Limits...
Direct imaging of phase objects enables conventional deconvolution in bright field light microscopy.

Hernández Candia CN, Gutiérrez-Medina B - PLoS ONE (2014)

Bottom Line: Polystyrene nanoparticles and microtubules (biological polymer filaments) serve as the pure phase point and line objects, respectively, that are imaged with high contrast and low noise using standard microscopy plus digital image processing.Our experimental results agree with a proposed model for the response functions, and confirm previous theoretical predictions.Finally, we use the measured phase point-spread function to apply conventional deconvolution on the bright field images of living, unstained bacteria, resulting in improved definition of cell boundaries and sub-cellular features.

View Article: PubMed Central - PubMed

Affiliation: Program in Molecular Biology, Instituto Potosino de Investigación Científica y Tecnológica, San Luis Potosí, Mexico.

ABSTRACT
In transmitted optical microscopy, absorption structure and phase structure of the specimen determine the three-dimensional intensity distribution of the image. The elementary impulse responses of the bright field microscope therefore consist of separate absorptive and phase components, precluding general application of linear, conventional deconvolution processing methods to improve image contrast and resolution. However, conventional deconvolution can be applied in the case of pure phase (or pure absorptive) objects if the corresponding phase (or absorptive) impulse responses of the microscope are known. In this work, we present direct measurements of the phase point- and line-spread functions of a high-aperture microscope operating in transmitted bright field. Polystyrene nanoparticles and microtubules (biological polymer filaments) serve as the pure phase point and line objects, respectively, that are imaged with high contrast and low noise using standard microscopy plus digital image processing. Our experimental results agree with a proposed model for the response functions, and confirm previous theoretical predictions. Finally, we use the measured phase point-spread function to apply conventional deconvolution on the bright field images of living, unstained bacteria, resulting in improved definition of cell boundaries and sub-cellular features. These developments demonstrate practical application of standard restoration methods to improve imaging of phase objects such as cells in transmitted light microscopy.

Show MeSH

Related in: MedlinePlus

Demonstration of deconvolution in the BF microscopy images of unstained, living E. coli cells.(A–C) BF images of bacteria before (“BF”) and after (“D”) deconvolution, together with their respective intensity profiles along the yellow dashed lines. Length of double-arrow lines in (B): m. (D) Image of a standing bacterium along three orthogonal slices (marked by the yellow dashed lines). In BF, the cell extends well beyond the position of the supporting coverslip (white dotted line), whereas the same views after deconvolution display the cell with improved definition of boundaries. (E) Time-lapse frames of a bacterium undergoing cell division under continuous illumination. No threshold or transparency levels were applied to deconvolved images. Scale bars: m.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3928359&req=5

pone-0089106-g005: Demonstration of deconvolution in the BF microscopy images of unstained, living E. coli cells.(A–C) BF images of bacteria before (“BF”) and after (“D”) deconvolution, together with their respective intensity profiles along the yellow dashed lines. Length of double-arrow lines in (B): m. (D) Image of a standing bacterium along three orthogonal slices (marked by the yellow dashed lines). In BF, the cell extends well beyond the position of the supporting coverslip (white dotted line), whereas the same views after deconvolution display the cell with improved definition of boundaries. (E) Time-lapse frames of a bacterium undergoing cell division under continuous illumination. No threshold or transparency levels were applied to deconvolved images. Scale bars: m.

Mentions: We next performed deconvolution of cell frames. One notoriously adverse aspect of BF images is that they present variations between negative and positive intensity values due to the strong influence of defocusing and out-of-focus scattered light, making difficult to identify object locations and boundaries. In contrast, the deconvolved frames show significant improvement in clarity (see Figures 5A–C), where the ambiguity of regions changing in intensity from positive to negative values is removed. In particular, cell wall boundaries become well defined, displaying a striking resemblance to fluorescence microscopy images of labelled cells [30]. Likewise, intensity variations along the cell body that are only hinted in the original BF frames gain contrast after deconvolution. Some of these variations (see Figure 5B) show a degree of spatial periodicity (∼0.5 m) and may correspond to the same structures observed recently in unstained bacteria using dSLIT, a microscopy technique capable of quantitative phase imaging [31]. Here, sensitivity to phase variations caused by the object is expected, as evidenced by Eq. (4). Image improvement is also observed in z-stacks, as demonstrated by performing deconvolution on the images of a bacterium whose orientation is standing above the coverslip (see Figure 5D). In the original BF images, the location and boundaries of the bacterium are difficult to distinguish, and the bacterium body appears to extend well into the supporting coverslip. These problems are much reduced after deconvolution, where cell orientation and boundaries are better defined.


