My client is a doctor wanting to create an AI model that looks at xrays of broken bones to determine if the patient requires a clinical appointment.
There would be 2 parts: 1. Creating and testing an ai model 2. Creating an application (potentially a chrome extension) that captures a photo and emails a report with the below output.
Output - classification of body part - assesment of if there is an injury or not - classification of the injury - treatment plan (either 1. clinincal appointment or 2. no clinical appointment)
Examples of the types of images to be used are as follows. The model will be trained with similar xray images which will be labelled as 'clinical appoitment' or 'no clinical appointment'.
Questions: How many xrays photos are required to train the AI Model? Can the model be built on microsft azure? Can photos be extracted from a web app and sent to an AI model to be processed? Is there a digitally secure way to capture and anonomyse the images? Can the output from the ai model be email to a pre-determined email address? What format do you require the data and images for training the ai model?
