The 12th Leeds Statistical Workshop on Shape and Image Analysis for Industry and Medicine

July 1st - 3rd, 1992. University of Leeds, U.K.

This annual series of workshops takes place over three days each summer. In recent years there has been an accent on the area of shape analysis, with previous speakers W.S. Kendall, Fred Bookstein and for 1992 Colin Goodall of Penn State University. Forty-three participants attended from as far afield as Australia, USA and Spain. The diverse range of backgrounds of the participants sums up the wide range of applications for statistical shape analysis: Statistics, Image Analysis, Anatomy, Anthropology, Industry, Medical Physics, Bioengineering, Computer Science, Electrical Engineering and Biomedical Science.

The format of the workshop was changed this year: the principal presenter Colin Goodall gave three keynote lectures with associated computer practical sessions, and in addition there were sessions of shorter talks from various contributors.

The lectures:

The analysis of shape: basic principles and applications. Estimation and inference for shape data. Visualization, and unusual shapes.

Prof. Goodall gave us an overview of statistical shape analysis of landmark data, from his perspective. This crucially involves employing the full geometrical information of the Kendall's shape space. The concepts of preshape sphere, tangent plane and Riemannian distance can be rather challenging at first, but there really are advantages to "doing things properly". Inference and estimation are carried out using Procrustes methods and many details can be found in his 1991 paper J.Royal.Statistical.Soc. B, 285-339. Further topics of discussion included shape distributions, other methods of shape analysis and visualization of shapes. Computer practicals involved playing around with different shape methods using the elegant S-Plus programming language on SUN workstations, and visualizing high dimensional shape spaces using Silicon Graphics machines.

Other contributed talks:

Statistical shape methodology
Prof Kanti Mardia, Statistics, Leeds
Active shape models for Image interpretation
Dr Tim Cootes, Wolfson Image Analysis Unit, Manchester
Gesture Recognition
Mr Mark Howes, Psychology, Leeds
The complex Bingham distribution in shape analysis
Prof. John Kent, Statistics, Leeds
Large sample shape analysis in the plane using complex eigenvectors
Mr Mike Prentice,
Shape analysis for automatic fish recognition
Dr Graham Horgan, SASS, Edinburgh
Studies of craniofacial shape in primates
Dr Paul O'Higgins, Anatomy, Western Australia
The shape of faces
Dr Ian Craw, Mathematical Sciences, Aberdeen
Morphology and shape
Dr Charles Taylor, Statistics, Leeds
Automated crack detection on forged metal components
Dr Noel Bryson, Wolfson Image Analysis Unit, Manchester
Terrain models from object motion
Professor David Hogg, Computer Studies, Leeds


Flexible template matching using genetic algorithms for ultrasound echocardiagrams
Dr Noel Bryson, Wolfson Image Analysis Unit, Manchester
Statistical size and shape analysis
Dr Ian Dryden, Statistics, Leeds
Applications of shape theory to collinearity testing
Dr Huiling Le, Mathematics, Nottingham
Thin plate splines
Mr John Little, Statistics, Leeds
Hominoid craniofacial growth : cellular activity and morphological change
Mr Jim McMahon, Anthropology, New York
Studies of craniofacial shape differences : Statistics, Geometry and Biology
Dr Paul O'Higgins, Anatomy, Western Australia
A distribution for paired size and shape
Dr Alistair Walder, Statistics, Leeds
Bayesian reconstructions from single photon emission camera tomographic data
Mr Iain Weir, Mathematics, Bristol
Chromosome analysis
Miss Lisa Wiffen, Statistics, Leeds

In summary, the conference was, I believe, very successful and interesting. Prof. Goodall's detailed lectures and practicals were well complimented by the large range of additional applied and theoretical contributions. The area of statistical shape analysis is an important and exciting one. With so many applied researchers interested in aspects of shape it is important that statistical methods are correctly employed, without disregard for the inherent geometrical and mathematical structure present in the objects and spaces under study.

The venue of Fairbairn House provided excellent food and facilities - although it was a shame that the Leeds weather did not quite match up to expectations! Lots of rain.

Ian Dryden.