---

Saturday, October17^th
 
9:30 - 11:30     Improving the Quality of Diagnosis and Prognosis (II)

---

The Problem of Barretts Oesophagus, Prospects for Endoscopic Diagnosis

Hugh Barr, C Fulljames, N Stone

Medical Raman Group, Gloucester

Dysplasia present in columnar-lined (Barrett's) oesophagus is a premalignant condition, with a sequence of increasing cellular abnormalities culminating in high grade dysplasia and ultimately oesophageal adenocarcinoma. The incidence of this carcinoma has shown a dramatic rise in both Europe and the USA. Patients with high grade dysplasia in Barrett's oesophagus present a very difficult clinical problem. Some advocate a policy of radical oesophageal resection, justifying this approach since carcinoma may be found in the resected specimen in 45% of patients. Others, concerned that half the patients are receiving a prophylactic operation with substantal morbidity and mortality, advocate intense endoscopic surveillance of dysplasia. If dysplasia is identified then endoscopic eradication is possible (1). In a pathological study following resection , the total area of Barretts oesophagus was found to be 32 cm² with areas of high grade dysplasia 1.3 cm² and carcinoma 1.1 cm². In this series patients are more likely to be overstaged and diagnosed as having adenocarcinoma than a cancer missed prior to surgery (2).
Optical biopsy endoscopic techniques for diagnosis have concentrated on autofluorescence or photosensitiser fluorescence. High-grade dysplasia has been successfully identified in Barretts oesophagus using these techniques. Thirty-six patients were studied and the area of Barretts oesophagus interrogated using a nitrogen-pumped dye laser emitting 410nm light in 5 nanosecond pulses. This was used to excite tissue autofluorescence which was collected by a fibre bundle and analysed by a spectrograph. Both emitting and collecting fibres were included in a flexible fibreoptic probe (1.7 mm) (3). There are several endoscopic devices now available. However diagnosis is moving to identify precise molecular markers to identify which patients will develop progressive disease with e-cadherin and telemorase activity (4). In this field vibrational spectroscopy may have a role.
We have demonstrated some differences in the vibrational spectra from various upper gastrointestinal tissues from normal, metaplastic, precancerous, and cancerous oesophageal and gastric tissue. These abnormal Raman lines may be correlated with the altered biochemistry in tumours, and it has been demonstrated that certain carotenoids and lipids have much greater intensities in cancer. There are major problems of Raman spectroscopy that have to be addressed, in particular the speed and time required for acquisition. This has made this technique impractical for an endoscopic device. However with a holographic Rayleigh rejection filters, a single monochromator is required to separate the Raman spectra into its separate energies. Hence a much greater throughput is achieved and much shorter acquisition times or reduced laser powers are possible. Vibrational spectroscopy is a difficult but realistic prospect for endoscopic diagnosis.

1. Barr H, Shepherd NA, Dix A, Roberts DJH, Tan WC, Krasner N. Eradication of high-grade dysplasia in columnar-lined (Barretts) oesophagus by photodynamic therapy with endogenously generated protoporphyrin IX, Lancet 1996; 348: 584-585.
2. Cameron AJ, Carpenter HA. Barretts esophagus, high grade dysplasia and early carcinoma: a pathological study. Am J Gastroenterol 1997; 92: 586-591.
3. Panjepour M, Overholt BF, Vo-Dinh T, Haggitt RC, Edwards, DH, Farris C, Buckley PF. Endoscopic fluorescence detection of high-grade dysplasia in Barretts esophagus Gastroenterology 1996; 111: 93-101.
4. Bailey T, Biddlestone L, Shepherd N, Barr H, Warner P; Jankowski J Altered cadherin and catenin complexs in the Barretts esophagus-dysplasia-adenocarcinoma sequence. American Journal of Pathology 1998: 152, 1-10.

