Professional Documents
Culture Documents
FCI Review
FCI Review
niques employed in computational imag- algorithms that are necessary for advancing
ing, a concept previously dispersed across through the rest of the book.
multiple sources. That shared groundwork Chapter 11 offers an in-depth discussion
Downloaded 02/14/24 to 38.140.150.106 . Redistribution subject to SIAM license or copyright; see https://epubs.siam.org/terms-privacy
renders the book both accessible and use- of model parameter estimation. The sub-
ful to researchers from diverse fields, such sequent four chapters (12 to 15) deliver a
as applied mathematics, physics, chemistry, comprehensive introduction to vital con-
optics, and signal processing. cepts related to the EM algorithm, Markov
Image reconstruction is a major part chains, hidden Markov chains (HMM), and
of computational imaging. Dr. Bouman stochastic simulation, which includes a va-
presents the image reconstruction challenge riety of sampling techniques. The final two
as a classical inverse problem, via the lens chapters (16 and 17) expound on Bayesian
of the Bayesian framework. Here is the segmentation and Poisson data models. To
question we ask: How do we reconstruct complement the core content, four appen-
the original image X from some physical dices provide additional insights.
sensing system that yields measurements Each chapter includes a set of exercise
Y , which are contingent on this unknown problems (always good in a textbook) at
image X? These measurements Y are typ- varying difficulty levels. At the end of the
ically characterized by substantial redun- book there is a list of nearly 100 pertinent
dancy and noise. The data representing references to the literature that span sev-
the crucial components of an authentic im- eral decades. It is noteworthy that the very
age are akin to a sparse set within another first citation concerns the development of
large set. The modeling approach that the computed tomography (CT) by Cormack
author takes considers the physical system and Hounsfield, who won the Nobel Prize
and the image X as random quantities. A in Physiology or Medicine in 1979—that
forward model of the system is given by brings home the importance of computa-
p(y|x), the conditional distribution of the tional imaging.
data Y given the original (unknown) im- To reiterate, this is an excellent text for a
age X. This distribution describes how the graduate-level course where students have a
measurements are related to the original decent background in probability and statis-
image and captures both the deterministic tics. It is also a helpful text for researchers
characteristics of the imaging system and or practitioners in medical and other types
probabilistic elements (e.g., noise and per- of imaging. While computing-related fields
turbations). Within this framework, prob- often evolve rapidly, I can see this book
lems can be categorized based on several staying relevant for years to come.
distinctive aspects, including single versus
ANITA T. LAYTON
dual physical dimensions, causal versus non-
University of Waterloo
causal analysis, continuous versus discrete
time, and Gaussian versus non-Gaussian
frameworks. Climate, Chaos and COVID: How Math-
The book is structured to unveil the in- ematical Models Describe the Universe.
tricacies that emerge within the aforemen- By Chris Budd. World Scientific, London, 2023.
tioned defining dimensions. It comprises a $78.00. xii+301 pp., hardcover. ISBN
total of 17 short chapters and four appen- 978-1-80061-304-1.
dices. The opening chapter offers a con-
cise introduction to computational imaging. Mathematical modeling played a high pro-
Following this, Chapters 2 to 4 provide an file and, at times, controversial role in the
overview of fundamental probabilistic tools, world’s response to the COVID-19 pan-
as well as 1D and 2D causal and noncausal demic. The modeling of Neil Ferguson and
Gaussian (Markov random field) models. his team at Imperial College is credited
The subsequent trio of chapters (5 through with derailing the UK government’s laissez-
7) delves into essential tools and algorithms faire approach to the pandemic in early
for maximum a posteriori (MAP) estima- March 2020. Here in New Zealand, where
tion. Chapters 8 to 10 encompass the re- I live, mathematical models underpinned
sults derived from constrained optimization our elimination of the virus and, together