Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Doppler-Enhanced Deep Learning: Improving Thyroid Nodule Segmentation with YOLOv5 Instance Segmentation

Mahmoud El Hussieni

TL;DR

The study tackles the challenge of accurate thyroid nodule segmentation in ultrasound by evaluating multiple YOLOv5-based instance segmentation variants, with and without doppler imaging. It demonstrates that doppler data significantly enhances segmentation across model sizes, with YOLOv5-Large achieving the highest Dice (0.91) and mAP (0.87) on the doppler-inclusive dataset, while smaller models remain viable for resource-constrained deployment. The work highlights real-time applicability, model-size trade-offs, and the surprising value of doppler information in automated delineation, suggesting practical pathways for AI-assisted thyroid cancer workflows. Limitations include single-institution data and lack of augmentation, indicating the need for broader validation and data diversity to generalize findings further.

Abstract

The increasing prevalence of thyroid cancer globally has led to the development of various computer-aided detection methods. Accurate segmentation of thyroid nodules is a critical first step in the development of AI-assisted clinical decision support systems. This study focuses on instance segmentation of thyroid nodules using YOLOv5 algorithms on ultrasound images. We evaluated multiple YOLOv5 variants (Nano, Small, Medium, Large, and XLarge) across two dataset versions, with and without doppler images. The YOLOv5-Large algorithm achieved the highest performance with a dice score of 91\% and mAP of 0.87 on the dataset including doppler images. Notably, our results demonstrate that doppler images, typically excluded by physicians, can significantly improve segmentation performance. The YOLOv5-Small model achieved 79\% dice score when doppler images were excluded, while including them improved performance across all model variants. These findings suggest that instance segmentation with YOLOv5 provides an effective real-time approach for thyroid nodule detection, with potential clinical applications in automated diagnostic systems.

Doppler-Enhanced Deep Learning: Improving Thyroid Nodule Segmentation with YOLOv5 Instance Segmentation

TL;DR

The study tackles the challenge of accurate thyroid nodule segmentation in ultrasound by evaluating multiple YOLOv5-based instance segmentation variants, with and without doppler imaging. It demonstrates that doppler data significantly enhances segmentation across model sizes, with YOLOv5-Large achieving the highest Dice (0.91) and mAP (0.87) on the doppler-inclusive dataset, while smaller models remain viable for resource-constrained deployment. The work highlights real-time applicability, model-size trade-offs, and the surprising value of doppler information in automated delineation, suggesting practical pathways for AI-assisted thyroid cancer workflows. Limitations include single-institution data and lack of augmentation, indicating the need for broader validation and data diversity to generalize findings further.

Abstract

The increasing prevalence of thyroid cancer globally has led to the development of various computer-aided detection methods. Accurate segmentation of thyroid nodules is a critical first step in the development of AI-assisted clinical decision support systems. This study focuses on instance segmentation of thyroid nodules using YOLOv5 algorithms on ultrasound images. We evaluated multiple YOLOv5 variants (Nano, Small, Medium, Large, and XLarge) across two dataset versions, with and without doppler images. The YOLOv5-Large algorithm achieved the highest performance with a dice score of 91\% and mAP of 0.87 on the dataset including doppler images. Notably, our results demonstrate that doppler images, typically excluded by physicians, can significantly improve segmentation performance. The YOLOv5-Small model achieved 79\% dice score when doppler images were excluded, while including them improved performance across all model variants. These findings suggest that instance segmentation with YOLOv5 provides an effective real-time approach for thyroid nodule detection, with potential clinical applications in automated diagnostic systems.

Paper Structure

This paper contains 31 sections, 3 equations, 1 figure, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Deep Learning for Medical Image Segmentation
  4. Instance Segmentation Approaches
  5. YOLO-based Approaches
  6. Research Gap
  7. Materials and Methods
  8. Dataset
  9. Data Preprocessing
  10. Instance Segmentation with YOLOv5
  11. Training Configuration
  12. Evaluation Metrics
  13. Dice Score
  14. Precision
  15. Recall
  16. ...and 16 more sections

Figures (1)

  • Figure 1: YOLOv5-Large inference showing multi-nodule segmentation with confidence scores. The model successfully detected and segmented four distinct nodules in a single ultrasound image.