Towards Better Cephalometric Landmark Detection with Diffusion Data Generation

TL;DR

This paper introduces a novel diffusion-based data generation method for cephalometric X-ray images, significantly improving landmark detection accuracy by augmenting training data with realistic, annotated images created without manual effort.

Contribution

The study presents a new diffusion-based image generator and a prompt dataset for controlled, diverse cephalometric X-ray synthesis, enhancing deep learning detection models.

Findings

Generated data improves detection accuracy by 6.5% SDR

Achieved 82.2% Success Detection Rate with augmented data

Demonstrated effective control over image attributes during generation

Abstract

Cephalometric landmark detection is essential for orthodontic diagnostics and treatment planning. Nevertheless, the scarcity of samples in data collection and the extensive effort required for manual annotation have significantly impeded the availability of diverse datasets. This limitation has restricted the effectiveness of deep learning-based detection methods, particularly those based on large-scale vision models. To address these challenges, we have developed an innovative data generation method capable of producing diverse cephalometric X-ray images along with corresponding annotations without human intervention. To achieve this, our approach initiates by constructing new cephalometric landmark annotations using anatomical priors. Then, we employ a diffusion-based generator to create realistic X-ray images that correspond closely with these annotations. To achieve precise control in producing samples with different attributes, we introduce a novel prompt cephalometric X-ray image dataset. This dataset includes real cephalometric X-ray images and detailed medical text prompts describing the images. By leveraging these detailed prompts, our method improves the generation process to control different styles and attributes. Facilitated by the large, diverse generated data, we introduce large-scale vision detection models into the cephalometric landmark detection task to improve accuracy. Experimental results demonstrate that training with the generated data substantially enhances the performance. Compared to methods without using the generated data, our approach improves the Success Detection Rate (SDR) by 6.5%, attaining a notable 82.2%. All code and data are available at: https://um-lab.github.io/cepha-generation