Limits...
Cell segmentation by multi-resolution analysis and maximum likelihood estimation (MAMLE).

Chowdhury S, Kandhavelu M, Yli-Harja O, Ribeiro AS - BMC Bioinformatics (2013)

Bottom Line: From this result, a correction procedure is applied that exploits maximum likelihood estimate as an objective function.The new method attained very high (above 90%) cell segmentation accuracy in all cases.Finally, its accuracy was compared to several existing methods, and in all tests, MAMLE outperformed them in segmentation accuracy.

View Article: PubMed Central - HTML - PubMed

ABSTRACT

Background: Cell imaging is becoming an indispensable tool for cell and molecular biology research. However, most processes studied are stochastic in nature, and require the observation of many cells and events. Ideally, extraction of information from these images ought to rely on automatic methods. Here, we propose a novel segmentation method, MAMLE, for detecting cells within dense clusters.

Methods: MAMLE executes cell segmentation in two stages. The first relies on state of the art filtering technique, edge detection in multi-resolution with morphological operator and threshold decomposition for adaptive thresholding. From this result, a correction procedure is applied that exploits maximum likelihood estimate as an objective function. Also, it acquires morphological features from the initial segmentation for constructing the likelihood parameter, after which the final segmentation is obtained.

Conclusions: We performed an empirical evaluation that includes sample images from different imaging modalities and diverse cell types. The new method attained very high (above 90%) cell segmentation accuracy in all cases. Finally, its accuracy was compared to several existing methods, and in all tests, MAMLE outperformed them in segmentation accuracy.

Show MeSH

Related in: MedlinePlus

Comparison between the proposed method and Schnitzcells software. (a) Fluorescent protein labelled Staphylococcus cells in Epifluorescence microscopy image, (b) Segmented result of (a) by the proposed method, (c) Segmented result of (a) by Schnitzcells software, (d) Fluorescent protein labelled E. coli cells captured with confocal microscope (Source [9]), (e) Segmented result of (d) by the proposed method, (f) Segmented result of (d) by Schnitzcells software, (g)) Fluorescent protein labelled E. coli cells captured with Epifluorescence microscope, (h) Segmented result of (g) by the proposed method, (i) Segmented result of (g) by Schnitzcells software, (j) E. coli cells captured with phase contrast microscope (Source [12]), (k) Segmented result of (j) by the proposed method, (l) Segmented result of (j) by Schnitzcells software.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3750476&req=5

Figure 5: Comparison between the proposed method and Schnitzcells software. (a) Fluorescent protein labelled Staphylococcus cells in Epifluorescence microscopy image, (b) Segmented result of (a) by the proposed method, (c) Segmented result of (a) by Schnitzcells software, (d) Fluorescent protein labelled E. coli cells captured with confocal microscope (Source [9]), (e) Segmented result of (d) by the proposed method, (f) Segmented result of (d) by Schnitzcells software, (g)) Fluorescent protein labelled E. coli cells captured with Epifluorescence microscope, (h) Segmented result of (g) by the proposed method, (i) Segmented result of (g) by Schnitzcells software, (j) E. coli cells captured with phase contrast microscope (Source [12]), (k) Segmented result of (j) by the proposed method, (l) Segmented result of (j) by Schnitzcells software.

Mentions: The cell segmentation accuracy of the algorithm is next compared with three state-of-the-art cell image analysis platforms, namely, Cellprofiler [8], Farsight [10], and Schnitzcells [9]. For the comparison to be unbiased, test samples were included from publicly available online repositories [8-10]. A set of sample results is shown in Figure 4. In general, we found the method proposed here to outperform the others in segmentation accuracy. Schnitzcells was the second best in E. coli segmentation (Figure 5). To further compare the proposed method and Schnitzcells we extended the evaluation. This additional test is carried out using publicly available bench mark images for cell counting [8]. The benchmark data contains roughly 2162 human HT29 colon cancer cells in 6 images. The cells were manually labelled and scored by two human observers and the average of the manual score is considered the ground truth. The human labelling had a mean absolute deviation of 11% and the best know result for this data set was attained by Cellprofiler, with a mean absolute deviation of only 6.2% [8]. In this benchmark data, our method exhibited a mean absolute deviation of only 1.79%.


