Non-invasive evaluation of egg quality and early life stage development of medaka fish (Oryzias latipes) by means of morphological image analysis methods/Neinvazinis medaku (Oryzias latipes) oocitu kokybes bei embriono vystymosi vertinimas matematines morfologijos metodais.
Animal models are often used to investigate environmental contamination by various industrial wastes and the toxicity of various potential drugs and substances used in the production of clothes and food. The tendency nowadays is to reduce the cost of these experiments and the number of organisms, in concern for animal welfare. A notable example is the awareness of the ecotoxicologist community of the pharmaceutical industry to apply the principles of replacement, reduction, and refinement (the 3Rs). This was established by Russell and Burch (1959), in the context of regulatory environmental assessments . One of the alternatives proposed is to use the early life stages of fish as an experimental model [2,3]. Fish embryos and larvae are generally considered to be the most sensitive stages in the life cycle of teleosts .
Investigation of early life stages is also important for estimating the chances of survival of particular fauna species facing global environmental changes.
Amongst the most popular species used for such investigations are zebrafish (Danio rerio), medaka (Oryzias latipes), rainbow trout (Onchorhynchus mykiss) and fathead minnow (Pimephales promelas). These popular species are mostly selected because of their developmental and physical features. For instance, adult zebrafish can produce a large number of translucent embryos every week, by external fertilization, and the embryogenesis is well characterized [5,6,7]. Medaka (Oryzias latipes) have similar features . Visual investigation of embryo development is totally noninvasive and gives promising results. Recent technologies allow visual investigation not only of early life stages but also of adult animals. A product of advanced genetic techniques is the so called "see-through" medaka , which is genetically deficient of pigments and whose internal organs can be observed. Digital imaging could be used in such investigations for quantitative evaluation and more reliable results.
Several standards (OECD 210, ASTEM E 1241-98, OECD 212, ISO 12890, DIN 38415-T6) are regulating the toxicological characteristics and endpoints used for evaluation. The toxicological endpoints are cumulative mortality, pro-larval hatching (rate, time for 50%), larva pro-larval length and weight, larval abnormalities (morphology, behavior), coagulated eggs, somite development, tail detachment and heartbeat.
Evaluation of oocyte quality is one of the methods used not only in environmental tests but in infertility treatment too. Many important events are triggered in the oocyte at fertilization. One of these events--the cortical reaction--is a secretory event involving a wave of exocytosis. It helps prevent polyspermy and leads to hardening of the envelope (chorion) surrounding the oocyte which is equivalent to the zona pellucida in human eggs. This reaction could be observed as changes in the distribution and density of cortical granules. It is possible to evoke this reaction artificially, triggering activation of the oocyte by an electrical stimulus. This evokes all the events normally triggered by fertilization apart from subsequent cell division for which the sperm centriole would be required. Quantitative evaluation of density, number of cortical granules and the dynamics of changes provide new possibilities for studying secretion and egg quality. The quantitative evaluation of the number and distribution of cortical granules by fluorescent imaging was reported in . However, invasive staining procedures were applied to make the cortical granules visible for the imaging system. The procedures make the method unsuitable for most environmental, toxicological or clinical investigations. The application of advanced methods of digital image processing could allow us to achieve similar results and make the evaluation of quality of eggs possible in the desired environment in a noninvasive way.
Disorders in the formation of the cardiovascular system of zebrafish embryos are reported as an indicator of environmental contamination by cadmium . Fluorescein labelled carboxylated latex micro-beads were used to visualize the cardiovascular system of the embryo. Visualization of the cardiovascular system is very complicated and invasive for the embryo. Special image processing methods could permit visualization of the cardiovascular system in a non-invasive way.
In general, characteristics and endpoints (as in the examples provided here) are determined by investigatory technologies that are already available. Advanced methods for digital imaging in many cases could allow non-invasive investigations in desired environment and also allow more reliable quantitative evaluation of the parameters.
In this study we demonstrate how advanced image processing methods could be applied in the evaluation of egg quality and investigation of the development of embryos of the fresh water teleost fish, medaka (Oryzias latipes). Egg quality (evoked cortical reaction) and two parameters characterizing development of the cardiovascular system (vascularization of the embryo and detailed analysis of its heart rhythm) were also evaluated.
Intact unfertilized (nonactivated) oocytes were obtained following natural oviposition from medaka fish (Oryzias latipes) raised in sea water collected from Cancale, North Brittany (France). Such oocytes were osmotically shrunk and for successful activation had to be reswollen in 10% or 25% sea water diluted with distilled water. Oocytes were activated electrically with a +/- 10 volt pulse of 1 to 10 milliseconds duration through Ag/AgCl electrodes placed on either side of the oocyte on a glass slide.
