- Methodology article
- Open Access
Extended morphological processing: a practical method for automatic spot detection of biological markers from microscopic images
© Kimori et al; licensee BioMed Central Ltd. 2010
- Received: 6 November 2009
- Accepted: 8 July 2010
- Published: 8 July 2010
A reliable extraction technique for resolving multiple spots in light or electron microscopic images is essential in investigations of the spatial distribution and dynamics of specific proteins inside cells and tissues. Currently, automatic spot extraction and characterization in complex microscopic images poses many challenges to conventional image processing methods.
A new method to extract closely located, small target spots from biological images is proposed. This method starts with a simple but practical operation based on the extended morphological top-hat transformation to subtract an uneven background. The core of our novel approach is the following: first, the original image is rotated in an arbitrary direction and each rotated image is opened with a single straight line-segment structuring element. Second, the opened images are unified and then subtracted from the original image. To evaluate these procedures, model images of simulated spots with closely located targets were created and the efficacy of our method was compared to that of conventional morphological filtering methods. The results showed the better performance of our method. The spots of real microscope images can be quantified to confirm that the method is applicable in a given practice.
Our method achieved effective spot extraction under various image conditions, including aggregated target spots, poor signal-to-noise ratio, and large variations in the background intensity. Furthermore, it has no restrictions with respect to the shape of the extracted spots. The features of our method allow its broad application in biological and biomedical image information analysis.
- Original Image
- Synthetic Image
- Mathematical Morphology
- Spot Detection
- Biological Image
Biological imaging such as confocal fluorescence microscopy and electron microscopy require the use of protein-labeling techniques to localize individual proteins within cells. Biological markers such as green fluorescence protein  and a variety of fluorescent dyes [2, 3] for fluorescence microscopy, and colloidal gold [4, 5] for electron microscopy are widely used. Molecules labeled with biological markers are generally observed as small specific spots against a background of high brightness. Quantitative comprehension of the localization and statistical distribution of the spots are essential for deciphering biological information. In general, cellular microscopic images have a low signal-to-noise ratio (SNR) and the differences in intensity between signal spot and background are not always clear. Moreover, the texture of those backgrounds is complicated. For these reasons, microscopy images are often difficult to manage computationally. Currently, there are several automatic processing and recognition systems for biological images and they have been applied in the quantitative analysis of biological objects ranging from molecules to cells to whole organisms [6–10].
The purpose of this study was to extract and characterize biological spots of intricate morphology and low contrast in an automatic manner. Current standard techniques for spot extraction consist of edge enhancement for image morphology, including discrete convolution by a high-pass mask and the use of first- or second-order differential operators, based on the magnitude of the spatial differences of the spots . One major problem with this approach, however, results from the blurring and degradation of the image contrast during image acquisition. For some spots with weak contrast, edge extraction is not sufficient. In real-world applications, most biological images contain object boundaries, artifacts, and noise. Therefore, edge enhancement filters may cause difficulties in distinguishing the exact edge of the object's structure from artifacts such as trivial geometric features. Additionally, these techniques can amplify background noise in the image while enhancing the object edge [12, 13].
In other methods based on conventional frequency-selective filters [14–18], the precise localization of low-contrast spots may not be possible. High-density areas resulting from the integration of many spots may not allow the isolation of individual spots through frequency-selective filters. In addition, the parameter settings are often so complex as to require their modification whenever the target spot images are changed [19, 20]. Furthermore, these methods cannot deal with the varied morphology of the spots.
Spot extraction methods based on conventional mathematical morphology  effectively capture the spots' location and their shape information [22–26]. These methods employ a morphological algorithm for background subtraction known as the top-hat transformation  or rolling-ball transformation . It is well recognized that the principle of these methods is very effective for extracting a target object from a wide variety of image types [29–34].
