The focus of this study is the potential use of FTIR imaging as a tool for objective automated histopathology. The Thesis also reports the use of multivariate statistical techniques to analyse the FTIR imaging data. These include Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA), Multivariate Curve Resolution (MCR) and Fuzzy C-Means Clustering (FCM). The development of a new PCA-FCM Clustering hybrid that can automatically detect the optimum clustering structure is also reported.
Chapter 1 provides a brief introduction to the use of vibrational spectroscopy to characterise biomolecules in tissues and cells for medical diagnosis.
Chapter 2 details the basic histology of a lymph node before proceeding to present imaging results gained from the analysis of both healthy and diseased lymph node tissue sections. The ability of each multivariate technique to discriminate different tissue types is discussed. In addition, the spectral features that are characteristic for each tissue type are reported. The development and application of a new PCA-FCM Clustering algorithm that can automatically determine the best clustering structure is also described in full. The results indicate that cellular abnormality provides changes to both the protein and nucleic acid vibrations. However, similar spectral profiles were identified for highly proliferating cells that were contained within reactive germinal centres of the lymph node.
Chapter 3 provides a short introduction to the histology of the cervlx before presenting imaging results that were gained from the analysis of both healthy and diseased cervical tissue sections. The ability of each multivariate technique to discriminate different tissue types is discussed. In addition, the spectral features that are characteristic for each tissue type are described in detail. Novel imaging experiments upon exfoliated cervical cells are also presented. It would appear that cellular abnormality in cervical tissues and cells affects both the protein and nucleic acid features of the spectra. Glycogen and glycoprotein contributions that are prevalent in healthy tissues are also absent.
Chapter 4 details sample preparation methods, the instrumentation and procedures used for data acquisition, and the subsequent data processing and multivariate techniques applied to analyse the collected spectral datasets.
[1]
Andrew G. Glen,et al.
APPL
,
2001
.
[2]
Lee G Luna,et al.
Manual of histologic staining methods of the Armed forces institute of pathology
,
1968
.
[3]
K. N. Dollman,et al.
- 1
,
1743
.
[4]
G. G. Stokes.
"J."
,
1890,
The New Yale Book of Quotations.
[5]
George Karypis,et al.
Soft clustering criterion functions for partitional document clustering: a summary of results
,
2004,
CIKM '04.
[6]
Vijay V. Raghavan,et al.
3M algorithm: finding an optimal fuzzy cluster scheme for proximity data
,
2002,
2002 IEEE World Congress on Computational Intelligence. 2002 IEEE International Conference on Fuzzy Systems. FUZZ-IEEE'02. Proceedings (Cat. No.02CH37291).
[7]
Greg Hamerly,et al.
Alternatives to the k-means algorithm that find better clusterings
,
2002,
CIKM '02.
[8]
Rohit Bhargava,et al.
Spectrochemical Analysis Using Infrared Multichannel Detectors
,
2007
.
[9]
Ericka Stricklin-Parker,et al.
Ann
,
2005
.
[10]
Jonathan M. Garibaldi,et al.
The Application of a Simulated Annealing Fuzzy Clustering Algorithm for Cancer Diagnosis
,
2004
.
[11]
Jonathan M. Garibaldi,et al.
Fuzzy Clustering in Biochemical Analysis of Cancer Cells
,
2005,
EUSFLAT Conf..