A smart phone-based sensors are being developed by the researchers at IIT-H to detect adulteration in milk. As a first step, they have developed a detector system to measure the acidity in milk through an indicator paper that changes colour according to the acidity in the milk. They have also developed algorithms that can be incorporated into a mobile phone to accurately detect the colour change.
Undertaken by a team led by Shiv Govind Singh, professor in Department of Electrical Engineering, IIT-H, comprising associate professors Soumya Jana and Siva Rama Krishna Vanjari, the research has been published in the November issue of the journal Food Analytical Methods .
Speaking of the importance of the research, Prof. Shiv Govind Singh said: “While techniques such as chromatography and spectroscopy can be used to detect adulteration, they generally require expensive set-up and are not amenable to miniaturisation into low-cost easy-to-use devices. Given this, they do not appeal to the vast majority of milk consumers in the developing world. We need to develop simple devices that the consumer can use to detect milk contamination. It should be possible to make milk adulteration detection fail-safe by monitoring all of these parameters at the same time, without the need for expensive equipment.”
As a first step, the research team has developed a sensor chip-based method for measuring pH, an indicator of the acidity. They have used a process called ‘electrospinning’ to produce paper-like material made of nano-sized fibres of nylon, loaded with a combination of three dyes. The paper is “halochromic”, that changes colour in response to changes in the acidity.
The team has developed a prototype smart phone-based algorithm in which the colours of the sensor strips after dipping in milk are captured using the camera of the phone, and the data is transformed into pH (acidity) ranges.
They have used three machine-learning algorithms and have compared their detection efficiencies in classifying the colour of the indicator strips.
On testing with milk spiked with various combinations of contaminants, they found near-perfect classification with accuracy of 99.71%.
