Difference between revisions of "Team:Cambridge-JIC/MicroMaps"
(6 intermediate revisions by 5 users not shown) | |||
Line 31: | Line 31: | ||
<center><h1 style="line-height:1.295em"> MicroMaps </h1></center> | <center><h1 style="line-height:1.295em"> MicroMaps </h1></center> | ||
<hr> | <hr> | ||
− | <center><p><i>The future of microscopy is (almost) here! Catch a sneakpeak of MicroMaps and play around with our early alpha by getting | + | <center><p><i>The future of microscopy is (almost) here! Catch a sneakpeak of MicroMaps and play around with our early alpha by getting hold of your own OpenScope.</i></p></center> |
<hr> | <hr> | ||
</div></div></section> | </div></div></section> | ||
Line 47: | Line 47: | ||
<li><p>See something you like? <b>capture a raw unprocessed image</b> for later, or <b>drop a pin</b> to return to later!</p></li> | <li><p>See something you like? <b>capture a raw unprocessed image</b> for later, or <b>drop a pin</b> to return to later!</p></li> | ||
<li><p>Need data? use an <b>extensive automated annotation toolkit</b> to measure and characterise your sample. Looking for a specific phenotype? Want to count your cells? Look no further - all of this with the comfort of knowing that you can manually intervene if the computer gets it wrong!</p></li> | <li><p>Need data? use an <b>extensive automated annotation toolkit</b> to measure and characterise your sample. Looking for a specific phenotype? Want to count your cells? Look no further - all of this with the comfort of knowing that you can manually intervene if the computer gets it wrong!</p></li> | ||
− | |||
</ul> | </ul> | ||
<hr> | <hr> | ||
Line 103: | Line 102: | ||
<p><b>The Method:</b> We tested our image processing software on some images of <i>Marchantia</i> gemma on a Petri dish with agar, This was intended to be a step towards our <a href="https://2015.igem.org/Team:Cambridge-JIC/Stretch_Goals" class="blue">Stretch Goal</a> - an automated screening desktop system. To write the software, the <a href="http://opencv.org/" class="blue">OpenCV</a> library was used. Two types of image processing algorithms were implemented:</p> | <p><b>The Method:</b> We tested our image processing software on some images of <i>Marchantia</i> gemma on a Petri dish with agar, This was intended to be a step towards our <a href="https://2015.igem.org/Team:Cambridge-JIC/Stretch_Goals" class="blue">Stretch Goal</a> - an automated screening desktop system. To write the software, the <a href="http://opencv.org/" class="blue">OpenCV</a> library was used. Two types of image processing algorithms were implemented:</p> | ||
<ul> | <ul> | ||
− | <li><p><b>Standard thresholding</b><br>This makes an image grey-scale and searches for the dark areas. We started off with a basic contrast increase to isolate the darker areas of the image, which we assume would correspond to samples. | + | <li><p><b>Standard thresholding</b><br>This makes an image grey-scale and searches for the dark areas. We started off with a basic contrast increase to isolate the darker areas of the image, which we assume would correspond to samples. We then followed the steps in a paper [2] which was supposed to yield much better sample isolation for samples which look faint, and are hard to distinguish from their background. This ended up detecting dents in the agar gel along with the samples. To resolve this issue we came up with the next idea...</p></li> |
− | <img src="https://static.igem.org/mediawiki/2015/c/c1/CamJIC-bdct.png" style="height:250px;margin:10px"> | + | <center><img src="https://static.igem.org/mediawiki/2015/c/c1/CamJIC-bdct.png" style="height:250px;margin:10px"></center> |
<li><p><b>Colour detection</b><br>An eye dropper was added to select the upper and lower colour darknesses to search for (the user would click to select these colours). These colours correspond to areas of the sample with better and worse illumination respectively.Also, a slider that allows you to change the 'darkness' of the sample colour was added. This generally varies depending on room lighting conditions. With this implementation, the program performed much better, detecting the <i>Marchantia</i> gemma before the agar dents.</p></li> | <li><p><b>Colour detection</b><br>An eye dropper was added to select the upper and lower colour darknesses to search for (the user would click to select these colours). These colours correspond to areas of the sample with better and worse illumination respectively.Also, a slider that allows you to change the 'darkness' of the sample colour was added. This generally varies depending on room lighting conditions. With this implementation, the program performed much better, detecting the <i>Marchantia</i> gemma before the agar dents.</p></li> | ||
Line 110: | Line 109: | ||
<center><img src="//2015.igem.org/wiki/images/e/ef/CamJIC-Software-ImageRec.jpg" style="width:400px;margin:10px"><p><i><b>Figure 4</b>: Sample recognition working on Petri dish with Marchantia gemma. The program highlights the samples it finds in red. Note that the agar dent is not included in the final output. This was achieved using the color detection algorithm.</i></p></center> | <center><img src="//2015.igem.org/wiki/images/e/ef/CamJIC-Software-ImageRec.jpg" style="width:400px;margin:10px"><p><i><b>Figure 4</b>: Sample recognition working on Petri dish with Marchantia gemma. The program highlights the samples it finds in red. Note that the agar dent is not included in the final output. This was achieved using the color detection algorithm.</i></p></center> | ||
<p><b>Microscopic image processing:</b> The colour detection used above can theoretically be easily adapted to work with fluorescent samples – this would prove useful for sample counting and detection of, for example, samples that successfully express a specific fluorescent protein. A similar strategy can be applied to stained samples with interesting coloured features: for example to recognize stained nuclei (eg. with toluidine blue) and in this way distinguish eukaryotic cells.</p> | <p><b>Microscopic image processing:</b> The colour detection used above can theoretically be easily adapted to work with fluorescent samples – this would prove useful for sample counting and detection of, for example, samples that successfully express a specific fluorescent protein. A similar strategy can be applied to stained samples with interesting coloured features: for example to recognize stained nuclei (eg. with toluidine blue) and in this way distinguish eukaryotic cells.</p> | ||
− | <p>However, we have not implemented sample recognition into MicroMaps Alpha, mostly due to lack of time and difficulties for coping with multicolour images. Still, the script for image recognition is in the <a href="" class="blue">software package</a> for you to try out (and improve). | + | <p>However, we have not implemented sample recognition into MicroMaps Alpha, mostly due to lack of time and difficulties for coping with multicolour images. Still, the script for image recognition is in the <a href="https://github.com/sourtin/igem15-sw/blob/master/img_processing/identificationTesting/marchantiaIdentification_gui.py" class="blue">software package</a> for you to try out (and improve, also in the source code download on the <a href="//2015.igem.org/Team:Cambridge-JIC/Downloads" class="blue">Downloads</a> page under img_processing/identificationTesting/marchantiaIdentification_gui.py). |
<hr> | <hr> | ||
<center><p><i>Image recognition was developed by Ocean, with useful feedback and advice from the rest of the Software team.</i></p></center> | <center><p><i>Image recognition was developed by Ocean, with useful feedback and advice from the rest of the Software team.</i></p></center> | ||
<hr> | <hr> | ||
</p> | </p> | ||
+ | <p style="font-size:80%">References: <br> [2] Chen, L., Chien, C. and Nguyen, X. (2013). An effective image segmentation method for noisy low-contrast unbalanced background in Mura defects using balanced discrete-cosine-transfer (BDCT). <i>Precision Engineering</i>, 37(2), pp.336-344.</p> | ||
Latest revision as of 00:50, 19 September 2015