The Computational Photography Project for Pill Identification (C3PI) will allow the public to identify unknown pills from photos. Users can also search by colour classification, pill description and dimensions.
C3PI plans to standardise image formats to support mobile and desktop viewing, giving app developers the chance to create mobile drug-checking software and online platforms.
Computer scientists said they are looking for standardised metrics and methods that will be unaffected by differences in camera angle, lighting, and colour transfer functions in digital cameras. Pills will be photographed using visible spectrum macrophotography techniques.
The photos are available for public access through an interface on the NIH website. Users can search for pill colour, shape, imprint – such as text, size, NDC number, RxCUI (RxNorm Concept Unique Identifier), and Structured Product Label (SPL) identifier.
Call to drugmakers
The National Library of Medicine (NLM), part of NIH, said the project is “an investment in consumer safety,” and building a “reliable and high-quality image catalogue” of all prescription pills in the US will help poison control, emergency responses, and the reduction of medication errors.
Currently, the NLM has photographed more than 2,300 prescription oral solid dosage formulations – including tablets and gel capsules, representing over 6,000 national drug codes (NDCs).
It aims to add many more, and called on drugmakers wholesalers and pharmacies to provide physical samples of pills to help the project.
The IT work behind the growing portfolio is being run by the Office of High Performance Computing and Communications (OHPCC) at Lister Hill National Center for Biomedical Communications.