Bright field images of developing embryos of medaka (Oryzias latipes) placed in tap water were taken every day starting from the 2nd to the 8th day postfertilization.
Images were taken using a Zeiss Axiovert 35 inverted microscope equipped with a Moticam 3000 (Motic, China) digital camera connected to the personal computer with a 1.5 GHz Intel Centrino processor and 512 MB RAM. Images were taken at 760x520 pixels resolution and 24-bit color encoding. For the egg cortical reaction frames were taken every second for two minutes starting from the beginning of electrical stimulation. Development of the embryo was recorded every 15 seconds at 15 frames/second and 320x240 pixels resolution. All image processing calculations were made with software elaborated using the MATLAB[R] Image processing toolbox.
Evaluation of the area of the egg covered by cortical granules: Cortical granules are spherical objects of a certain size on the image. Oil droplets are also spherical objects but of varying and often much bigger size. Mathematical morphology methods based on Minkovski addition and subtraction were applied to determine size . Granulometry likens image objects to stones whose sizes can be determined by sifting them through screens of increasing "mesh" size and collecting what remains after each pass. Image objects are sifted by opening the image with a structuring element of increasing size and counting the remaining intensity surface area (summation of pixel values in the image) after each opening. The opening procedure is a combination of erosion of the original image A by structuring element B followed by dilation of the resulting image by the same structuring element B:
A [omicron] B = (A[THETA]B) [direct sum] B. (1)
Preliminary granulometry measurements of images using "disk" structuring elements revealed a radius of cortical granules of 5-6 pixels on the registered images. All round objects with a radius of 5-6 pixels were extracted subtracting images obtained after opening the original image with a structural element of radius 7 from images obtained after opening with a structural element of radius 5. More detailed information about fast morphological image transforms using bitmapped images is presented in . Images containing extracted round objects were converted into black and white format, assigning a value of "1" or "0" according to the condition:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII], (2)
where X--the array of pixels of the picture. Coefficient values of 0.3 and 0.1 were selected during preliminary tests of the method. So, values of pixels in the transformed image belonging to the recognized objects were equal to "1" and zeroes were assigned to the rest of the pixels. The area covered by objects on the ordinary frame [S.sub.i] was estimated as the sum of pixel values:
[S.sub.i] = [m.summation over (j=1)] [n.summation over (k=1)] [a.sub.j,k], (3)
where a--values of the pixels, m and n--dimensions of the frame--760 and 520 respectively.
Visualization of blood vessels: Moving blood cells are visible in the vessels of the developing embryo and differences between the values of pixels of consequent frames in the movie can trace them. The accumulated difference of images was calculated as follows:
[??] = [N.summation over (i=1)] [A.sub.i], (4)
where [A.sub.i]--a difference of subsequent frames:
[A.sub.i] = [absolute value of [F.sub.i] - [F.sub.i-1]], (5)
[F.sub.i] and [F.sub.i-1] are the arrays of pixel values of subsequent frames. Elements of matrix [??] were filtered using the threshold level to get rid of some artifacts and non significant image distortions and converted into black and white format. Values of "1" or "0" were assigned to the elements of the matrix according to the same condition (2) used for cortical granules. The threshold values were also selected during preliminary tests of the method. Accumulations of white pixels ([a.sub.j,k]=1) were consolidated into clusters by means of fast morphological image transform using the image closing function:
A x B= (A [direct sum] B)[THETA]B, (6)
where B was a "disk" structuring element of radius 3 . The area S covered by these clusters was estimated in the same way as in the case of cortical granules (3). All calculations were made with software elaborated using the MATLAB[R] Image processing toolbox.
Registration of heart beats: The photo detector (LDR 2-20K [empty set]mm) was mounted in the ocular eyepiece of the microscope and together with a DC power source connected to the ADC of the computer's sound card. The microscope aperture was used to hide all the light passing through the embryo except at the location of the beating heart. Illumination changes due to heart movement were converted to an electrical signal containing peaks at the time of heart contraction. The complicated shape of the peaks meant that the amplitude detector could only be used for the preliminary detection of the heart beats. Final adjustment of heart beat moments was performed by maximizing the correlation between the preliminary detected ordinal????? signal peak and the model with a permanently updated shape. The heart beat moments were used to form, so called in cardiology, R-R interval sequences of the heart beats of the embryo. Evaluation of the average heart rate in developing fish embryos is a well known and popular method . However, detailed analysis of heart rate variability could reveal functions of autonomous heart control that should appear together with the developing nerve system. Spectral analysis of fetal heart rate variability is known and used in clinical practice . So we decided to perform spectral analysis of the R-R interval sequences detected from the developing embryos.