 

---

 

Multiwavelength, Spectral and Coherence-Based Optical Imaging for Biomedical Applications

D. Farkas, B. Ballou, C. Du, G. Fisher, R. Levenson, Y. Pan and E. Wachman

Center for Light Microscope Imaging and Biotechnology, Carnegie Mellon, University, and Department of Bioengineering, University of Pittsburgh, Pittsburgh PA

The most versatile and time-efficient way of achieving non-invasive imaging of biological objects is by optical means. Microscopy has long been the dominant technique, and has enjoyed a veritable renaissance. Recent advances allow study of living cells and tissues, both dynamically and in three dimensions. Developments in several fields made this possible, yielding significant improvements in the spatial, spectral and temporal resolution of optical bioimaging. Recent progress in this field will be reviewed, emphasizing technological advances [1] and our salient contributions. We developed robotic, machine vision-type microscopy allowing dynamic live monitoring of events at the subcellular level [2], and further enhanced it spectrally and temporally by improving the performance of acousto-optic tunable filters [3]. Microscopic imaging with overall resolution an order of magnitude better than current standards is presented, together with some selected applications, emphasizing spectral imaging [4, 5], and extensions to the tissue level [6,7], including dual-wavelength, 3-D optical coherence tomography [8] of living skin.
The potential of optical functional imaging as a means of monitoring normal and pathological metabolic activity in vivo is also discussed, in the context of possible contributions to important medical challenges such as transplantation and cancer research.

1. Farkas, D.L., et al. (editors) (1998) Optical Investigations of Cells In Vitro and In Vivo, Progress in Biomedical Optics (Proc. SPIE), vol. 3260.
2. D.L. Taylor et al. (1997) Ann. New York Acad. Sci., 820, 208-228.
3. Wachman, E., Shonat, R. et al. (1997) Biophysical Journal, 73, 1215-1231 (2 articles)
4. Farkas, D. L., et al. (1997) Springer Lecture Notes in Computer Science, 1311, 663-671.
5. Farkas, D. L., et al. (1998) Computerized Medical Imaging and Graphics, 22, 89-102.
6. Ballou, B. et al. (1997) Biotechnology Progress, 13, 649-658.
7. Ballou B. et al. (1998) Cancer Detection and Prevention, 22, 251-257.7.
8. Pan, Y. and Farkas, D.L. (1998) Journal of Biomedical Optics, 3, in press.
 
Supported by the National Science Foundation (US) and the Ben Franklin Technology Center of Western Pennsylvania

 

---

 

Mid-Infrared Microspectrometry: an Emerging Tool for Tissue Diagnostics?

Peter Lasch¹, Janina Kneipp¹, Michael Beekes¹, Heike Audring², Wolfgang Hänsch³, Heinz Fabian³, and Dieter Naumann¹

¹ Robert Koch-Institut, D-13353 Berlin, Germany
² Dermatologische Klinik der Charité, Medizinische Fakultät der Humboldt Universität zu Berlin, D-10098 Berlin, Germany
³ Max-Delbrück-Centrum für Molekulare Medizin, D-13125, Berlin, Germany

Fourier transform infrared spectroscopy (FT-IR) is an accepted methodology in biophysics which provides structural information of biological molecules such as proteins, nucleic acids, carbohydrates, and lipids. In the past decade, this technique has also been applied to biomedical subjects, e.g. to the diagnosis of tissue pathologies. Research has been conducted on samples from a variety of different organs, including colon, cervix, heart, brain, prostate, skin, lymphatic organs to mention a few. Results obtained consistently proved that the spectral information is sufficient to distinguish between various tissue structures and also to study disease processes at a molecular level. Thus, most of the authors concluded that infrared spectroscopy is a potentially valuable tool for objective screening and disease diagnosis in real clinical settings. Particularly, the use of IR-microspectrometry which provides spatially resolved structural information and considers therefore tissue heterogeneities on the microscopic level enhances the discriminative power of the method significantly. For FT-IR data evaluation, single wavelength analysis, i.e. functional group analysis, is an already established technique.
Recently, multivariate data evaluation approaches to pattern recognition like Factor Analysis, Cluster Analysis or Artificial Neural Networks analysis (ANN) have been applied to detect typical spectral signatures and to increase reliability of diagnosis. We will present some of our very recent results on FT-IR microspectroscopic investigations of colorectal cancer thin sections, thin sections from melanoma and of hamster brain. The table below shows exemplary classification results on the ANN-analysis of 27 colorectal cancer samples with a histopathological grading II-III. For each of the samples, more than 150 spectra of 7 distinct histological structures were recorded by an infrared microsope. The analysis of the IR-spectra was performed with two randomly selected spectral data sets. One of them was utilized for network learning (training), while the other was necessary for internal validation. The classification results shown here were obtained on the validation data set.
 

