How can I remove labels that are incorrectly identified? For example, Missing Tooth that is looking at cheek mucosa. Tartar labeled as Fractured Tooth. Stained enamel identified as Calculus after cleaning.
Hi,
Thank you for your query regarding the AI X-ray pathology detection system. Your concerns about erroneous labels, such as ‘Missing Tooth’ being identified on cheek mucosa or mislabeled conditions like tartar and fractured teeth, are valid.
Currently, our system does not offer a direct method for users to remove or edit these AI-generated labels. This is primarily because the system is designed to evolve and improve its accuracy over time as it is exposed to more diverse images. It’s important to understand that, similar to discrepancies in interpretations among professionals, AI models can also reflect varying degrees of certainty in their predictions.
To assist you in evaluating the reliability of these labels, we provide a ‘confidence level’ for each prediction. This feature is intended to help you gauge how confident the AI model is about its label.
Shown below:
We recognize the importance of user feedback in refining our AI models. Therefore, we are actively exploring mechanisms that would allow users to flag predictions they believe to be incorrect. Such feedback would be invaluable in enhancing the accuracy of future model iterations.
In the meantime, we recommend using the AI’s labels as a supplementary tool alongside professional judgement, particularly in cases where the AI’s confidence level is lower or when there’s a known challenging area in interpretation.
Your feedback is crucial to us, and we appreciate your involvement in making our AI system more robust and reliable.