The principle problems with image analysis of asbestos fibers include: the complexity of many fiber shapes, including bundles, agglomerates, and split fibers; the fibers often go in and out of the plane of focus; the background includes many particles and other non-fibrous shapes; the phase contrast optics produces haloes around particles in the sample that can be detected as fibers; and finally, and perhaps most importantly, the contrast between the fibers and background is poor and many fibers are near the detection threshold. An evaluation of the MAP program indicated that a significant fraction of the fibers were misidentified as multiple fibers, not detected at all, and groups of compact particles or edges of large particles were detected as fibers [40]. Inoue and coworkers have more recently developed image analysis software using a microprocessor-based PC [41]. Initial tests indicate that it works approximately as well as human counters. Inoue also evaluated how well human counters and the image analyzer did in detecting the same fibers in a sample and found that only about 50% of the fibers were consistently counted by all counters, so the image analysis system did approximately as well as the human counters [42]. Further testing of the image analysis system is needed. In addition to image analysis, optical microscopy can be enhanced using a personal computer to more easily observe the image and to mark and measure fiber dimensions, with automatic recording of the fibers counted [43]. This does not appear to improve the counting accuracy since the analyst still decides which fibers are to be counted. 4.
POLARIZING LIGHT MICROSCOPY (PLM) OF BULK MATERIALS (Adapted from Baron [44]) The asbestos fibers in bulk material can be released and become airborne when the bulk material is disturbed. For this reason, it is desirable to measure the asbestos content of bulk samples. PLM is often used to determine the percent asbestos in bulk material. The EPA [45] has defined asbestos containing material (ACM) as material containing more than 1% asbestos using the PLM method, which effectively estimates concentration by area observed. Some confusion exists regarding the units of asbestos percentage. EPA originally indicated that the limit for ACM was 1% by mass [45], but because of the difficulties in determining corrections for differences in material density and in determining particle volumes, the limit was changed to 1% by area as determined by the PLM method [46]. OSHA does not specify units for percent asbestos in its regulations [47]. Several PLM techniques are used for identifying fiber type as well as semi-quantifying the percent fibrous material (usually asbestos) in a sample [48-52]. These techniques depend on particle shape, the refractive index, and other optical properties of individual particles. Many of these PLM techniques require visual observation of color in the fiber and become less reliable for fibers thinner than about 1 :m [53]. a. Sampling Several procedures have been suggested for obtaining representative bulk samples of ACM in a fashion that prevents unnecessary exposure to asbestos aerosol [54-56]. Representative sampling of commercial ACM materials is often problematic; these materials may vary significantly in asbestos concentration between nearby locations and even at different depths
3/15/03
151
NIOSH Manual of Analytical Methods
