What does this research mean for the field?

A segmentation-guided machine learning model utilizing peritumoral radiomic features can effectively differentiate between benign and malignant salivary gland tumors on ultrasound. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To develop and validate a segmentation-guided machine learning framework utilizing peritumoral radiomic features to differentiate benign from malignant salivary gland tumors on ultrasound.

September 8, 2026Acta Oto-Laryngologica

Segmentation-guided peritumoral radiomics and machine learning for ultrasound-based classification of salivary gland tumors

View Full Paper

Ask AI

Bookmark

Authors

HCHung-Wei Chen CCChin-Huan Chang WHWei-Chen Hung

Discussion

Member takes

Overview

Diagnostic modeling study demonstrates accurate classification of salivary gland tumors using ultrasound radiomics, indicating clinical utility for preoperative noninvasive decision-making.

Key Points

To develop and validate a segmentation-guided machine learning framework utilizing peritumoral radiomic features to differentiate benign from malignant salivary gland tumors on ultrasound.
Trained a YOLOv8 deep learning model for automated tumor segmentation and extracted peritumoral radiomic features from ultrasound scans.
Screened and selected 12 radiomic features to train machine learning classifiers, evaluating them on testing and validation cohorts (validation n = 51, collected January–December 2025).
Constructed and deployed the full diagnostic pipeline into standalone and web-based applications.
The YOLOv8 segmentation model achieved Dice coefficients of 0.9485 in the training cohort and 0.8896 in the testing cohort.
Support Vector Classifier emerged as the best-performing model, yielding an AUC of 0.8117 in the testing cohort.
In the external validation cohort (n = 51), the Support Vector Classifier maintained an AUC of 0.7951.

Cite This Study

Chen et al. (2026) studied this question.

synapsesocial.com/papers/6a9fd80458e84d0ff5b4730d https://doi.org/10.1080/00016489.2026.2722468

View Full Paper

Ask AI

Bookmark

September 8, 2026Acta Oto-Laryngologica

Segmentation-guided peritumoral radiomics and machine learning for ultrasound-based classification of salivary gland tumors

View Full Paper

Ask AI

Bookmark

Authors

HCHung-Wei Chen CCChin-Huan Chang WHWei-Chen Hung

Discussion

Member takes

Overview

Diagnostic modeling study demonstrates accurate classification of salivary gland tumors using ultrasound radiomics, indicating clinical utility for preoperative noninvasive decision-making.

Key Points

To develop and validate a segmentation-guided machine learning framework utilizing peritumoral radiomic features to differentiate benign from malignant salivary gland tumors on ultrasound.
Trained a YOLOv8 deep learning model for automated tumor segmentation and extracted peritumoral radiomic features from ultrasound scans.
Screened and selected 12 radiomic features to train machine learning classifiers, evaluating them on testing and validation cohorts (validation n = 51, collected January–December 2025).
Constructed and deployed the full diagnostic pipeline into standalone and web-based applications.
The YOLOv8 segmentation model achieved Dice coefficients of 0.9485 in the training cohort and 0.8896 in the testing cohort.
Support Vector Classifier emerged as the best-performing model, yielding an AUC of 0.8117 in the testing cohort.
In the external validation cohort (n = 51), the Support Vector Classifier maintained an AUC of 0.7951.

Cite This Study

Chen et al. (2026) studied this question.

synapsesocial.com/papers/6a9fd80458e84d0ff5b4730d https://doi.org/10.1080/00016489.2026.2722468

View Full Paper

Ask AI

Bookmark