| Insulin injection plays a very important role in diabetes treatment, but it is pretty hard to monitor injection sites in practical due to the difficulty to remember the previous injection sites correctly on a daily basis. If a patient injects insulin on the same tissue (or same spot of the skin) repeatedly, there can be many side effects like lipodystrophy which means abnormal storage of fat at the same place and will lead to erratic glucose control. In this thesis, we develop an innovative smartphone-based system that can effectively monitor the location of every injection site, and provide feedback for enforcing a new insulin injection attempt. Furthermore, as a smartphone-based system, it is convenient for nurses or patients to use. The system achieves monitoring under a combination of image processing, data analytics, and other advance technologies. We successfully implement a prototype of our system, which can run on any device with Android environment and a default camera, and we also evaluate its performance in terms of correct detection probability, response delay, and deviation of location estimation. |
