Computer Vision News - January 2018

Typically with FDA filings, they need to know about every changing component. A smartphone makes that more difficult. In the end, MobileODT became the first smartphone- connected imaging device that received FDA approval. The project still remains in its early phase, and MobileODT currently does not have a computer vision product. However, they have several projects on computer vision happening in parallel including a recent Kaggle challenge sponsored by Intel . It must be said that a certain fraction of women has an extra fold in their cervix, making them untreatable. Even though this only occurs in 2% of the patients, treatment could potentially kill these women. They want to flag these patients so that the nurses at the point of care refer them to an expert rather than move forward with treatment right away. The Kaggle challenge, which ended a few months ago, involved identifying those patients. They want to take the information from the challenge and gradually turn it to something that can run on a mobile platform. The various winners and their solutions took many approaches trying several different algorithms. Each of them had an incremental bump, and while no algorithm solved everything magically, together they performed quite well. The challenge validated that the methods work, although they have a partner that is still trying to figure out how to optimize these methods and put them on a mobile platform. MobileODT has an ongoing project with Global Good , an organization funded by Bill Gates and Intellectual Ventures to give successful engineering ideas the chance to contribute to mankind. In this case they try to develop an algorithm that will allow nurses to detect cancer at the point of care. In terms of computer vision, the project requires several things . First, they need to make sure that the images are good enough quality so that the algorithms can work well on them. Some of their users have never used a smartphone. They have to teach nurses how to use text messages and then the camera, in order to improve their ability to take high- quality images with the device. The Global Good project developed a focus score . After a user takes an image, the focus score instructs them whether they need to take another image, and what to improve. 12 Computer Vision News MobileODT “MobileODT became the first smartphone- connected imaging device that received FDA approval” Application