It is a duty incumbent on upright and credible men of all ranks who have performed anything notable or praiseworthy to record in their own words the events of their lives. But they should not undertake this honorable task until they are past the age of forty.
-Benvenuto Cellini
When I first started writing Andy’s Brain Blog over a decade ago, I had a vision: That every man, woman, and child would have free access to learning about neuroimaging analysis, no matter how much money they made, no matter where they were from, no matter whether they passed or failed the Marshmallow self-control test as a child. All that was needed was an Internet connection, a willingness to learn, and an equal willingness to laugh at my jokes and furrow one’s brow with a serious look whenever I delivered any one of my cornball platitudes.
The blog and its eventual successor, Andy’s Brain Book, has helped many researchers throughout the years, and all of the knowledge on those websites has finally been distilled into a physical book. The book, Modern fMRI, is now available on Amazon for pre-order, and will be shipping on July 10th. I have been writing it for the past couple of years, and it represents the summation of my career until this point. I have plans for what the next phase of my career will be, but for now, this will be my contribution to the field.
What is the book about? It is a snapshot of where the field of neuroimaging stands in the mid-2020s, specifically fMRI. The book begins with the history of fMRI, as do all other books of its kind, and then devotes a couple of chapters to describing commonly used pulse sequences and experimental designs, along with their tradeoffs. After a brief chapter summarizing the basics of computer programming, the next one traces the development of the major software packages that were developed in the 1990s and are still used today: AFNI, SPM, and FSL, along with recent developments such as NiLearn.
The middle chapters form a triptych which covers statistics, region of interest analysis, and common fallacies, such as biased analyses. The latter chapter was one of the most pleasing to write, since I could describe some of the best ways to falsify and misrepresent your data through seemingly appropriate statistics; in my straitlaced, impeccably neat life, it is the closest I get to indulging in the bliss of evil.
The remainder of the book covers the history and latest developments in techniques such as functional connectivity and machine learning, and then makes some predictions about where the field is heading in the next decade or so, with a nod towards the increasing use of open-access databases and open science practices. For those who insist on pushing through to the end, you are rewarded with two appendices: one which reviews the papers that have called into question the results of the field, such as the so-called Dead Salmon paper and the Voodoo Correlations paper; and another appendix, admittedly thin, which briefly describes how Artificial Intelligence might be used to assist with neuroimaging analysis. Since this part of the field is rapidly changing, I thought it best to not make any definitive predictions, and simply to describe what is out there, and how it might be used. Good fortune to you.