Frequently Asked Questions

Why do you use artificial intelligence (AI) in biomedical research?

Our research involves microscopes which produce a lot of images. It takes a long time to sort through and analyse them all, and AI and machine learning (including HRMAn) can help make this process much quicker. In addition, researchers can sometimes be accidentally biased towards finding a specific bit of evidence or data that supports their hypothesis or research question. This can potentially skew the data and make it less reliable. HRMAn helps reduce this bias, but we hope to reduce it even more by taking into account large numbers of classifications made by you, our volunteers.

What exactly am I seeing in the images in the Parasite Patrol workflows?

When you look at the Workflow images, you will see a green Toxoplasma parasite, and you may also see blobs of pink defence proteins gathering around it. What you are seeing is a special type of Toxoplasma which has been changed in the lab so that it produces a green fluorescent protein (GFP). Under a certain wavelength of light, it glows green and the microscope captures the image.

The defence proteins also look fluorescent because because we have added antibodies which stick to the protein that we are interested in. This antibody has a fluorescent tag which fluoresces under a certain wavelength of light making it look pink. It works similarly to the green fluorescent protein produced by the Toxoplasma in that way. This allows the microscope to see the two different colours and distinguish between the parasite and the defence proteins.

Here are some examples of images in the workflow and what the different parts of the image look like:

What if I'm not sure how to classify something?

If you're not sure, you can consult the tutorials or field guide, or you can select the 'No/I don't know'. You can use the zoom and colour flip functions in the workflow to look at the image more closely if you’re not sure which option to choose. Some images are difficult to classify, but this if part of why we need your help - getting data from a large number of people makes it more likely that we will reach an accurate consensus, i.e. as close to the ‘truth’ as possible.

What will you do with my classifications?

When we start analysing your classifications, the number of ‘Yes’ and ‘No/I don't know' classifications for each image in will be counted and used to make a dataset. This dataset will be uploaded to HRMAn which will use the dataset to learn the properties of different images including sizes and shapes across different parts of the image by analysing the pixels. When we give HRMAn new images that it hasn't seen before, it will try to find similarities between the new images and the images it was shown during training. We hope that it will This will help HRMAn recognise whether there is an attack on the parasite or not.

I need help with a workflow - what should I do?

Don't worry if you're having trouble - you can consult the tutorials and field guide at any time. You can also ask questions or give suggestions on our Talk (insert hyperlink) page. The 'Tricky images' collection may also help you distinguish between images that show a cell responding to a parasite, and those that don't.

Will my progress be saved if I leave the Classify page?

Yes - You can leave the Classify page at any point and your classifications will be saved. This means you can do a lot of classifications at once, or just a few at a time.