Debacles mar “Big Science” and fMRI research

Rinckside 2016; 27,7: 17-18.

ashington, 1989: “Now, Therefore, I, George Bush, President of the United States of America, do hereby proclaim the decade beginning January 1, 1990, as the Decade of the Brain.” Twenty-four years la­ter:

“Last year, I launched the BRAIN Initiative [a twelve year pro­gram] to help unlock the mysteries of the brain, to improve our treatment of con­di­tions like Alzheimer’s and autism and to deepen our understanding of how we think, learn and remember.“ US-President Barack Obama.

   Neuroscience is a wide field of brain and spine studies that evolved from the sciences of neuroanatomy and neurophysiol­ogy. Nowa­days the discipline encompasses sub­dis­ci­plines far outside the boundaries of the exact sciences and of me­di­ci­ne.

With the US-initiatives and those proposed and implemented elsewhere in the world, there came “Big Science” devoted to neuroscience research: a hun­dred million here, a hundred million there — or, in a European program, one billion over ten years. Alto­gether, many billions of euros or US-dollars were pro­mised to sub­si­dize projects heavily rely­ing on fMRI research, among them the US-American Hu­man Con­nec­tome Pro­ject and the European Human Brain Project. The ten-year European project started in 2013, but was already interrupted in 2015 for reasons described without over­en­thu­sias­tic detail by the project leaders and by those responsible in the European Commission — but told fully in a paper in Scientific American:

“Two years in, a $1- billion-plus effort to simulate the human brain is in disarray. Was it poor management, or is something fun­damentally wrong with Big Science?” [1].

Those in the know in Europe remained silent. In a report about the US Connectome project, one finds the following statement, hid­den in a box outside the running text:

“What is needed to get past the current impasse is a method that selec­tively separates global signal from global noise. Though no such method is yet avail­able, we offer several observations about global fMRI fluctuations.” [2]

The history of MRI is a story of successes but also a story of empty promises.

   The history of MRI is a story of successes but also a story of empty promises. Certain events have implications and consequences that only slowly un­fold over the years.

BOLD imaging and its applications have developed into more than a disappointment — it might become an utter fiasco. Many results are based upon obscure metadata. I have described some reasons and low points in my last column about fMRI: “Func­tional charlatans” [3]. I don't want to stray off here into the multifaceted picture of BOLD imaging and fMRI; on the other hand, I want to stress that there are numerous serious, genuine, and critical scientists in imaging neuroscience.

However, not only these scientists are upset, the educated and critical public is also getting annoyed, as Manfred Schneider describes their reactions to fMRI and Big Sci­ence in the science pages of the Swiss newspaper Neue Zürcher Zeitung:

“Numerous neuroscience institutes were founded, immense amounts of money were mobilized, the western societies got into a neuroscience frenzy. Thousands of subjects were placed in functional magnetic resonance equipment, where they had to endure movies, pornographic pictures, poems, while the researchers at their terminals ob­served the oxygen metabolism in their brain cells, converted the thrown out data into little pictures and tried to delude the world into believing that they are watching the brain thinking, feeling, acting.

By now such studies have lapsed into a kind of oddball science. Mean­while, no linguist can talk any more about synonyms without telling the confused zeit­geist that the words cake and pie are “synomynously” stored directly next to each other in the left temporal lobe [4].”

More and more researchers admit that acquisition and processing techniques of BOLD data lack the required meticulousness and thus the biased results and conclu­sions are scientifically irrelevant. Even the inventor of BOLD fMRI, Seiji Ogawa was among the harshest critics. Twenty-two years after his first description in 1990 [5], he published a 19-page review paper where he, in a roundabout way, discusses and disputes his tech­nique [6].