Morphological operations use small synthetic images called structuring elements (SEs), which are a fundamental tool in mathematical morphology. The SE used as a probe moves along each pixel of the image. To apply morphological filtering for spot extraction from various types of biological images, the procedure to determine the shape and size of the SE is very important. A commonly used SE shape is the square or disk. In the rolling-ball transformation, a ball-shaped SE (such as a disk SE with weights arranged in order to describe a hemisphere in gray scales) is used. In the above-described methods for spot extraction, these SEs were also used. However, most small contiguous spots cannot be individually distinguished, such that several spots are extracted as one connected region because the size (width) of the SEs is wider than the minimum distance between the peaks of adjacent spots. A suitable SE shape for spot extraction includes a straight-line segment (a fuller description of which is given in the Methods and Results sections); however, since processing by common morphological operations with a single line-segment SE is not isotropic, it cannot consider the geometrical details of an intricate image. Thus, for spot extraction, conventional morphological processing is not effective.
Advanced morphological processing with multiple SEs has been reported [35, 36]. In this approach, multiple sets of line-segment SEs generated by rotation of the single line-segment SE in different directions are applied. However, in the discrete space of the images, it is difficult to generate a straight line-segment as the SE that can be rotated in an arbitrary direction. This restriction in the rotational direction of the SE prevents adequate spot detection in complicated biomedical images.
In this study, we solved these problems by introducing a simple and practical approach, an extended mathematical morphology, into the automatic detection of spots in biological images. This technique is based on top-hat transformation with a single SE as the straight line-segment. In our algorithm, an original image can be rotated in arbitrary directions with respect to the single SE. This novel method, which we named rotational morphological processing (RMP), can homogeneously treat with geometrical features in an image under various orientations. Top-hat transformation based on RMP has been applied to spot extraction, in the absence of any hypotheses to fit the spots by 2-D Gaussian shape or minimal intensity. Finally, by isotropic processing with the line-segment SE, contiguous spots can be segmented into individual parts. Our novel method was developed in order to automatically extract spots, such as biological markers consisting of antibodies conjugated with fluorescent molecules, from a biological image of intricate morphology and low contrast.
Smal et al. evaluated the spot detection methods most frequently used in fluorescence microscopy , including wavelet-based multiscale detecting [16, 18], morphological based methods [23, 24, 38], and the machine learning method . In this study, we compared the performance of our proposed method with other morphological based methods, such as conventional top-hat transformation and h-dome transformation, by using synthetic-noise images.
This report is organized as follows. A brief introduction describing the basics of conventional mathematical morphology is followed, in the Methods section, by a detailed presentation of our spot extraction technique. In the Results section, the application of the detection method to synthetic images as well as to real image data from electron and fluorescence micrographs is discussed. In the final section, the effectiveness of our novel method is summarized and evaluated.
Conventional mathematical morphology
Mathematical morphology is based on set-theory concepts of the shape of an objective image . An image can be represented by a set of pixels. Morphological operations always deal with a set of two images: an objective image and a SE. Each SE has shape and size characteristics as parameters of the operation. Let f denote a gray-scale image function from Z2 into [0, I-1], where I is a positive integer. Let B denote a binary SE. The fundamental operators of mathematical morphology are dilation and erosion.
where D f and D b are the domains of the functions f and B, respectively. The opening and closing operations are delivered from dilation and erosion.
The top-hat transformation is one of the commonly used morphological operations for extracting local bright objects from a low contrast image in gray-scale . It is obtained by subtracting from the original image f the opening image γ B using the SE B.
It yields an image in which all the residual features (peaks and ridges) are subtracted by the opening operation. Adding these residual features to the original images has the effect of accentuating objective structures with high intensity . If the difference in intensity between the target objects and the background of the image is markedly small, it is difficult to detect these differences with the human eye. However, these low-contrast objects can be extracted and enhanced by the top-hat transformation.
Another method to extract the local bright object in biological images, based on mathematical morphology, is the h-dome transformation .
where D h (f) is the h-dome image of a gray-scale image f, (f-h) represents the result of subtracting a constant value h from the gray-scale image, and ρ f (f-h) the morphological reconstruction of the gray-scale image from f-h. The gray-level reconstruction is obtained by iterative geodesic dilation of f-h under f until stability is reached .