Typical pictures of the oocyte before and after activation following electrical stimulation are shown on Fig. 2. One can see the whole area of the oocyte covered by small round objects - cortical granules on the upper left image which was taken before activation. Almost no such objects remain on the bottom left image, which was taken 15 seconds after the start of the activating electrical stimulus. Pictures on the right of Fig. 2 illustrate the results of the image processing procedure. Black and white format images are obtained from the images shown on the left. Only some oil droplets, bigger than the round cortical granules on the initial images were misrecognized as cortical granules, but this does not crucially influence the final result. The calculated area covered by cortical granules before activation was in the range of 30%-40% of the total area of the oocyte and decreased to 20%-25% at the end of the reaction. Dynamics of the reaction were different and steady state values after activation were reached in times varying from 15 to 45 sec. The aim of this study was to show that cortical granules could be recognized by this method and the area covered by them could be evaluated. Therefore, we tested the method on as many different oocytes as possible. But in this study we do not provide any comparison of the image processing results in regard to the conditions of oocyte preparation etc. A variety of normalized time courses of area covered by cortical granules following activation of oocytes by electrical stimuli is presented in Fig. 1. Expert control confirmed that recognition of cortical granules in all presented cases was of sufficient quality.
[FIGURE 1 OMITTED]
Typical pictures of medaka embryos on the 2nd and 8th day after fertilization are shown on the left of Fig. 3, upper and bottom images respectively. The image processing result--black and white images with recognized blood vessels marked in white on black background are presented on the right of Fig. 3. The accuracy of marking of the blood vessels was considered satisfactory. The area covered by blood vessels increased in average about 4 fold by the 8th day when compared to the 2nd day after fertilization.
The average heart rate increased during development of the embryo in all registered cases. Time intervals between consecutive heartbeats (R-R intervals) registered in embryos on the 2nd and 8th day after fertilization are shown in Fig. 4. Besides the increase in the heart rate the variability also increases. The power spectrum of the R-R interval sequences is presented in Fig. 5. Some peaks appearing at about 0.6 Hz frequency range on later days of development were present in all registered cases.
The recognition of cortical granules of the artificially activated oocyte with sufficiant reliability is a promising result opening a perspective for non-invasive investigations of the quality of oocytes not only in toxicological or environmental investigations, but also in clinical investigations. The limitations of this method are that the area we observe in our pictures is a projection of just part of the sphere concerning both the oocyte and the cortical granules. In fact the observed process reflects exocytosis--release of substances into the perichorionic space of the oocyte--and should be evaluated in terms of volume. Estimation of the total volume of the oocyte would not be a big problem, however, for the estimation of the total amount of substance released from cortical granules we could calculate this only by knowing what part of the total amount the visible granules represent. This would require additional investigations.
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
[FIGURE 5 OMITTED]
Nevertheless, the same part of the oocyte is visible during the whole investigation and our relative figures of dynamics of area covered by cortical granules reflects the oocyte activation process and could allow better evaluation of oocyte quality and the activation process.
Many parameters or key points in embryo development are usually visually evaluated because of the difficulties to elaborate morphological models in a quantitative way that describe the developing structures. In this study we present only the first and easiest to calculate parameter characterizing development of blood vessels. However, we think it is a first step towards application of so called blind image analysis methods successfully used in many areas including biomedical image analysis . It gives generalized quantitative estimates to the objects and structures without detailed morphological analysis. Advanced morphological image analysis could be applied to reveal special morphological development disorders giving them quantitative estimates. We expect that blind image analysis methods could be useful for evaluation of cardiac development disorders in embryos as well as the quality of eggs.
Average heart rate is known as a representative parameter for evaluation of embryo development. Observed heart rate variability in most cases is related to autonomous heart activity control. It reflects nerve system actions regulating heart rhythm according to the demands of the whole organism. Autonomous fetal heart activity control reflecting rhythm variations are known in clinical investigations , but there is a lack of information about heart rate variability in fish embryos. Maybe it is due to a lack of suitable methods of evaluation. However, detailed analysis of R-R intervals could allow detection of rhythm disorders--indicators of critical states of the embryo. Time-frequency analysis of R-R interval sequences provides a means of evaluating possible autonomic regulation of heart activity. Appearance of a peak in the 0. 6.z range in the power spectrum of R-R interval sequences in later stages of development in our recordings correlates with widely known frequency characteristics of autonomous heart control. It supports the idea about the possibility of evaluation of heart control mechanisms.