Several thousand papers on fMRI appear every year, Kim and Ogawa mention 3,000 [6], PubMed's numbers stretch between 42,000 and nearly 170,000 since 1990, de­pending on the search terms one uses. Meanwhile it has become clear that many of these papers, apparently a majority, rest on shaky foundations. Some scientists read Seong-Gi Kim's and Seiji Ogawa's review as a farewell to BOLD imaging, but this con­clusion seems too drastic and far-reaching. However, the publication alludes to the in­tricacy of the scientific background:

“The BOLD effect in fMRI is very complex, and this is still an area of intense research.”

The authors also observe in their concluding remarks:

“Dynamic properties and magni­tudes of BOLD functional responses are dependent on many phy­sio­logi­cal parameters as well as baseline conditions. In patients with neu­rovascular disorders, the BOLD re­sponse could be sluggish, or even de­creas­ed rela­tive to baseline. This should not be interpreted simply as a decrease in neural activity, because neurovascular coupling may be ham­pered … Resting-state fMRI studies are widely performed, but its physio­logical source needs to be systematically investigated.”

At the end, the critical reader's conclusion is: BOLD and fMRI should stay in the hands of genuine scientists. Clinical and psychological or commercial applications should be limited to trained and principled researchers. Today, the concepts of fMRI rely on a great many hypotheses, calculations, and simulations; however, practical proof to establish the validity of these models lags behind. Twenty-six years after Ogawa's original publica­tion, everything is still “panta rhei — everything is in flux”: definitely no hope for a trip to Stockholm, but rather “back to the drawing board”.

A number of scientists had hinted at the threatening problems, among them Christoph Sege­barth [7] and Nikos Logothetis [8]. Here is but one more example published by a Swedish group in spring 2016:

“We found that the most common software packages for fMRI analysis (SPM, FSL, AFNI) can result in false-positive rates of up to 70%. These results question the valid­ity of some 40,000 fMRI studies and may have a large impact on the interpretation of neu­roimaging results [9]."

   Definitely, “Biostatistics for Radiologists” is not sufficient to perform fMRI and to apply sta­tistical processing to the results. On the other hand, fMRI is not in the hands of radiolo­gists — fortunately, in this case.

Functional MRI seemed one of the most promising research techniques for and be­yond neuroimaging: the true study of brain organization. Now we fear the waste of hun­dred of millions euros of research grants and the shattered remains of thousands of scientific papers. Since nobody really feels responsible or in charge it will be difficult to minimize the repercussions of this debacle.

References

1. Theil S. Why the human brain project went wrong — and how to fix it. Scientific American 2015; 313. 1 October 2015.
2. Glasser MF, Smith SM, Marcus DS, et al. The Human Connectome Project’s neu­roimaging approach. Nature Neuroscience 2016; 19: 1175-1187.
3. Rinck PA. Functional charlatans. Rinckside 2015; 26,4: 9-11.
4. Schneider M. Geld für Science-Fiction —- die Hirnforschungshysterie. Neue Zürcher Zeitung; 24 May 2013; pg. 23.
5. Ogawa S, Lee TM, Kay AR, Tank DW. Brain magnetic reso­nance imaging with con­trast dependent on blood oxygenation. Proc Natl Acad Sci USA 1990; 87: 9868-9872.
6. Kim S-G, Ogawa S. Biophysical and physiological origins of blood oxygenation level-dependent fMRI signals. Journal of Cerebral Blood Flow & Metabolism 2012; 32: 1188-1206.
7. Segebarth C, Belle V, Delon C, Massarelli R, Decety J, Le Bas JF, Décorps M, Ben­abid AL. Functional MRI of the human brain: predominance of signals from extracere­bral veins. Neuroreport. 1994, 21;5: 813-816.
8. Logothetis NK. What we can do and what we cannot do with fMRI. Nature. 2008; 453 (7197): 869-878.
9. Eklund A, Nichols TE, Knutsson H. Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. PNAS 2016; 113, 28: 7900-7905.

Citation: Rinck PA. Debacles mar “Big Science” and fMRI research. Rinckside 2016; 27,7: 17-18.

A digest version of this column was published as:
Rinck PA. Debacles mar “Big Science” and fMRI research.
Aunt Minnie Europe. Maverinck. 17 November 2016.