Spot extraction filter: top-hat transformation by RMP
The essential elements required for the spot extraction filter are processing of the biomedical image isotropically and isolation of the adjacent spots from the image background. In order to fulfill these requirements, our method proposes that the objective image can be rotated in arbitrary directions with respect to a single straight line-segment SE whose width has only 1 pixel. The width of this SE ensures determination of the minimum distance between two different spots for isolation. By using this SE, spots separated by distances of only 1 pixel can be distinguished individually. The length of the SE should be adjusted so that it is longer than the size of the target spot. A spot that is smaller than the length of this SE is extracted by the top-hat transformation.
This top-hat transformation by RMP with the straight line-segment SE consists of the following steps:
Algorithm 1 (Top-hat transformation by RMP with line-segment SE)
Original image rotation. The original image f (Figure 1a) is rotated in a clockwise direction with respect to the center of the image frame. Assume that dividing a half of the circle (π [rad]) into N equiangles gives us each direction at an angle of π/N [rad] (Figure 1c), which is an increment angle. Namely, f i (Figure 1d, top row) denotes the rotated image of f with the angle of π i/N [rad], where i = 0, 1,..., N-1.
Opening. All rotated images are subjected to an opening operation with the straight line-segment SE B (Figure 1b). The opening operations of the rotated image f i are represented as γ B (f i ) (Figure 1d, middle row).
Opened image rotation. The opened images (γ B (f i )) are rotated π·i/N [rad] in an anticlockwise direction. The rotation at i times of the opened image is denoted by h i (Figure 1d, bottom row).
Union of the rotated and opened images (opening by RMP). The processed images (h i ) are unified. In union processing, the maximum intensity value, which corresponds to the same pixel coordinate among all opened images, is taken to generate the whole image.
Top-hat transformation by RMP. The unified opened image (γ' B (f)) is subtracted from the original image (f).
This operation is used as the spot extraction filter in our proposed method.
In this study, we utilized the top-hat transform by RMP for spot extraction from biological images obtained with electron and fluorescence microscopy. The entire practical process consists of the following steps:
Algorithm 2 (Spot extraction for practical biological images)
Noise reduction: Noise, which is less than the resolution limit of the micrograph or target spot, is removed via opening by RMP (equation (7)) with the straight line-segment SE. The length of the SE is set to be smaller than the diameter of the target spots.
Spot extraction: The spots are extracted by the top-hat transformation by RMP (equation (8)) with the straight line-segment SE. The length of SE is set to be larger than the diameter of the target spots.
Binarization: The extracted spots are binarized by equation (9) for recognition and measurement of the spots computationally.
Namely, the pixels of residual regions by the top-hat transformation by RMP are assigned an intensity of 255 in an 8-bit gray-scale value.
In several cases, the subtraction process and the binarization process leave the small isolated pixels on the image such that they represent residual background noise. The conventional opening operation (equation (3)) can also be applied to remove the remaining noise. This post-processing should be adapted to the particular application.
Isolation of overlapped spots
The opening operation can be geometrically processed by pressing the SE up against the surface of the original image f and sliding it underneath the entire surface. The surface of the opened image is constructed from the highest point in the region reached by the SE. Since the line-segment SE B L has a width of 1 pixel, in the process of RMP opening the SE fits the narrow intervening space between the spots. Thus, line-segment SE reaches the baseline of the individual spots (i.e., the upper level of the overlapped region). In contrast, the disk SE B D has a width of 11 pixels as diameter. Since it is larger than the intervening space of the spots, the SE cannot fit the space. Thus, the disk SE cannot reach the level at which the two spots are distinguished. Accordingly, the spots were not isolated by the conventional top-hat transformation.
Verification of optimal number of rotational direction
where F(i, j) denotes the original image, F'(i, j) the filtered image by RMP opening, and m 1 × m 2 the total number of pixels.
In addition to this experiment, performance was also tested using the multiple SEs method, shown as a red line in Figure 3e. The straight line-segment SE was used for these tests. In the case of N = 1, the single SE (orientation θ = 0 [rad], horizontal direction) was applied; at N = 2, two SEs (θ = 0 and π/2 [rad]) were used, and at N = 4, four SEs (θ = 0, π/2, π/4 and 3π/4 [rad]) were used. These results showed that reconstruction by the multiple SEs was insufficient.
