# Nonenhanced CT-Based radiomics model enhances PTC detection in Hashimoto’s thyroiditis

## TL;DR

This study shows that a CT-based radiomics model can help detect thyroid cancer in patients with Hashimoto's thyroiditis, improving early diagnosis.

## Contribution

A novel nonenhanced CT-based radiomics model is developed to detect papillary thyroid carcinoma in Hashimoto’s thyroiditis patients.

## Key findings

- The MLP model achieved the highest AUC of 0.783 in the external validation cohort.
- The model showed a sensitivity of 0.643 and specificity of 0.923 in detecting PTC.
- Six radiomic features were selected using LASSO for the machine learning models.

## Abstract

Hashimoto's thyroiditis (HT) is a common benign thyroid disease that often coexists with papillary thyroid carcinoma (PTC). Owing to the diffuse changes in the thyroid caused by HT, PTCs can be challenging to detect using conventional imaging modalities such as ultrasound and CT. The aim of this study was to develop a radiomics model based on nonenhanced CT (NECT) to predict the presence of PTC in the patients with HT, thereby improving early diagnostic accuracy.

This retrospective study included pathologically confirmed HT patients with or without PTC who underwent NECT scans within 30 days before surgery from January 2017 to April 2023 at Hospital I and Hospital II. The patients from hospital I were divided randomly at a ratio of 8:2 into a training cohort and an internal validation cohort. The patients from hospital II were assigned to the external validation cohort. Radiomic features were extracted using PyRadiomics. Intraclass correlation coefficient, Pearson correlation and LASSO analyses were conducted to reduce the dimensionality of the radiomic features. Four machine learning algorithms, including logistic regression (LR), naive bayes (NB), support vector machine (SVM), and multilayer perceptron (MLP) classifiers, were employed to develop and validate the prediction models based on the remaining features.

A total of 130 patients, 89 from Hospital I [71 in the training cohort and 18 internal validation cohort] and 41 from Hospital II [external validation cohort], were included. Six features with nonzero coefficients were retained by the LASSO algorithm for inclusion in the machine learning models. In the external validation cohort, the LR, NB, SVM, and MLP models obtained AUCs of 0.736, 0.690, 0.751 and 0.783, respectively. The MLP model performed the best in the external validation cohort, with an area under the curve of 0.783, a sensitivity of 0.643, and a specificity of 0.923.

A radiomics model based on NECT could identify PTCs in patients with HT and had the potential to enhance early diagnosis and intervention for these patients.