A lot of exciting things have happened since my last blog. First off I must admit I have missed some blogs due to my parental leave. But, on the up site, the reward is a beautiful little daughter named Gaia. We are all super happy and enjoying these first few months together very much! Moreover, Gaia is now officially the youngest foodsmartphoner 🙂 🙂
Apart from this beautiful news a lot of exciting research has been done as well. And I will now give a short overview of what has been accomplished on hyphenating smartphones with biochemical assays for food contaminant detection from my site.
First off, I have been looking into the possibility to reliably quantify colorimetric assays with different smartphone models. For this purpose I have looked into the effects of using different colour spaces and channels and combinations of colour channels for the quantification of paper based and liquid assays with various smartphones. For this work we took pictures of a battery of different assays at various concentrations of the targeted contaminant including nanoparticle based paper strips, pH strips, lateral flow assays for: 1 cow milk detection in goat milk (milk adulteration); 2 gluten detection in gluten free products and; 3 domoic acid in shellfish, 4 saxitoxin in shellfish and, 5, okadaic acid in shellfish. From the taken pictures we then extracted the raw RGB values using a known colour app downloaded from google play. These raw channel values were then converted into other colour spaces (CieLAB and HSV). The individual channels of all these colour spaces were then used to generate calibration curves. Predictions with the fitted curves were used to determine the efficiency of each of the colour spaces. We did this because there seems to be no consensus in the scientific literature regarding which channel to use. Some suggest it is better to go for the H channel of HSV which conveys the hue of the assay. Others claim it is better to use the L (lightness) channel of CieLab colour space instead since this colour space is closer to the description of true colour perceived by the human eye. Others again claim it is better to stay in RGB space and use one of those 3 channels because converting these values mathematically can only add more noise to the measurements….
Thus we decided to look experimentally what was best. What we found was very interesting. It seems that for all the assays we tested there is truly no need to convert to other colour spaces. Instead R, G or B channel always give the best results. This being said, one should look for each specific assay which of the three channels is working best since that is highly assay dependent.
Another interesting find that I had in this assay was the benefit of proper background correction. In fact, I discovered that simply taking a ratio of the obtained background and test values within the same picture can get rid of a lot of noise introduced from illumination variation. What we found is that there is actually no need for the use of a light shielding box if one avoids working in direct sunlight. At least not for all the assays we tested. This is very good news. Since the ultimate smartphone hyphenated assay would not need to use any additional devices such as a light box.
Its not all sunshine (or rather shade for these colorimetric assays) and roses however. When the experiments were repeated with different phone models it became clear that the absolute values for colour intensity differed highly between different phones. This means it is currently not possible with the lateral flow assays to just take a picture with phone A and quantify the colour intensity using a calibration curve made in the lab with phone B… This means threre is still work to do to improve that situation. On the positive side, pH quantification (which is a colour change and not a colour intensity change based assay) was much more consistent over different phone models. Almost never with errors exceeding 5%, even if calibration was done on another phone model as used to do the measurement. This is great news, since it means we are almost there for our colorimetric based assays.
Only a few minor difficulties to overcome and accurate colorimetric quantification of lateral flow assays with a smartphone will be possible at your dinner table!
If you want to read more about this exciting research please check out our latest publication on this titled: The Efficiency of Color Space Channels to Quantify Color and Color Intensity Change in Liquids, pH Strips, and Lateral Flow Assays with Smartphones. The work is open access and I hope it can contribute to further progress smartphone based detection of food contaminants.
Apart from this work more will hopefully become available soon as I am currently back in way to sunny and not so snowy Switzerland on another secondment. But I will be sure to update you all on that the next time.
Hasta la pasta