Comparison with conventional morphological methods by using synthetic-noise images
The Poisson-distributed noise was added to each source image with a uniform (type-A) and a gradient (type-B) background (Figure 5c). This is one of the main sources of noise in fluorescence microscopy imaging .
The PSNR between each source image and the corresponding noise-added image was calculated. The average value of the PSNR of the type-A image set was 13.848 ± 0.242 (mean ± SD) dB, while for the type-B image set it was 11.245 ± 0.169 dB.
In each image type, the synthetic images had a total of 150 spots; thus, TP + FN = 150. The results detected by our proposed method were compared with those of the conventional top-hat and h-dome transformations.
In this experiment, noise-added images were first smoothed with the Gaussian filter (3 × 3 kernel).
Processing by our proposed method followed algorithm 2. Since the width of the spot domain was 17 pixels (Figure 5b), any structure smaller than this width was regarded as noise or artifact. In step 1, the straight line-segment SE of 13 × 1 pixels was used, and in step 2, the straight line-segment SE of 21 × 1 pixels. In both steps, the rotational direction number (N) of RMP processing was 36. The subtracted image was binarized by the method in step 3.
For the method based on the conventional top-hat transformation, the processing in step 2 differs from that of the proposed method. Top-hat transformation was applied with disk SE (diameter: 21 pixels) and the subtracted image was binarized by the method in step 3.
For the method based on h-dome transformation, this transformation was applied to the smoothed image obtained in step1 of algorithm 2. We set the value of parameter h to 50, and the subtracted image (h-dome) was binarized by the method in step 3.
Finally, binary images obtained from these methods were cleaned by the conventional opening with disk SE (diameter: 13 pixels) as post-processing.
The results of the spot extraction are seen in Figure 5c. The type-A and type-B images are shown in the top and bottom row, respectively. All spots were correctly detected by our proposed method but not by the other methods. Figure 5d shows the actual values of recall, precision, and F-measure for each type image set. The performance in terms of the F-measure for the proposed method was consistently 100% in all type images. Thus, the proposed method is more tolerant of Poisson noise images than other methods.
In addition, the performance of our method was tested using synthetic images under various noise levels. For the source image (Figure 5a), nine degrees of Poisson-distributed noise images with uniform background were generated, with the PSNR decreasing from 17.732 to 7.797 dB. The upper part of Figure 5e shows the synthetic images, captured only in the rectangular region in Figure 5a. The proposed method was applied to these synthetic images with the same procedure and SEs as in the previous experiment. The bottom of Figure 5e shows that the scores for the recall, precision, and F-measure rates were consistently 100% in the PSNR range of 17.732 to 10.057 dB. Subsequently, the recall rate decreased with decreasing PSNR although the precision rate remained at 100%. This indicates that FP was zero and ensures the accuracy of our method for spot detection.
Spot extraction of colloidal gold particles
Processing of the spot extraction was carried out with algorithm 2. In the case of the micrograph containing 10-nm gold particles, line-segment SE with a size of 5 × 1 pixels (4.5 × 0.9 nm) and 13 × 1 pixels (11.7 × 0.9 nm) was used in noise reduction and spot extraction, respectively. The extracted spots were binarized and overlaid as seen in the red region on the original image. The result is shown on the right side of Figure 6a. For the electron micrograph containing the 1.8-nm gold particles, line-segment SE with a size of 3 × 1 pixels (0.96 × 0.32 nm) and 7 × 1 pixels (2.24 × 0.32 nm) was used for noise reduction and spot extraction, respectively. The result is shown on the right side of Figure 6b. For all extracted particles, the averaged Feret's diameter was calculated. The mean ± SD of the diameter of the extracted spots was 10.16 ± 0.77 nm (164 spots) for the 10-nm gold particles, and 1.81 ± 0.15 nm (813 spots) for the 1.8-nm gold particles. This result shows that the diameter of the extracted spots was consistent with the nominal diameter.
