Project summaries 2021
Using a crappifying method to generate better microscopy images faster
A technique for artificially degrading images will help apply deep learning to enhance low-resolution microscopy images
Using deep learning to enhance low-resolution microscopy images has become a lot easier thanks to a new method for generating training data that involves artificially degrading images. This method was developed as part of an AMED-funded study led by Yoshiyuki Kubota of the National Institute for Physiological Sciences.
Microscopes suffer from an ‘eternal triangle of compromise’ due to trade-offs between image resolution, illumination intensity and imaging speed. This problem is particularly troublesome for point-scanning systems, the most widely used systems for high-resolution imaging.
Deep learning is a promising way to boost the resolution of low-resolution images, but it requires collecting large volumes of training data.
Now, to generate the large amount of training data needed for deep learning, a team led by Kubota has developed a computational filter they affectionately dubbed a ‘crappifier’. “The crappifier allows us to generate training data much more easily from pre-existing data,” explains Uri Manor, a member of Kubota’s team. “We trained a deep learning-based AI system using crappified data to enable undersampling when imaging biological samples.”
Using their technique, the team was able to obtain images up to 16 times faster than before. “Seemingly impossible imaging projects can now be completed within our lifetimes,” says Manor. It will be especially helpful for large imaging projects and for imaging living samples that are sensitive to high light levels, such as mitochondria.
Importantly, the method does not require expensive equipment, making it highly accessible to labs.
“This is but one of many examples of how deep-learning-based AI is making possible what we never would have imagined to be possible before,” says Manor. “The future of AI in biological research is very exciting and we expect to see many other breakthroughs in microscopy and beyond.”
Researcher: Yoshiyuki Kubota
Host Institution: National Institute for Physiological Sciences
AMED Funding: Strategic Research Program for Brain Sciences (SRPBS) / Brain Mapping by Integrated Neurotechnologies for Disease Studies (Brain/MINDS)
Mouse memory enhanced by skipping food every second day
Fasting every other day linked to improved long-term memory in mice
Making mice go hungry every second day boosts their long-term memory, an AMED-funded study has found. This finding may have implications for human health and could lead to therapies that slow mental decline.
Originally identified as an age-suppressing gene, Klotho has been linked to both aging and long-term memory. However, it was unknown whether these two effects are connected.
Now, a team that included Makoto Kuro-O of Jichi Medical University in Japan has observed the effects of intermittent fasting on mice by feeding them every other day. After three months of this intermittent fasting, mice exhibited improved long-term memory retention compared with mice that had been fed daily.
On a neurological level, this was observed as upregulation of Klotho in the hippocampus — a brain region primarily associated with memory.
“We discovered that intermittent fasting improved long-term memory in mice compared to a matched control,” explains Kuro-O. “The improvement was associated with increased neurogenesis and expression of the gene Klotho in the hippocampus.”
This long-term memory improvement was not simply due to reduced calorie intake, since mice fed daily with the same reduced-calorie diet as the intermittent-fasting mice did not exhibit such an increase in long-term memory.
The team believes that their results highlight the central role that Klotho plays in generating new neurons in adults.
These findings may one day help human health and memory. “It may be possible to slow cognitive decline and extend life span by using drugs to maintain or enhance endogenous Klotho expression,” says Kuro-O.
Researcher: Makoto Kuro-O
Host Institution: Jichi Medical University
AMED Funding: Advanced Research & Development Programs for Medical Innovation
Supplying oxygen to mammals through the gut
Oxygen can be administered through the butts of mice and pigs, raising the possibility of helping patients with respiratory failure in the future
Mice and pigs can respire via their intestines instead of their lungs — a finding that may one day help patients with severe respiratory problems.
Certain aquatic creatures such as sea cucumbers and freshwater catfish can breathe through their rectums in low-oxygen environments. But this ability has never been observed in mammals.
Now, in an AMED-funded study, a team led by Takanori Takebe of the Tokyo Medical and Dental University has come up with a way to administer pure oxygen through the rectums of mice and pigs.
Mice and pigs given intestinal oxygen survived longer in low-oxygen environments than those without intestinal oxygen, with 75% of the mice surviving 50 minutes of low-oxygen conditions, which normally would be fatal. Furthermore, animals given intestinal respiration had more oxygen reach their hearts and displayed fewer signs of oxygen deprivation, such as skin discoloration.
“This is the first study to demonstrate that the mammalian gut can be repurposed for breathing,” notes Takebe.
The process also worked when the team substituted the intestinal gas ventilation system with a liquid-based alternative made up of perfluorochemicals, an alteration that may make the technique more feasible for use in humans.
“Sadly, many people having been suffering from hypoxic condition during the pandemic,” says Takebe. “Although the technique’s safety needs to be thoroughly evaluated, I hope we will be able provide an alternative option to conventional artificial respiratory support for patients as soon as possible.”
The team is currently conducting preclinical trials with a view to testing their technique in a clinical setting.
Researcher: Takanori Takebe
Host Institution: Tokyo Medical and Dental University
AMED Funding: Research Program on Emerging and Re-emerging Infectious Diseases
Return to AMED-funded projects