X-Ray Microanalysis of Cryosections Using Image Analysis

Abstract

It is possible for electron probe x-ray microanalysis to be used to generate quantitative digital images of elemental distribution in cryosections of biological specimens (Gorlen et al. 1984, Ingram et al. 1987, Saubermann and Heyman 1987, Fiori et al. 1988, Johnson et al. 1988). By combining low temperature specimen preparation methods with computerized electron beam placement, an elemental map can be produced which gives information in a readily interpretable and understandable manner. By using well established electron probe x-ray analytical algorithms for quantitation of x-ray signals obtained through energy dispersive systems, each pixel can yield valuable quantitative information. When many pixels are combined, the statistical power of this form of analysis can be appreciated. However, and perhaps even more importantly, when grey level values are assigned to a range of pixel values in forming a map, it is possible for the eye and brain to do the statistical processing in an intuitive fashion. Since information on several elements can be acquired simultaneously, it is also possible to superimpose, and, thereby, relate elemental concentrations to each other as well as to anatomic structure. When quantitated, elemental mass is related to wet weight mass, as well as to dry weight mass. By using frozen hydrated cryosectioning techniques and analysis methods, special elemental relationships often become apparent (Saubermann and Stockton 1988, Saubermann and Heyman 1987). However, direct x-ray imaging of frozen hydrated tissue sections presents a formidable list of problems. Among these problems are 1) low peak to background ratios of biological elements normally found in small wet weight concentrations; 2) indistinguishable morphology due to the lack of contrast making it very difficult, if not impossible, to identify an area to be mapped; and 3) a specimen which is highly radiation sensitive such that water loss occurs at probe currents necessary to permit statistically acceptable x-ray generation rates. The methods which we have developed, use an algorithm which basically circumvents these major problems. Because changes in dry weight mass fraction of elements may not reflect changes in wet weight mas fractions (since cell water content can change for various physiological and pathological reasons), it may be important to determine and appreciate both wet weight and dry weight mass fractions in a biological specimen (Saubermann and Stockton, 1988).