Spot extraction of fluorescent antibodies
By visual observation, the diameter of each spot was found to be about 5 pixels (ca. 300 nm). Therefore, a SE length longer than the diameter of the target spots was selected for the spot extraction process. Binarization was carried out using equation (9). The regions of the extracted spots were superimposed on the original image as red-colored regions (Figure 7b). From this micrograph, the 627 spots were extracted. The quantitative estimation of the bright spots corresponding to the caveolae is provided in Figure 7c. The sum of the intensity of the pixel values in each extracted spot region was calculated and the distribution depicted in a histogram. The total number of extracted spots was 6701 from five micrographs (which included approximately 5 cell regions). The median of the histogram was 1319. The largest cluster can be seen centered at the histogram's median value.
Our novel method to extract the spots in electron and fluorescence microscopic images uses the extended morphological filter through the top-hat transformation by RMP. We have successfully shown that the method is useful for extracting spots in biomedical images in which the conventional method is inadequate. The key concepts of our spot extraction method are the use of a straight line-segment SE and the rotation of the original image. By changing the length of SE, target spots of various sizes can be extracted. The method avoids the technical difficulties of traditional morphological processing and its performance is robust in the processing of biomedical images. The main advantages of our method are that it is computationally simple and easily modified for the extraction of target spots of different sizes and shapes, and that it can handle images in various conditions, e.g., aggregated target spots, poor SNR, and a background with large variations in intensity. The method yields directional information regarding the spatial distribution of spots within the cell as well as the frequency distribution of the size and intensity of the spots.
Our method is based on a line-segment SE with a 1-pixel width as the minimum separation distance and therefore allowed two or more target spots located close to each other to be clearly distinguished (Figure 2). With conventional morphological top-hat transformation using the common SE shape (such as a disk or square), it is difficult to separate such spots. A similar difficulty arises when the "ball" SE is used. Since it has a radius that is larger than the inter-space distance between adjacent spots, it cannot fit within the space. The top surface obtained during opening with the rolling ball cannot reach the baseline allowing for separation of the spots.
To verify the optimality of the number of rotational directions (N) shown in Figure 3, we investigated how an artifacts-contaminated image (Figure 3b) could be restored by the RMP opening with increasing N. The experimental result (Figure 3e) showed that N = 36 was a better trade-off because the value of PSNR was low for N < 36 while for N > 36 the processing time became longer. In our method, a large computational cost, which is proportional to the size of the input images, is inevitable.
We compared our spot extraction method with the conventional top-hat and h-dome transformations. As seen in Figure 5, our method outperformed the others with respect to the three criteria (Figure 5d). For the proposed method, the performance in terms of F-measure rate was maintained at 100% among all background typed images. The precision rate of the conventional top-hat transformation was much lower due to its higher FP value (124 in type-A and 228 in type-B, respectively) in detection of the noise. Furthermore, conventional top-hat transformation could not separate adjacent spots, as it used the disk SE. Meanwhile, the recall rate of the h-dome transformation was much lower due to its higher FN value (113 in type-A and 28 in type-B, respectively). Thus, the number of undetected true spots was large.
We further investigated the change in the three measurements as a function of decreasing PSNR from 17.732 to 7.797 dB (Figure 5e). The F-measure rate was maintained at 100% until PSNR decreased to about 10 dB. Subsequently, when PSNR decreased further, the F-measure rate decreased as well due to a decreasing recall rate (thus, increasing the FN value); however, the precision rate was constantly 100%. These results showed that our method is accurate in spot detection.
In the measurement of gold particles in the electron micrograph (Figure 6), the value of the averaged Feret's diameter of the extracted spots and the value of the nominal diameter of the gold particles were in close agreement. Thus, our method effectively extracted spots of the specified size with high accuracy.
Figure 7 shows the location of the small spots in the cell and the estimation of the spots intensities. Previously, Orlichenko reported that stimulation of epithelial cells with epithelial growth factor (EGF) resulted in a profound increase in the number of caveolar structures at the plasma membrane . Our method was able to carry out precise quantitative measurements of the spatial and intensity distributions of the membrane domain with respect to caveolae.
Furthermore, our method allows effective extraction of various shaped spots. Since it is based on the top-hat transformation, the spots are extracted independently in terms of the shape of the surface relief, which is based on variations in the intensity value within a spot region (Figure 8). In the conventional spot detection methods that rely on matched filtering, a 2-D Gaussian distribution is commonly used for the matched filter, assuming that a point-spread function of a signal spot has a 2-D Gaussian distribution. However, because most spots have an irregular topology, as in the example in Figure 8, accurate spot extraction is difficult using the matched filtering method.
