An improved multipath residual CNN-based classification approach for periapical disease prediction and diagnosis in dental radiography

Dental radiography offers significant indication for medical/clinical diagnosis, quality assessment and treatment. Huge efforts have been taken while developing the digital dental X-ray image analysis system for the enhancement of clinical quality. In this manuscript, the datasets, methodology and analysis of performance are carried out for the evaluation of qualities regarding the dental treatment with the utilization of periapical images of dental X-ray that is taken before and after the operations. With the purpose of supporting dentists to make some clinical decisions, a tool pipeline for automated clinical quality evaluation is being proposed. In this approach, a disease diagnosis from the dental image analysis is made by means of deep learning techniques. Initially, the dental input dataset is preprocessed using bias-corrected filter technique. For segmentation process, semantic contextual network segmentation (SCNS) is employed. The features were extracted using multi-scale local ternary pattern (MS-LTP). Statistical linear discriminant analysis (SLDA) approach is employed for the selection of features. At last, the extracted and selected features outcome is post-processed to improve the rate of classifier performance. The classification process is carried out by means of improved multipath residual CNN (IMRCNN) classifier. Thus, the proposed technique provides better accuracy than others in the diagnosis of dental disease to predict Periapical Disease Detection in Dental Radiographs image. Thus, the disease is predicted so as to diagnose the severity earlier and helpful for dentist in making decisions on treatment process. Thus, finally the performance estimation has been made and the results were compared with existing techniques to prove the effectiveness of proposed strategy.

Automated summary

This paper introduces a dental image disease diagnosis method based on deep learning techniques. By pre-processing, segmentation, feature extraction, and classification of dental X-ray images, an automated clinical quality evaluation is achieved. This method has higher accuracy in predicting periapical disease detection and can assist dentists in making treatment decisions.