Doppler-Enhanced Deep Learning: Improving Thyroid Nodule Segmentation with YOLOv5 Instance Segmentation
Mahmoud El Hussieni
2025-12-02
Summary
This paper investigates using computer vision, specifically a type of AI called YOLOv5, to automatically identify and outline thyroid nodules in ultrasound images, which are images doctors use to look at the thyroid gland.
What's the problem?
Thyroid cancer is becoming more common, and doctors need help to accurately find and analyze thyroid nodules – lumps in the thyroid. Manually identifying these nodules in ultrasound images can be time-consuming and sometimes inaccurate. A key step in helping doctors is to precisely outline each nodule in the image, a process called segmentation, but doing this automatically is challenging.
What's the solution?
Researchers tested different versions of the YOLOv5 algorithm (Nano, Small, Medium, Large, and XLarge) to see which one was best at instance segmentation, meaning identifying *each* nodule separately. They used two sets of ultrasound images: one with standard images and another that also included 'Doppler' images, which show blood flow. They found that the YOLOv5-Large version worked the best, achieving high accuracy, and surprisingly, including the Doppler images significantly improved the results for all versions of the algorithm.
Why it matters?
This research shows that AI can be a useful tool for quickly and accurately detecting thyroid nodules in ultrasound images. The finding that Doppler images, often ignored by doctors, actually *help* the AI perform better is particularly important, suggesting a way to improve diagnostic systems and potentially lead to earlier and more accurate diagnoses of thyroid cancer.
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.