Signal spots extracted by our method can be transformed into a 2-D Gaussian distribution as a normalization of spot shape. This allows the application of our method to the conventional automatic tracking system of individual fluorescent particles .
The RMP-based method enables a shape and intensity analysis for various types of biomedical images (2-D gel electrophoresis image, DNA microarray image, electron micrograph, X-ray mammographic image, etc.). It can be applied not only to spot extraction but also to a wide variety of important image processing techniques, such as segmentation, smoothing, and pattern extraction . Overall, it provides a wide-ranging analytical approach to biological and biomedical informatics.
We thank Dr. S. Yuasa (National Center of Neurology and Psychiatry) for helpful discussions and encouragement, Prof. T. Kodama (Osaka University) for continuing guidance and encouragement. We also express our appreciation to Prof. E. Katayama (University of Tokyo) and Dr. K. Aoyama (FEI Japan). This work was supported in part by a Health Labor Science Research Grant (Nano-001) and Grants-in-Aid for Scientific Research from the Ministry of Education Culture, Sports, Science and Technology to N. Morone.
- Shimomura O, Johnson FH, Saiga Y: Extraction, purification and properties of aequorin, a bioluminescent protein from the luminous hydromedusan, Aequorea. J Cell Comp Physiol 1962, 59: 223–239. 10.1002/jcp.1030590302View ArticlePubMedGoogle Scholar
- Paddock S: Over the rainbow: 25 years of confocal imaging. Biotechniques 2008, 44(5):643–644. 10.2144/000112798View ArticlePubMedGoogle Scholar
- Giepmans BNG, Adams SR, Ellisman MH, Tsien RY: The fluorescent toolbox for assessing protein location and function. Science 2006, 312: 217–224. 10.1126/science.1124618View ArticlePubMedGoogle Scholar
- Handley DA: The development and application of colloidal gold as a microscopic probe. In Colloidal Gold: Principles, Methods, and Applications. Volume 1. Edited by: Hayat MA. San Diego: Academic Press; 1989:1–32.View ArticleGoogle Scholar
- Hainfeld JF, Powell RD: New frontiers in gold labeling. J Histochem Cytochem 2000, 48: 471–480.View ArticlePubMedGoogle Scholar
- Peng H: Bioimage informatics: a new area of engineering biology. Bioinformatics 2008, 24: 1827–1836. 10.1093/bioinformatics/btn346View ArticlePubMedPubMed CentralGoogle Scholar
- Bozinov D, Rahnenfuhrer J: Unsupervised technique for robust target separation and analysis of DNA microarray spots through adaptive pixel clustering. Bioinformatics 2002, 18: 747–756. 10.1093/bioinformatics/18.5.747View ArticlePubMedGoogle Scholar
- Rahnenfuhrer J, Bozinov D: Hybrid clustering for microarray image analysis combining intensity and shape features. BMC Bioinformatics 2004, 5: 47. 10.1186/1471-2105-5-47View ArticlePubMedPubMed CentralGoogle Scholar
- Mete M, Hennings L, Spencer HJ, Topaloglu U: Automatic identification of angiogenesis in double stained images of liver tissue. BMC Bioinformatics 2009, 10(Suppl 11):S13. 10.1186/1471-2105-10-S11-S13View ArticlePubMedPubMed CentralGoogle Scholar
- Berth M, Moser FM, Kolbe M, Bernhardt J: The state of the art in the analysis of two-dimensional gel electrophoresis images. Appl Microbiol Biotechnol 2007, 76: 1223–1243. 10.1007/s00253-007-1128-0View ArticlePubMedPubMed CentralGoogle Scholar
- Gonzalez RC, Woods RE: Digital Image Processing. 3rd edition. New Jersey: Prentice Hall; 2008.Google Scholar
- Suri JS, Wilson DL, Laxminarayan S, Eds: Handbook of Biomedical Image Analysis: Volume 2: Segmentation Models Part B. New York: Springer-Verlag; 2005.Google Scholar
