Background

Although most countries are on track to meet the Millenium Development Goal target for providing access to drinking water, recent studies show that even sources that are considered improved are frequently contaminated with E. coli and other fecal indicator bacteria. Current methods for testing the microbial quality of water are expensive and require specialized equipment and trained personnel.

Problem Definition

The Portable Microbiology Lab, or PML, was developed to dramatically reduce the cost and equipment needed for the field detection of bacteria in water. The kit uses off-the-shelf Petrifilm growth media from the food safety industry and a simplified version of the rigorously tested Colilert reagent system.

The petrifilm media can be used to detect the following types of bacteria:

• Non-coliform bacteria (mostly harmless), which show up as red colonies without gas bubbles
• Thermotolerant coliform bacteria (often associated with fecal contamination), which are red colonies surrounded by gas bubbles (due to the fermentation of lactose)
• E. coli (strong indicator of fecal contamination), which appears as a blue colony (due to reaction with a particular enzyme)

A plate with all three types of colonies is shown in the photo below:

Counting the number and types of colonies can be time-consuming for large numbers of highly contaminated samples and since some interpretation is required, different users might introduce bias regarding which colonies to count. However, the rules are fairly straightforward and should be possible to implement in code using image processing algorithms. In fact, 3M sells a commercial plate reader for laboratories. 

The App

The goal of this project is to develop an iOS or Android app that can photograph the plates, count colonies, and save or transmit the photos and related data (such as time and place of sample) to a central database. This would provide the user-- often a community health worker --instant feedback regarding the safety of a water source and allow central tracking and mapping of water quality data for the entire municipality or region. An alternative formulation of this problem might be to develop server-side software (perhaps a modification to the open-source Ushahidi platform) that could receive photos via SMS or email and immediately relay back the bacteria counts.