Donois E, Freund O, Surlève-Bazeille J E, Taïeb A
Département de Biologie Cellulaire et Microscopie Electronique, Université Bordeaux I, Talence, France.
Microsc Res Tech. 1997 Feb 1;36(3):188-200. doi: 10.1002/(SICI)1097-0029(19970201)36:3<188::AID-JEMT7>3.0.CO;2-L.
The aims of the study were to write an image analysis (IA) program allowing the stereological quantification of human epidermal melanocyte melanization at the ultrastructural level and to specify the suitable preparative methods, in keeping with IA limits and stereological principles. Micrographs of cultured human melanocytes obtained in transmission electron microscopy were digitized with a scanner. The key step of the designed IA program is a thresholding based on the gray levels. Hence, gray level histograms (pixel frequency as a function of gray level) of melanocyte images exhibit a peak specific to melanin. The gray level thresholding used consists in isolating the melanin pixels that form profiles on a binary image and in storing the numerical data produced for a given melanocyte profile. These primary data are used to calculate numerous parameters via stereology with melanocyte cytoplasm and melanized melanosome as main reference spaces. The most important stereological parameters obtained are v(mi,cy) (melanin volume per average cell), v(mi,m) (melanin volume per average melanized melanosome), and nm (number of melanized melanosomes per average cell), and their validity is discussed. Melanocytes embedded in situ were abandoned for stereological reasons but pelleted melanocytes were found suitable. Using this computerized tool and stereology, we are able to perform quantitative studies producing varied data even from small cell samples. To our knowledge, this is the first stereological approach for quantifying intracellular melanization. A quantitative comparison of spectrophotometrical results (melanin assay) with stereological results obtained in ultraviolet B-irradiated Caucasian epidermal melanocytes will be performed in order to appraise this method.
本研究的目的是编写一个图像分析（IA）程序，以便在超微结构水平上对人类表皮黑素细胞的黑素化进行体视学定量，并根据IA的局限性和体视学原理确定合适的制备方法。通过扫描仪将透射电子显微镜下获得的培养人类黑素细胞的显微照片数字化。所设计的IA程序的关键步骤是基于灰度级的阈值处理。因此，黑素细胞图像的灰度级直方图（像素频率作为灰度级的函数）呈现出一个特定于黑色素的峰值。所使用的灰度级阈值处理包括在二值图像上分离形成轮廓的黑色素像素，并存储为给定黑素细胞轮廓生成的数值数据。这些原始数据用于通过体视学计算众多参数，以黑素细胞细胞质和黑素化黑素小体作为主要参考空间。获得的最重要的体视学参数是v(mi,cy)（平均每个细胞的黑色素体积）、v(mi,m)（平均每个黑素化黑素小体的黑色素体积）和nm（平均每个细胞的黑素化黑素小体数量），并对其有效性进行了讨论。由于体视学原因放弃了原位包埋的黑素细胞，但发现沉淀的黑素细胞是合适的。使用这种计算机化工具和体视学，即使从小细胞样本中我们也能够进行产生各种数据的定量研究。据我们所知，这是第一种用于量化细胞内黑素化的体视学方法。将对紫外线B照射的白种人表皮黑素细胞的分光光度法结果（黑色素测定）与体视学结果进行定量比较，以评估该方法。
