Papiron platform to aid diagnosis in dental panoramics

Abstract

Papiron is the result of a partnership between Papaiz, the largest network of diagnostic imaging for dentistry in Brazil, and MaChiron, an artificial intelligence startup in health, and has a database of 217,000 panoramic views, allowing the creation of deep learning models for evaluating the quality of imaging exams and detecting more than 30 types of radiological findings. The Papiron platform has as its main objective to assist in the diagnosis of radiological findings in dental panoramic radiographs. In the first phase, we developed panoramic quality assessment models to standardize and improve the exam acquisition process. Generally, when an x-ray exam is performed, the image is acquired with a device controlled by a radiology technician and then needs to be approved by a radiologist dentist, which interrupts your workflow. This increases the waiting time of the patient and reduces the productivity of the clinic per device. With the automatic evaluator, the flow would be optimized, as the image would only be viewed by the radiologist during the exam evaluation and would already include a marking of the findings detected by the platform. In addition, the tool has algorithms that detect important changes in the image, such as the presence of caries and missing teeth. The use of Artificial Intelligence algorithms in the detection of altered findings in radiographs brings benefits to clinical practice as it helps to reduce the variability in the detection capacity between dentists, reduces the rate of detection errors of subtle findings and accelerates the laudation process, as the findings are already pointed out and pre-filled in the report editor, requiring only the verification by the dentist before approving. The tool works as a web platform that receives images through API integration with radiological clinics, which allows the analysis of exams in batches automatically, or through manual upload of individual users. The main tasks performed are: image quality control classification, construction of the automated odontogram (map with present and absent teeth) and indication of the presence of caries. The tool is built so that it is possible to quickly integrate new algorithms for the detection of pathologies other than caries as they are trained. As detailed below, the objective of this proposal is to carry out the technological validation of the tool in a relevant environment within the partner company. After the conclusion of this PIPE/SEBRAE proposal, our objective is to launch the product in the market in a safer way with a tool tested and validated both from the clinical point of view and from the scale infrastructure point of view. In addition to Papaiz, we will seek to expand as customers other dental radiological clinics, offices and a network of dental clinics, integration with electronic medical record tools and manufacturers of x-ray equipment. It is worth mentioning that because it is a web tool, this product is scalable and internationalization is also part of the market expansion plans. (AU)