The story of MCG: lost opportunity.

This story is a great example of how illiterate bureaucracy, for quite poorly specified reason, was capable to stop a pioneering research and delayed by 30 years the opportunity for the progress in an important area of science.

1. The origin

This method was one of the first digital microscopy methods, which main  goal was the registration of the stable and reproducible profile of the cortical area of human brain. At the time of its development,  computers were very simple and not at all powerful, and even  such terms as  “medical informatics” and “digital pathology” did not exist. However, considering modern definition of digital pathology as an image-based information environment enabled by computer technology, our set of programs and algorithms was truly pioneering technology. Due to it’s stability and reproducibility, the procedure was given a name “Automatic Morpho-Corticography”, of MCG.

I should start the review of  MCG technology from a short tribute to my Alma Mater – the N. I. Pirogov’s Second Moscow Medical School (presently – Pirogov’s Russian National Research Medical University). The reason for that is not just a mere politeness or pride. The former Rector (or Dean, in American English) of the 2nd Moscow Medical School, prof. Yu. M. Lopuhin, who is better known for his contributions to transplantology and hemosorbtion, had an amazing ability to foresee the future of Medicine. He and prof. S.A. Gasparian were among the first people in Russian (or rather “Soviet”) medical science who understood the significance of computers and software engineering in medicine (“medical cybernetics”, as they used to say at that time).  Many years before it became well known and widely used, they were able to understand the importance of medical imaging, computerized microscopy and computer-assisted diagnosis. Thus, in 1973, they established in the Second Moscow Medical School the first in the world Department (or “Faculty”) of Medical Cybernetics.  Medical students, in addition to medical education, were able to get training in programming, computer science, image processing and other fields related to the application of computers in medicine.

Importantly, the teaching process was organized in such a way, that starting from the second year of education students were able to participate in solving practical scientific problems using the most powerful computers and computerized equipment available at that time. This opportunity was open even for students enrolled in “general” medical program. So, I wrote my first code for automatic scanning and segmentation of histological specimens in the Laboratory of Prof. V.E. Nemirovsky, in 1974. I was a third-year student and started working with first two computers I ever saw in my life: Wang 720C and PDP-12. Computers were just unpacked and installed, and radiated this unique head-spinning smell of brand-new “western” electronics, which very few could experience first hand at that time. I said “code”, since high-level language for Wang-720C was not available: the program had to be developed in hexadecimal “auto-code” (0407 – “GoTo”, 0408 – “Label”, 0300 – “+” etc.). Given this experience, I never take C++, Python and Matlab for granted.

2. Design of the technology and implementation conditions

For its time, it was truly pioneering technology. It was based on algorithms of Mathematical Morphology (J. Serra, CMM, France), implemented in the texture analysis sytsem TAS, by Ernst Leitz, Wetzlar (see fig. 1)

MCG_Implementation_1

Fig. 1: Brief technical description of automatic microscopy system  Leitz TAS (1980-1982).

One again, our method was one of the first digital microscopy method, developed for registration of the stable and reproducible profile of the cortical area of human brain. We defined “stability” as obtaining statistically indistinguishable result when the measurement was repeated in the same area, which was achieved by adaptive segmentation algorithm based on measuring of mean gradients distribution, and rather sophisticated methods of elimination of artifacts, vessels and glia. Reproducibility was defined as obtaining the same profile pattern and average values measured in the same area, but different set of serial sections, what was achieved by satisfying ergodicity requirements (see, for example “Ch. Lantuejoule, Ergodicity and integral range, J. of Microscopy, 161, 3, 1991. Practically it meant processing very significant volume of tissue in each area, at least within 2 mm of section space at  ~<0.5 mkm resolution). Starting from step 4 (below) it was fully automatic.

EPSON scanner image

Fig. 2: MCG processing pipeline.

The method was called “Morphocorticography” because in a substantial number of cases it revealed a pattern of local minima and maxima, specific for a given individual. This pattern reproduced itself in different areas of the same individual, but not in another individual.

All details of the method, including MM notations of image processing steps, were published between 1982 and 1992 in Russian (translated in English, Istomin VV, Neuropatol. and Behav. Physiol.,1986, UDC 616.831.31-076.5), English (V. Istomin, K. Amunts, Vision & Voice Magazine, 6, 2, 1992) and German (Istomin V., Amunts K., Build und Ton, 43, 1, 1990 and Amunts K, Istomin V., Vision & Voice Magazine, 6, 1, 1992). Unfortunately, these journals were not very well known, and probably do not even exist anymore.

3 thoughts on “The story of MCG: lost opportunity.”

  1. Thank you for paying tribute and preserving memory of yhe people who evidently have contributed so much for the scientific development!
    At the place and time where every substantial and good deed must have been done contrary to the prevailing standard, those scientists and truly gifted administratirs had the strength and dedication to carry out their vision! The following generations of the medical practitioners and researchers will carry on!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.