3. Was it such a big challenge to create BigBrain data set?
Interview statement: “When, in 2003, together with her Canadian colleague Prof. Alan Evans, the Jülich neuroscientist Prof. Katrin Amunts decided to make thousands of histological sections of a human brain, to stain and digitize them, it was completely unclear whether it would ever be possible to reconstruct this brain on the computer. At that time, there were no technical possibilities to cope with the huge amount of data.”
My Question 1: What was so special about “deciding to make thousands of histological sections of a human brain, to stain and digitize them” in 2003?
My Answer: Nothing whatsoever. Making thousands of histological sections of human brain and staining them is a routine procedure accepted in all institutions dealing with brain anatomy and cytoarchitecture since the beginning of 18th century. Economo, Coskinas, Sarkisov, Brodmann, and many other neuroanatomists used serial sectioning technique as routine procedure necessary for every day work of any scientist dealing with the brain structure. Digitizing sections came later, but became available since first television camera was installed on the microscope, what happened in late 1970s. First whole slide scanner was developed in 1994 . Very quickly many companies, including Leica, Philips, 3DHistech, Hamamatsu, Photonics, Motic, Zeiss, Huron, Olympus, Roche, Perkin-Elmer, Keyence and Bionovation started producing these devices for different areas of pathology. So, in 2003 whole slide digitizing was very much a common place method, used by pathologists all over the world. Why authors decided to use flatbed scanner instead of the whole-slide scanner is a big question, but it was not revolutionary at all. Most likely the reason was the lack of funds, but it is also strange, given the level of funding available to authors’ institutions.
My Question 2: So, what problems was necessary to solve?
My Answer: At least – two, but it was not explained in the interview. First was collection and mounting of all sections. That was, indeed, a challenge. Usually, a technician selects every 20’s section or so, and there is always a possibility to discard some, if section becomes damaged. Collecting every single serial section of a brain – is a real achievement, but with all due respect – that was an achievement of a very good team of technicians, very well trained and with many years of experience. The second problem was registration of all these sections to create uninterrupted and smooth volume in 3D space. That was achieved by a team of software developers, who used an original (but not new) software package “MINC Tools”. This package was in development since 1992 (source: https://www.bic.mni.mcgill.ca/ServicesSoftware/MINC) for MRI imaging. Between 2003 and 2013 it was modified for histological sections during implementation of the described project.