thyroid nodules treated with RFA. Radiomic features were extracted in PyRadiomics and analyzed through Principal Component Analysis (PCA). A MATLAB (R2024b)-based workflow was implemented for automated chromatic and morphological analysis. (3) Results: Chromatic Analysis correctly identified separated nuclei with approximately 5% remaining unrecognized. Radiomics revealed robust connections between nuclear shape descriptors and texture-based features, showing the potential of a combined morphological-radiomic approach. PCA indicated that texture and first order features played a significant role in cytological heterogeneity. (4) Conclusions: A combination between radiomics, chromatic, and morphological analysis provides a deeper understanding of thyroid nodule characteristics. By capturing texture and intensity variations often missed by traditional methods, radiomics may enhance prediction of post-RFA behavior. The proposed methodology provides a foundation for predictive models of Volume Reduction Ratio (VRR), improving personalized diagnosis, treatment planning, and follow-up.