Advanced methods for biomedical imaging and signal processing applied to the analysis of images of early life stages could be a powerful instrument for realizing principles of replacement, reduction, and refinement (the 3 Rs) in pharmaceutical and environmental investigations.
Received 2009 02 10
[1.] Hutchinson T. H., Barrett S., Buzby M., Constable D., Hartmann A., Hayes E., Huggett D., Laenge R., Lillicrap A. D., Straub J. O., Thompson, R. S. A strategy to reduce the numbers of fish used in acute ecotoxicity testing of pharmaceuticals. Environ. Toxicol. Chem. 2003: 22, 3031-3036.
[2.] Oberemm A. The use of a refined zebrafish embryo bioassay for the assessment of aquatic toxicity. Lab. Anim. 2000: 29, 32-40.
[3.] Nagel R., Dar T. The embryo test with the zebrafish Danio rerio--a general model in ecotoxicology and toxicology. Altex Altern. Tierexp. 2002: 19, 38-48.
[4.] Shuk Han Cheng, Po Kwok Chan, Rudolf Shiu Sun Wu. The use of microangiography in detecting aberrant vasculature in zebrafish embryos exposed to cadmium // Aquatic Toxicology.--2001.--No. 52.-P. 61-71.
[5.] Hisaoka K. K., Battle H. I. The normal developmental stages of the zebrafish, Brachydanio rerio (Hamilton-buchanan)//J. Morphol.--1958.-No. 102.-P. 311-327.
[6.] Kimmel C. B., Ballard W. W., Kimmel S. R., Ullmann B., Schilling T. F. Stages of embryonic-development of the zebrafish//Dev. Dyn.--1995. -No. 203.-P. 253-310.
[7.] Lele Z., Krone P. H. The zebrafish as a model system in developmental, toxicological and transgenic research // Biotechnol. Adv.--1996.--No. 14.-P. 57-72.
[8.] Yamamoto T. Medaka (Kilifish). Biology and Strains. Tokyo: Keygaku Publishing Co., 1975.-P. 17-29.
[9.] Iwamatsu T., Nakamura H., Ozato K., Wakamatsu Y. Normal growth of the "See-Through" Medaka // Zoological Science.--2003.--No. 20.-P. 607-615.
[10.] Wang et al. Quantified Analysis of Cortical Granule Distribution and Exocytose of Porcine Oocytes during Meiotic Maturation and Activation // Biology of Reproduction.--1997.--No. 56.-P. 1376-1382.
[11.] Pierre Soille. Morphological Image Analysis; Principles and Applications. ISBN 3540-65671-5.-1999.
[12.] Rein van den Boomgard, and Richard van Balen. Methods for Fast Morphological Image Transforms Using Bitmapped Images // Computer Vision, Graphics, and Image Processing: Graphical Models and Image Processing.--1992.--Vol. 54, No. 3.--P. 252-254.
[13.] Heideman W., Antkiewicz D. S., Carney S. A., Peterson R. E. Zebrafish and cardiac toxicology // Cardiovasc Toxicol.--2005.--No. 5(2).--P. 203-214.
[14.] Van Laar J. O., Porath M. M., Peters C. H., Oei S. G. Spectral analysis of fetal heart rate variability for fetal surveillance: review of the literature // Acta Obstet Gynecol Scand.--2008.--No. 87(3).--P. 300-306.
[15.] Andrzej Cichocki and Shun-ichi Amari. Adaptive Blind Signal and Image Processing. ISBN 0471 60791 6.-2002.
Institute for Biomedical Research, Kaunas University of Medicine, Eiveniu str. 4, LT-50009 Kaunas, Lithuania, phone: +370 37 302952; e-mail: email@example.com
D. J. Webb
Universite de Rennes 1, Batiment 25, Campus de Beaulieu 35042 Rennes Cedex, France, Phone: +33 (0)2 23 23 66 95; e-mail: firstname.lastname@example.org
Institute for Biomedical Research, Kaunas University of Medicine, Eiveniu str. 4, LT-50009 Kaunas, Lithuania, phone: +370 37 327367; e-mail: email@example.com
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||MEDICINE TECHNOLOGY/MEDICINOS TECHNOLOGIJA|
|Author:||Krisciukaitis, A.; Webb, D.J.; Grigaliunas, A.|
|Publication:||Elektronika ir Elektrotechnika|
|Date:||Jun 1, 2009|
|Previous Article:||Evaluation of the prognostication of cardiovascular events within one year in patients with ischemic syndromes/Iseminiais sindromais serganciu...|
|Next Article:||Evaluation of coherence of T wave in different leads/Elektrokardiogramos T dantelio sasaju vertinimas.|