Ali Ertürk – our pioneering scientist who analyzes cleared tissue with artificial intelligence and says dreaming is good for science

Some researchers call his plans "far-fetched"; others use less polite descriptions, says Ali Ertürk. He can be highly self-critical, but he also calls himself a "dreamer" working across many fields, including neuroscience, nanotechnology, advanced engineering and deep learning. Ertürk, a researcher at Ludwig Maximilian University in Munich, says, "It has been fun so far, and it is getting more exciting". Ali Ertürk continues to direct a new institute of tissue engineering and regenerative medicine at the Helmholtz Center near Munich. His laboratory of 20 people from 15 nations consists of experts in computer science, neuroscience, chemistry, engineering and microscopy. Ertürk draws inspiration from historian Yuval Noah Harari's book Homo Deus: A Brief History of Tomorrow, which describes how humanity is designing itself for the 22nd century and how it advances science by bringing together expertise across fields. "This is definitely happening in biology plus deep learning", says Ertürk.

Ali Ertürk's work

In 2006, Ertürk began working on a topic studied a century earlier by the pioneering German anatomist Werner Spalteholz: clearing tissues (making them clear / transparent). He used fluorescence microscopy and GFP labels to harness the power of clearing to see detailed brain structure. Ertürk and his team cleared an entire mouse and imaged it with a laser-scanning microscope, thereby examining tumors and metastases at the single-cell level.

Below, you can see a study of Ali Ertürk and his team's newly developed technology, which enables the detection of even the smallest metastases throughout a mouse's entire body, DeepMACT, based on deep learning (multilayer artificial neural networks), in a video summary: (Turkish subtitles are available)

Until now, tissues and organs have been imaged in 2-dimensional form in many imaging studies; in contrast, 3-dimensional imaging, although methodologically quite difficult, provides much more detailed and valuable information.

An entire mouse yields a four-terabyte dataset consisting of 100,000 images, which is too large to analyze manually, accurately and within a reasonable time. Because the light is brighter at the tissue's periphery, a standard thresholding approach risks losing data from the tissue's interior. Tissue-clearing laboratories need to use the artificial intelligence method called deep learning to analyze the data, says Ertürk, "otherwise the field will not move very far".

A new section of Ali Ertürk's laboratory: VesSAP: Vessel Segmentation & Analysis Pipeline

VesSAP (Vessel Segmentation & Analysis Pipeline). In this work, his team cleared all the blood vessels in the brains of different types of mice, including the small capillaries. The data were registered to the Allen Mouse Brain Atlas.

VesSAP uses a machine learning network consisting of five convolutional layers. To avoid the need for a large training dataset, they used 'transfer learning', in which synthetic datasets were generated and refined with real data. The network achieved "human-level accuracy", says Ertürk. "This is the way to go with such large datasets", he says. The team wanted to avoid using a multilayer network such as the U-Net convolutional network with its millions of parameters. With their own algorithms and fewer parameters, his team achieved accuracy levels similar to U-Net. "But our algorithms are about 25 to 50 times faster", he says. This gain means that the new analysis pipeline can run on standard computers. Users can download the entire analysis pipeline or individual modules. "Whatever you want to take, you can take", he says. There are 3D images of the vascular system, algorithms, documentation and examples. The team worked on a combination of chemical stains to make both large and small blood vessels visible. He and his team are adapting this analysis pipeline to analyze the vascular system in a cleared human kidney. "I think this is very beautiful because the human kidney is an incredible structure," says Ertürk. "We do not know how it is organized at this level of detail." They had to modify the chemistry for human organs and adjust the microscope setup to scan the entire kidney. The 3D maps are part of a broader plan, just as an architect needs a plan to build a skyscraper.

Who is Ali Ertürk? His education

Ertürk grew up in Turkey. He completed his undergraduate education in the Department of Molecular Biology and Genetics at Bilkent in Ankara. During his undergraduate studies, he completed two summer internships in neurobiology laboratories in the United States, at Yale Medicine and Harvard Medicine. He had never previously spent time in a research laboratory, left Turkey or been on an airplane. "It was a very big thing for me", he says. For his doctoral research in neuroscience, he went to the Max Planck Institute of Neurobiology, where he worked on imaging technologies and tissue clearing. He took photographs, bringing together the real world and the microscopic one. "What I love is taking photographs, especially at night." He also spends time with his family, including his newborn baby. He enjoys reading audiobooks; he gets 70-80 books a year.

"It has been fun so far, and it is getting more and more exciting."

Ertürk's mentor and the CEO of Helmholtz Munich, Matthias Tschöp, said they had set out to transform biomedicine by combining 'next-generation engineering' and AI (artificial intelligence) technologies. "With his expertise in systems biomedicine, high-resolution imaging and artificial intelligence, Ali Ertürk embodies the 'hybrid' scientist we need today to tackle the growing health threats we face as a society", he says. Beyond his scientific achievements, "Ali has an infectiously creative mind." He is a thinker who synthesizes ideas, with impressive talents as an accomplished artist and photographer. Tschöp said, "We are delighted that he has decided to join our team as one of our institute directors".

Ertürk completed his postdoctoral fellowship at Genentech. Its culture of integrating research and practice is the culture he hopes to incorporate into his new institute. He loved the San Francisco Bay Area, and its widespread entrepreneurial spirit gave him the confidence to explore engineering and deep learning. When people tell him he cannot succeed at something, he replies, "I do not know, but I will find a solution; that is my attitude".