- Bankman IN, Ed: Handbook of Medical Imaging: Processing and Analysis. San Diego: Academic Press; 2000.Google Scholar
- van der Heijden F, Apperloo W, Spreeuwers LJ: Numerical optimisation in spot detector design. Pattern recognition letters 1997, 18: 1091–1097. 10.1016/S0167-8655(97)00086-XView ArticleGoogle Scholar
- Blanford RP, Tanimoto SL: Bright-spot detection in pyramids. Comput Vis Graph Image Process 1988, 43: 133–149. 10.1016/0734-189X(88)90058-8View ArticleGoogle Scholar
- Olivo-Marin J-C: Extraction of spots in biological images using multiscale products. Pattern Recognition 2002, 35: 1989–1996. 10.1016/S0031-3203(01)00127-3View ArticleGoogle Scholar
- Genovesio A, Liedl T, Emiliani V, Parak WJ, Coppey-Moisan M, Olivo-Marin J-C: Multiple particle tracking in 3-D + t microscopy: Method and application to the tracking of endocytosed quantum dots. IEEE Trans Image Process 2006, 15: 1062–1070. 10.1109/TIP.2006.872323View ArticlePubMedGoogle Scholar
- Zhang B, Fadili J, Starck J-L, Olivo-Marin J-C: Multiscale variance-stabilizing transform for mixed-Poisson-Gaussian processes and its applications in bioimaging. Proceedings of the IEEE International Conference on Image Processsing 2007, 6: VI-233-VI-236.Google Scholar
- Zeng L, Wu W: Motion Objects Detection Based on Wavelet Clustering. Proceedings of 2nd IEEE International Conference on Computer Science and Information Technology 2009, 562–566. full_textGoogle Scholar
- Bao W, Zhou R, Yang J, Yu D, Li N: Anti-aliasing lifting scheme for mechanical vibration fault feature extraction. Mech Syst Signal Process 2009, 23: 1458–1473. 10.1016/j.ymssp.2009.02.010View ArticleGoogle Scholar
- Serra J: Image analysis and mathematical morphology. London: Academic Press; 1982.Google Scholar
- Meyer F: Iterative image transformations for an automatic screening of cervical smears. J Histochem Cytochem 1979, 27: 128–135.View ArticlePubMedGoogle Scholar
- Bright DS, Steel EB: Two-dimensional top hat filter for extracting spots and spheres from digital images. J Microsc 1987, 146: 191–200.View ArticleGoogle Scholar
- Breen EJ, Joss GH, Williams KL: Locating objects of interest within biological images: The top hat box filter. J Comput-Assist Microsc 1991, 3: 97–102.Google Scholar
- Angulo J: Mathematical morphology operators for reading radioactivity DNA array images. Proceedings of the IASTED International Conference on Visualization, Imaging, and Image Processing 2004, 802–807.Google Scholar
- Janson ME, Setty TG, Paoletti A, Tran PT: Efficient formation of bipolar microtubule bundles requires microtubule-bound γ-tubulin complexes. J Cell Biol 2005, 169: 297–308. 10.1083/jcb.200410119View ArticlePubMedPubMed CentralGoogle Scholar
- Meyer F: Contrast features extraction. In Quantitative analysis of microstructures in material sciences, biology and medicine. Edited by: Chermant JL. Stuttgart: Rieder Verlag; 1977:374–380.Google Scholar
- Sternberg SR: Grayscale morphology. Comput Vis Graph Image Process 1986, 35: 333–355. 10.1016/0734-189X(86)90004-6View ArticleGoogle Scholar
- Manders EM, Hoebe R, Strackee J, Vossepoel AM, Aten JA: Largest contour segmentation: a tool for the localization of spots in confocal images. Cytometry 1996, 23: 15–21. Publisher Full Text 10.1002/(SICI)1097-0320(19960101)23:1<15::AID-CYTO3>3.0.CO;2-LView ArticlePubMedGoogle Scholar
- Giakoumis I, Nikolaidis N, Pitas I: Digital image processing techniques for the detection and removal of cracks in digitized paintings. IEEE Trans Image Process 2006, 15: 178–188. 10.1109/TIP.2005.860311View ArticlePubMedGoogle Scholar
