Editing Super-resolution imaging (section)

==Research==
There is promising research on using [[convolutional neural network|deep convolutional networks]] to perform super-resolution.<ref>{{Cite arXiv|last1=Johnson|first1=Justin|last2=Alahi|first2=Alexandre|last3=Fei-Fei|first3=Li|date=2016-03-26|title=Perceptual Losses for Real-Time Style Transfer and Super-Resolution|eprint=1603.08155|class=cs.CV}}</ref> In particular work has been demonstrated showing the transformation of a 20x [[microscope]] image of pollen grains into a 1500x [[scanning electron microscope]] image using it.<ref>{{Cite journal|last1=Grant-Jacob|first1=James A|last2=Mackay|first2=Benita S|last3=Baker|first3=James A G|last4=Xie|first4=Yunhui|last5=Heath|first5=Daniel J|last6=Loxham|first6=Matthew|last7=Eason|first7=Robert W|last8=Mills|first8=Ben|date=2019-06-18|title=A neural lens for super-resolution biological imaging|journal=Journal of Physics Communications|volume=3|issue=6|pages=065004|doi=10.1088/2399-6528/ab267d|issn=2399-6528|bibcode=2019JPhCo...3f5004G|doi-access=free}}</ref> While this technique can increase the information content of an image, there is no guarantee that the upscaled features exist in the original image and [[Image scaling#Deep convolutional neural networks|deep convolutional upscalers]] should not be used in analytical applications with ambiguous inputs.<ref>{{Cite conference |author=Blau |first1=Yochai |last2=Michaeli |first2=Tomer |year=2018 |title=The perception-distortion tradeoff |conference=IEEE Conference on Computer Vision and Pattern Recognition |pages=6228–6237 |doi=10.1109/CVPR.2018.00652|arxiv=1711.06077 }}</ref><ref>{{Cite web |last=Zeeberg |first=Amos |date=2023-08-23 |title=The AI Tools Making Images Look Better |url=https://www.quantamagazine.org/the-ai-tools-making-images-look-better-20230823/ |access-date=2023-08-28 |website=Quanta Magazine |language=en}}</ref> These methods can [[Hallucination (artificial intelligence)|hallucinate]] image features, which can make them unsafe for medical use.<ref name="cohen-miccai-2018">{{cite conference <!-- Citation bot no --> |conference=21st International Conference, Granada, Spain, September 16–20, 2018, Proceedings, Part I |last1=Cohen |first1=Joseph Paul |chapter=Distribution Matching Losses Can Hallucinate Features in Medical Image Translation |title=Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 |series=Lecture Notes in Computer Science |date=2018 |volume=11070 |pages=529–536 |doi=10.1007/978-3-030-00928-1_60 |arxiv=1805.08841 |isbn=978-3-030-00927-4 |s2cid=43919703 |chapter-url=https://www.springerprofessional.de/en/en/distribution-matching-losses-can-hallucinate-features-in-medical/16122390 |access-date=1 May 2022 | first2=Margaux |last2=Luck |first3= Sina |last3=Honari |editor1= Alejandro F. Frangi |editor2= Julia A. Schnabel |editor3= Christos Davatzikos |editor4= Carlos Alberola-López |editor5= Gabor Fichtinger }}</ref>