Development and Validation of a Digital Image Processing-Based Pill Detection Tool for an Oral Medication...

Area of Science:

• Biomedical Engineering
• Digital Health
• Pharmaceutical Sciences

Background:

• Medication adherence is crucial for managing chronic diseases.
• Existing objective adherence tracking tools are often costly, inaccessible, or require extra devices.
• There is a need for affordable, accessible tools to monitor pill intake objectively.

Purpose of the Study:

• To develop a low-cost, device-free tool for monitoring oral medication adherence using mobile health technology.
• To assess the efficacy of a digital image processing technique for detecting pill removal from blister packs.

Main Methods:

• A pill intake detection tool was developed using digital image processing.
• The Circular Hough Transform was employed for feature extraction, focusing on round pills.
• The tool involves two steps: blister registration and pill detection/classification.

Main Results:

• 100% detection rate for round pills in full blister packs (gray or transparent cover).
• High accuracy in counting untaken pills in partially opened blisters (95.2% for remaining, 95.1% for taken in gray blisters).
• Detection rates for transparent covers were 88.2% for remaining and 80.8% for taken pills.

Conclusions:

• The proposed tool demonstrates significant potential for objective, low-cost monitoring of round pill adherence.
• Further improvements are needed for accurate classification of taken versus remaining pills, especially for non-round pill shapes.