	1	2	3	4	5	6	7	S	accuracy	SP	PPV	NPV	histology
1	191	0	2	0	0	0	2	195	97.9%	99.5%	97.5%	99.6%	crypts
2	0	114	1	4	4	0	0	123	92.7%	98.5%	88.4%	99.1%	L.muscularis mucosae
3	3	4	159	0	0	2	0	168	94.6%	99.2%	95.2%	99.0%	L. propria mucosae
4	1	4	1	120	0	0	0	126	95.2%	99.1%	93.0%	99.4%	submucosa
5	0	1	0	0	136	0	0	137	99.3%	99.6%	97.1%	99.9%	T. muscularis
6	0	4	0	1	0	121	3	129	93.8%	99.1%	93.0%	99.7%	connective tissue
7	1	2	4	4	0	7	280	298	94.0%	99.4%	98.2%	97.9%	adenocarcinoma
S	196	129	167	129	140	130	285	1176	95.3%

The overall accuracy of the classification of 95% seems to be very promising for a future reference data bank of FT-IR spectra which might be useful to very rapidly identify colon cancer in clinical samples. A more detailed presentation of the results and a discussion of the problematic details such as variance of spectral features due to histopathological processes, "inter-patient variance", standardization of data collection and definition of standards to assess spectra quality will be given at the symposium.

 

---

 

Spectroscopic Bioimaging Utilizing Infrared Focal-Plane Arrays

E. Neil Lewis, Pina Colarusso, Linda H. Kidder, Ira W. Levin, and Abigail S. Haka

Laboratory of Chemical Physics, National Institutes of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda,MD 20892-0510

The coupling of digital imaging and optical spectroscopy has traditionally proceeded through two distinct pathways: utilizing imaging detectors in concert with discreet optical filters, or coupling high resolution spectrometers with point mapping approaches. Fixed bandpass optical filters provide high image quality with little or no spectral information. Whereas in mapping approaches an emphasis on spectral performance takes precedence over image quality. More recently technologies that attempt to completely integrate these modalities are emerging. A range of continuously tunable optical filters, step-scan interferometers, high performance infrared focal-plane array detectors (FPAs) and the desktop computer revolution have all contributed to the feasibility and ease of completely integrating spectroscopy and imaging. Several recent developments in infrared FPA technology including uncooled near and mid-infrared versions point the way to the possibility of higher pixel densities, ease of accessibility and lower cost. Additionally, further developments in digital signal processing (DSP) hardware and interferometer architecture suggest that simpler and more compact implementations of these types of imaging systems will become available in the near future.
We have developed a spectroscopic imaging system that incorporates a step-scan FTIR, a microscope or macroscopic image formation optics, and infrared sensitive focal-plane arrays. The method provides a seamless combination of spectroscopy for molecular analysis with the power of visualization. It generates chemically specific images while simultaneously obtaining high-resolution infrared spectra for each detector pixel. The spatial resolution of the images approaches the diffraction limit, while the spectral resolution is determined by the interferometer, and can be 4 cm^-1 or higher. In contrast to standard histological protocols, FTIR spectroscopic imaging simultaneously analyzes cell layers and identifies subtle structural and biochemical changes within the sample.
As an example of the versatility of this technique, data collected from several biological systems will be presented. The neurodegenerative genetic disorder Niemann-Pick type C has been examined by spectroscopically imaging intact brain sections of normal mice and a mutant strain that expresses this human disease. FTIR images that show biochemical variation will be compared to images generated using standard histopathological staining techniques. Significant changes in the relative amounts of lipids in the various cellular layers were observed. These results are consistent with demyelination within the cerebellum of the NPC mouse. Standard techniques that characterize biochemical content are highly selective but destructive. A single imaging data set however, contains an inherent multiplicity of contrast producing mechanisms (vibrational normal modes) that arise from differences in the biochemical composition. Data will also be presented where the methodology has been used to elucidate and assess the biochemical variability in metastatic and normal prostate tissue specimens. The information, which can be extracted from these image data sets, clearly illustrates the power of FT-IR spectroscopic imaging as an analytical tool in research and medicine. We suggest that FTIR spectroscopic imaging should provide a highly reliable, complementary tool for standard histological tier testing.

