 image, picture, detector, and it's using three different algorithms. Pretty common in the field. Fast, IRS, and she to Massey. And here you have a bunch of image example images that you can select. Yes, in this case, you can see it's not performing very well, because it's just picking up on the text, specifically the IRS algorithm. And the fast does a bit better, but it does pick up the features in the image, but then just pretty much selecting everything in this particular case. And she to Massey is actually selecting, it's actually working pretty well because it's selecting those a that might be areas of interest, but then it's missing this hemorrhage. So not very good. So none of them is actually working well. So what again, Harris is just picking the text labels on the image fast, it's just all the shootings and labeling everything as the area of interest and she to Massey to Massey is labeling some useful bits, but potentially missing the most important thing here, which would be the lower hemorrhage, which is in the name of that file. So I assume that correct. You have some normal MRIs with assumed higher contrast. The sheet to Massey is doing pretty well, it's labeling some of the areas of interest quite well. A Harris is also not doing too bad. And fast is just labeling pretty much the entire image, which is not very useful, is it? So you could play play around with it, see which one you like better. They obviously work differently depending on the quality of the image and so on so forth. So we might be doing more work on this later on.