Compensation of the contrast and brightness attenuation with depth in confocal microscopy

Abstract

A confocal laser scanning microscope (CLSM) captures images from a biological specimen in different depths and provides one with a stack of precisely registered fluorescent images. However, image intensities suffer from light loss distortions showing contrast and brightness degradation of images with depth. This effect causes problems in subsequent analysis and segmentation of biological objects. Therefore, we developed a post-processing method to correct this problem. It is a two step algorithm. In the first step the algorithm determines a standard histogram of the stack of images and in the second step histograms of individual images are warped according to this standard histogram. The algorithm results in more stable brightness and contrast of images through the stack.

A statistical approach for intensity loss compensation of confocal microscopy images

“… Pre‐processing methods : Ones that correct the intensity loss with modified optics as the images are being captured (Atkins & De Paula, 1994; Chen et al , 1995; Becker, 1996; Song et al , 1996). More recently, a method, which manipulates the photomultiplier gain, was introduced to counter the high intensity losses in the deep layers of the specimen by Ĉapek et al (2006). Post‐processing methods : These methods compensate the images after they are captured (Rigaut & Vassy, 1991; Oostveldt et al , 1998; Ortiz et al , 1999; Kervrann et al , 2004; Ĉapek et al , 2005). Intensity decay function (IDF)–based methods model intensity loss in the images as a parametric decay function of depth or time.…”

“…Post‐processing methods : These methods compensate the images after they are captured (Rigaut & Vassy, 1991; Oostveldt et al , 1998; Ortiz et al , 1999; Kervrann et al , 2004; Ĉapek et al , 2005). Intensity decay function (IDF)–based methods model intensity loss in the images as a parametric decay function of depth or time.…”

“…Negahdaripour & Yu (1993) apply a general model in which the horizontal and vertical flow fields as well as additive and multiplicative intensity relationships are estimated for every pixel. According to Ĉapek et al (2005), this approach is computationally expensive. A least‐squares optimization‐based approach, which optimizes brightness and contrast, is proposed by Periaswamy & Farid (2003) and Kervrann et al (2004).…”

“…The method discussed in Kervrann et al (2004) is not only sensitive to noise which can be eliminated by median filtering but also to the dynamic movement of objects in neighbouring optical sections. According to Ĉapek et al (2005), this gives erroneous and unstable results even in the presence of a very few outliers in optical sections. In Cox et al (1995), intensity variations are corrected based on histogram warping, but it is restricted to the case where a global, spatially invariant, non‐linear, monotonically increasing relationship exists between the intensities of the two images.…”

“…In Cox et al (1995), intensity variations are corrected based on histogram warping, but it is restricted to the case where a global, spatially invariant, non‐linear, monotonically increasing relationship exists between the intensities of the two images. Ĉapek et al (2005) extend the approach of Cox et al (1995) and attempt to give a general and fully automatic method of correcting intensity loss in confocal microscopy images. The proposed method manipulates the image histogram as in Ĉapek et al (2005), but it focusses on the continuous domain of probabilities to filter the foreground information to calculate the correction parameters.…”