 

---

 

Challenges and Recent Achievements in FT-IR Microspectroscopy

David L. Wetzel and Steven M. LeVine

Microbeam Molecular Spectroscopy Lab., Kansas State Univ., Shellenberger Hall, Manhattan, KS 66506; Dept. Mol. Integ. Physiology, Univ. of Kansas Med. Ctr., Kansas City, KS 66160

FT-IR microspectroscopy may assist the medical professional in the following ways: (1) diagnosis of disease, (2) characterization of the chemical basis of disease mechanisms, and (3) monitoring the effectiveness of therapeutics. The ability to relate chemical analysis of tissue histology in situ makes FT-IR microspectroscopy far superior to microanalysis for the purposes stated above. The modern era of FT-IR microspectroscopy began by the 1989 patenting of a high performance IR microscope with all-reflecting optics and dual remote image plane masks. Stringent design considerations were required to produce instruments with a combination of a high S/N ratio and high spatial resolution in the microscopic field, which have been described by Reffner in 1998 in Cell. Mol. Biol. Vol. 44 (1). A high optical throughput interferometer is coupled with an optically efficient IR microscope to maintain radiation flux. A sensitive cryo HgCdTe small area detector that matches the microbeam contributes to the optical efficiency and favorable S/N operation. Biological applications were rare when we made our first presentations in 1987-1991. Our neuroscience work began in 1991 with functional group mapping of the cerebral cortex and underlying white matter. A continuation of this work included: (1) characterizing the chemistry of white matter in twitcher mice (globoid cell leukodystrophy), (2) studying multiple sclerosis brain lesions, (3) utilizing extravasated blood experiments to show chemical differences in lesioned white matter vs. penumbra and normal white matter, (4) studying inflammation in brains of HIV-positive individuals, (5) characterizing dramatic chemical differences of cerebellar layers, (6) mapping single Purkinje cells, (7) utilizing drinking water containing 40% D₂O to generate CD, ND, OD, as a novel method to study protein and lipid metabolism (8) at the Brookhaven National Lab, in collaboration with the Canadian NRC group, the chemistry of Alzheimer plaques was characterized, and (9) our recent experiments on rat retinas utilized linear mapping and a differential interference optical reflective objective (for improved microscopic visualization) to correlate chemistry with retina histology. Other instrumental advances we have employed include synchrotron radiation for an IR source and a focal plane array camera as a detector. The former permits ultraspatial resolution and the latter permits rapid imaging with simultaneous spectra obtained for 4096 pixels. In 1998, an entire thematic issue (270 pages) devoted to FT-IR microspectroscopy appeared as Vol. 44 of Cellular and Molecular Biology (Paris). The distribution of lipids and proteins were mapped in living cells. Several groups reported efforts to classify exfoliated cervical cells. Melanoma and oral squamous cell carcinoma were examined. Detection of cancerous cells or identification of foreign materials in biopsied breast tissue were reported. Metabolites of drugs of abuse were localized in hair. Two groups examined bone specimens. Cooperation from members of the medical/pathology community together with chemists and other physical scientists will enable FT-IR microspectroscopy to advance the understanding of human diseases on a molecular level.