Dept. of Computer Science and Engineering, Sri Jayachamarajendra College of Engineering, JSS Science and Technology University, Mysuru, India.
Dept. of Computer Science and Engineering, Sri Jayachamarajendra College of Engineering, JSS Science and Technology University, Mysuru, India.
Comput Biol Med. 2024 Dec;183:109318. doi: 10.1016/j.compbiomed.2024.109318. Epub 2024 Oct 28.
Cephalometric landmark annotation is a key challenge in radiographic analysis, requiring automation due to its time-consuming process and inherent subjectivity. This study investigates the application of advanced transfer learning techniques to enhance the accuracy of anatomical landmarks in cephalometric images, which is a vital aspect of orthodontic diagnosis and treatment planning.
We assess the suitability of transfer learning methods by employing state-of-the-art pose estimation models. The first framework is Detectron2, with two baselines featuring different ResNet backbone architectures: rcnn_R_50_FPN_3x and rcnn_R_101_FPN_3x. The second framework is YOLOv8, with three variants reflecting different network sizes: YOLOv8s-pose, YOLOv8m-pose, and YOLOv8l-pose. These pose estimation models are adopted for the landmark annotation task. The models are trained and evaluated on the DiverseCEPH19 dataset, comprising 1692 radiographic images with 19 landmarks, and their performance is analyzed across various images categories within the dataset. Additionally, the study is extended to a benchmark dataset of 400 images to investigate how dataset size impacts the performance of these frameworks.
Despite variations in objectives and evaluation metrics between pose estimation and landmark localization tasks, the results are promising. Detectron2's variant outperforms others with an accuracy of 85.89%, compared to 72.92% achieved by YOLOv8's variant on the DiverseCEPH19 dataset. This superior performance is also observed in the smaller benchmark dataset, where Detectron2 consistently maintains higher accuracy than YOLOv8.
The noted enhancements in annotation precision suggest the suitability of Detectron2 for deployment in applications that require high precision while taking into account factors such as model size, inference time, and resource utilization, the evidence favors YOLOv8 baselines.
头影测量标志点标注是放射分析中的一个关键挑战，由于其耗时且具有主观性，因此需要自动化。本研究探讨了先进的迁移学习技术在提高头影测量图像中解剖标志点准确性方面的应用，这对头齿矫正诊断和治疗计划至关重要。
我们通过使用最先进的姿态估计模型来评估迁移学习方法的适用性。第一个框架是 Detectron2，它有两个基于不同 ResNet 骨干架构的基线：rcnn_R_50_FPN_3x 和 rcnn_R_101_FPN_3x。第二个框架是 YOLOv8，它有三个反映不同网络大小的变体：YOLOv8s-pose、YOLOv8m-pose 和 YOLOv8l-pose。这些姿态估计模型被用于标志点标注任务。模型在 DiverseCEPH19 数据集上进行训练和评估，该数据集包含 1692 张带有 19 个标志点的放射图像，其性能在数据集中的各种图像类别中进行分析。此外，研究还扩展到 400 张图像的基准数据集，以研究数据集大小如何影响这些框架的性能。
尽管姿态估计和标志点定位任务的目标和评估指标有所不同，但结果还是很有希望的。Detectron2 的变体在 DiverseCEPH19 数据集上的准确率为 85.89%，优于 YOLOv8 的变体的 72.92%。在较小的基准数据集上也观察到了这种更好的性能，Detectron2 始终保持比 YOLOv8 更高的精度。
标志点标注精度的提高表明 Detectron2 适合在需要高精度的应用中部署，同时考虑模型大小、推断时间和资源利用等因素，证据倾向于 YOLOv8 基线。
