Algorithms for smartphone and tablet image analysis for healthcare applications

Abstract

Smartphones and tablets are finding their way into healthcare delivery to the extent that mobile health (mHealth) has become an identifiable field within eHealth. In prior work, a mobile app to document chronic wounds and wound care, specifically pressure ulcers (bedsores) was developed for Android smartphones and tablets. One feature of the mobile app allowed users to take images of the wound using the smartphone or tablet's integrated camera. In a user trial with nurses at a personal care home, this feature emerged as a key benefit of the mobile app. This paper developed image analysis algorithms that facilitate noncontact measurements of irregularly shaped images (e.g., wounds), where the image is taken with a sole smartphone or tablet camera. The image analysis relies on the sensors integrated in the smartphone or tablet with no auxiliary or add-on instrumentation on the device. Three approaches to image analysis were developed and evaluated: 1) computing depth using autofocus data; 2) a custom sensor fusion of inertial sensors and feature tracking in a video stream; and 3) a custom pinch/zoom approach. The pinch/zoom approach demonstrated the strongest potential and thus developed into a fully functional prototype complete with a measurement mechanism. While image analysis is a very well developed field, this paper contributes to image analysis applications and implementation in mHealth, specifically for wound care.