Cell segmentation by multi-resolution analysis and maximum likelihood estimation (MAMLE).

Chowdhury S, Kandhavelu M, Yli-Harja O, Ribeiro AS - BMC Bioinformatics (2013)

Comparison between the proposed method and Schnitzcells software. (a) Fluorescent protein labelled Staphylococcus cells in Epifluorescence microscopy image, (b) Segmented result of (a) by the proposed method, (c) Segmented result of (a) by Schnitzcells software, (d) Fluorescent protein labelled E. coli cells captured with confocal microscope (Source [9]), (e) Segmented result of (d) by the proposed method, (f) Segmented result of (d) by Schnitzcells software, (g)) Fluorescent protein labelled E. coli cells captured with Epifluorescence microscope, (h) Segmented result of (g) by the proposed method, (i) Segmented result of (g) by Schnitzcells software, (j) E. coli cells captured with phase contrast microscope (Source [12]), (k) Segmented result of (j) by the proposed method, (l) Segmented result of (j) by Schnitzcells software.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3750476&req=5

Figure 5: Comparison between the proposed method and Schnitzcells software. (a) Fluorescent protein labelled Staphylococcus cells in Epifluorescence microscopy image, (b) Segmented result of (a) by the proposed method, (c) Segmented result of (a) by Schnitzcells software, (d) Fluorescent protein labelled E. coli cells captured with confocal microscope (Source [9]), (e) Segmented result of (d) by the proposed method, (f) Segmented result of (d) by Schnitzcells software, (g)) Fluorescent protein labelled E. coli cells captured with Epifluorescence microscope, (h) Segmented result of (g) by the proposed method, (i) Segmented result of (g) by Schnitzcells software, (j) E. coli cells captured with phase contrast microscope (Source [12]), (k) Segmented result of (j) by the proposed method, (l) Segmented result of (j) by Schnitzcells software.
Mentions: The cell segmentation accuracy of the algorithm is next compared with three state-of-the-art cell image analysis platforms, namely, Cellprofiler [8], Farsight [10], and Schnitzcells [9]. For the comparison to be unbiased, test samples were included from publicly available online repositories [8-10]. A set of sample results is shown in Figure 4. In general, we found the method proposed here to outperform the others in segmentation accuracy. Schnitzcells was the second best in E. coli segmentation (Figure 5). To further compare the proposed method and Schnitzcells we extended the evaluation. This additional test is carried out using publicly available bench mark images for cell counting [8]. The benchmark data contains roughly 2162 human HT29 colon cancer cells in 6 images. The cells were manually labelled and scored by two human observers and the average of the manual score is considered the ground truth. The human labelling had a mean absolute deviation of 11% and the best know result for this data set was attained by Cellprofiler, with a mean absolute deviation of only 6.2% [8]. In this benchmark data, our method exhibited a mean absolute deviation of only 1.79%.

Bottom Line: From this result, a correction procedure is applied that exploits maximum likelihood estimate as an objective function.The new method attained very high (above 90%) cell segmentation accuracy in all cases.Finally, its accuracy was compared to several existing methods, and in all tests, MAMLE outperformed them in segmentation accuracy.

View Article: PubMed Central - HTML - PubMed

ABSTRACT

Background: Cell imaging is becoming an indispensable tool for cell and molecular biology research. However, most processes studied are stochastic in nature, and require the observation of many cells and events. Ideally, extraction of information from these images ought to rely on automatic methods. Here, we propose a novel segmentation method, MAMLE, for detecting cells within dense clusters.

Methods: MAMLE executes cell segmentation in two stages. The first relies on state of the art filtering technique, edge detection in multi-resolution with morphological operator and threshold decomposition for adaptive thresholding. From this result, a correction procedure is applied that exploits maximum likelihood estimate as an objective function. Also, it acquires morphological features from the initial segmentation for constructing the likelihood parameter, after which the final segmentation is obtained.

Conclusions: We performed an empirical evaluation that includes sample images from different imaging modalities and diverse cell types. The new method attained very high (above 90%) cell segmentation accuracy in all cases. Finally, its accuracy was compared to several existing methods, and in all tests, MAMLE outperformed them in segmentation accuracy.

Show MeSH
Related in: MedlinePlus