Accurate classification of periodontal disease through panoramic X-ray images carries significant clinical importance for the diagnosis and treatment of periodontal disease. Existing methods determine periodontal disease by estimating bone loss in X-ray images, relying on clinical annotations of pathological information on these images. However, these annotations lack consistency with clinical probe measurement results. To address this, our study proposes a hybrid classification algorithm directly supervised by clinical probe measurement results. The new algorithm can comprehensively estimate the probability of a patient having periodontal disease from both tooth and patient levels, and incorporates clinical priors for decision-making, thereby improving the model's sensitivity to periodontal disease diagnosis. We conducted extensive experimental testing on datasets collected from clinics, and the results show that our method achieves excellent performance in periodontal disease classification, demonstrating tremendous potential for application.

Figure 1. Proposed integrated clinical prior periodontal disease hybrid classification network framework.

Figure 2 Comparison of ROC curves between the proposed model and existing representative methods.

The related research findings were published in Medical Image Analysis under the title 'Clinical Knowledge-guided Hybrid Classification Network for Automatic Periodontal Disease Diagnosis in X-Ray Image'. Lanzhuju Mei, a PhD student from Professor Dinggang Shen's research group at ShanghaiTech University, Ke Deng, an assistant professor at the University of Hong Kong, and Zhiming Cui, an assistant professor at ShanghaiTech University, are co-first authors. Professor Dinggang Shen, a tenured full professor at ShanghaiTech University, and Professor Maurizio S. Tonetti from Ninth People’s Hospital, School of Medicine, Shanghai Jiao Tong University are co-corresponding authors. ShanghaiTech University is the first completing institution, the University of Hong Kong, Shanghai Ninth People's Hospital affiliated to Shanghai Jiao Tong University School of Medicine and Shanghai Clinical Research and Trial Center are collaborating institutions.

Paper link： https://doi.org/10.1016/j.media.2024.103376