- Yang Y, Huang S, Rao N: An automatic hybrid method for retinal blood vessel extraction. Int J Appl Math Comput Sci 2008, 18: 399–407. 10.2478/v10006-008-0036-5View ArticleGoogle Scholar
- Kapsalas P, Maravelaki-Kalaitzaki P, Zervakis M, Delegou ET, Moropoulou A: A morphological fusion algorithm for optical detection and quantification of decay patterns on stone surfaces. Constr Building Mater 2008, 22: 228–238. 10.1016/j.conbuildmat.2006.08.024View ArticleGoogle Scholar
- Kim H, Morgan DE, Zeng H, Grizzle WE, Warram JM, Stockard CR, Wang D, Zinn KR: Breast tumor xenografts: diffusion-weighted MR imaging to assess early therapy with novel apoptosis-inducing anti-DR5 antibody. Radiology 2008, 248: 844–851. 10.1148/radiol.2483071740View ArticlePubMedPubMed CentralGoogle Scholar
- Ji Q, Engel J, Craine E: Texture analysis for classification of cervix lesions. IEEE Trans Med Imaging 2000, 19: 1144–1149. 10.1109/42.896790View ArticlePubMedGoogle Scholar
- Stevenson RL, Arce GR: Morphological Filters: Statistical and Further Syntactic Properties. IEEE Trans Circuits and Systems 1987, 34: 1292–1305. 10.1109/TCS.1987.1086067View ArticleGoogle Scholar
- Song J, Delp EJ: The analysis of morphological filters with multiple structuring elements. Comput Vis Graph Image Process 1990, 50: 308–328. 10.1016/0734-189X(90)90150-TView ArticleGoogle Scholar
- Smal I, Loog M, Niessen W, Meijering E: Quantitative comparison of spot detection methods in fluorescence microscopy. IEEE Trans Med Imaging 2010, 29: 282–301. 10.1109/TMI.2009.2025127View ArticlePubMedGoogle Scholar
- Vincent L: Morphological grayscale reconstruction in image analysis: Applications and efficient algorithms. IEEE Trans Image Process 1993, 2: 176–201. 10.1109/83.217222View ArticlePubMedGoogle Scholar
- Jiang S, Zhou X, Kirchhausen T, Wong STC: Detection of molecular particles in live cells via machine learning. Cytometry A 2007, 71: 563–575.View ArticlePubMedGoogle Scholar
- Soille P: Morphological Image Analysis: Principles and Applications. Berlin: Springer-Verlag; 1999.View ArticleGoogle Scholar
- Lacoste TD, Michalet X, Pinaud F, Chemla DS, Alivisatos AP, Weiss S: Ultrahigh-resolution multicolor colocalization of single fluorescent probes. Proc Natl Acad Sci USA 2000, 97: 9461–9466. 10.1073/pnas.170286097View ArticlePubMedPubMed CentralGoogle Scholar
- Smal I, Draegestein K, Galjart N, Niessen W, Meijering E: Particle filtering for multiple object tracking in dynamic fluorescence microscopy images: application to microtubule growth analysis. IEEE Trans Med Imaging 2008, 27: 789–804. 10.1109/TMI.2008.916964View ArticlePubMedGoogle Scholar
- Orlichenko L, Huang B, Krueger E, McNiven MA: Epithelial growth factor-induced phosphorylation of caveolin 1 at tyrosine 14 stimulates caveolae formation in epithelial cells. J Biol Chem 2006, 281: 4570–4579. 10.1074/jbc.M512088200View ArticlePubMedGoogle Scholar
- Cheezum MK, Walker WF, Guilford WH: Quantitative comparison of algorithms for tracking single fluorescent particles. Biophys J 2001, 81: 2378–2388. 10.1016/S0006-3495(01)75884-5View ArticlePubMedPubMed CentralGoogle Scholar
- Kimori Y, Oguchi Y, Ichise N, Baba N, Katayama E: A procedure to analyze surface profiles of the protein molecules visualized by quick-freeze deep-etch replica electron microscopy. Ultramicroscopy 2007, 107: 25–39. 10.1016/j.ultramic.2006.04.012View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.