Direct imaging of phase objects enables conventional deconvolution in bright field light microscopy.

Hernández Candia CN, Gutiérrez-Medina B - PLoS ONE (2014)

Demonstration of deconvolution in the BF microscopy images of unstained, living E. coli cells.(A–C) BF images of bacteria before (“BF”) and after (“D”) deconvolution, together with their respective intensity profiles along the yellow dashed lines. Length of double-arrow lines in (B): m. (D) Image of a standing bacterium along three orthogonal slices (marked by the yellow dashed lines). In BF, the cell extends well beyond the position of the supporting coverslip (white dotted line), whereas the same views after deconvolution display the cell with improved definition of boundaries. (E) Time-lapse frames of a bacterium undergoing cell division under continuous illumination. No threshold or transparency levels were applied to deconvolved images. Scale bars: m.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3928359&req=5

pone-0089106-g005: Demonstration of deconvolution in the BF microscopy images of unstained, living E. coli cells.(A–C) BF images of bacteria before (“BF”) and after (“D”) deconvolution, together with their respective intensity profiles along the yellow dashed lines. Length of double-arrow lines in (B): m. (D) Image of a standing bacterium along three orthogonal slices (marked by the yellow dashed lines). In BF, the cell extends well beyond the position of the supporting coverslip (white dotted line), whereas the same views after deconvolution display the cell with improved definition of boundaries. (E) Time-lapse frames of a bacterium undergoing cell division under continuous illumination. No threshold or transparency levels were applied to deconvolved images. Scale bars: m.
Mentions: We next performed deconvolution of cell frames. One notoriously adverse aspect of BF images is that they present variations between negative and positive intensity values due to the strong influence of defocusing and out-of-focus scattered light, making difficult to identify object locations and boundaries. In contrast, the deconvolved frames show significant improvement in clarity (see Figures 5A–C), where the ambiguity of regions changing in intensity from positive to negative values is removed. In particular, cell wall boundaries become well defined, displaying a striking resemblance to fluorescence microscopy images of labelled cells [30]. Likewise, intensity variations along the cell body that are only hinted in the original BF frames gain contrast after deconvolution. Some of these variations (see Figure 5B) show a degree of spatial periodicity (∼0.5 m) and may correspond to the same structures observed recently in unstained bacteria using dSLIT, a microscopy technique capable of quantitative phase imaging [31]. Here, sensitivity to phase variations caused by the object is expected, as evidenced by Eq. (4). Image improvement is also observed in z-stacks, as demonstrated by performing deconvolution on the images of a bacterium whose orientation is standing above the coverslip (see Figure 5D). In the original BF images, the location and boundaries of the bacterium are difficult to distinguish, and the bacterium body appears to extend well into the supporting coverslip. These problems are much reduced after deconvolution, where cell orientation and boundaries are better defined.

Bottom Line: Polystyrene nanoparticles and microtubules (biological polymer filaments) serve as the pure phase point and line objects, respectively, that are imaged with high contrast and low noise using standard microscopy plus digital image processing.Our experimental results agree with a proposed model for the response functions, and confirm previous theoretical predictions.Finally, we use the measured phase point-spread function to apply conventional deconvolution on the bright field images of living, unstained bacteria, resulting in improved definition of cell boundaries and sub-cellular features.

View Article: PubMed Central - PubMed

Affiliation: Program in Molecular Biology, Instituto Potosino de Investigación Científica y Tecnológica, San Luis Potosí, Mexico.

ABSTRACT
In transmitted optical microscopy, absorption structure and phase structure of the specimen determine the three-dimensional intensity distribution of the image. The elementary impulse responses of the bright field microscope therefore consist of separate absorptive and phase components, precluding general application of linear, conventional deconvolution processing methods to improve image contrast and resolution. However, conventional deconvolution can be applied in the case of pure phase (or pure absorptive) objects if the corresponding phase (or absorptive) impulse responses of the microscope are known. In this work, we present direct measurements of the phase point- and line-spread functions of a high-aperture microscope operating in transmitted bright field. Polystyrene nanoparticles and microtubules (biological polymer filaments) serve as the pure phase point and line objects, respectively, that are imaged with high contrast and low noise using standard microscopy plus digital image processing. Our experimental results agree with a proposed model for the response functions, and confirm previous theoretical predictions. Finally, we use the measured phase point-spread function to apply conventional deconvolution on the bright field images of living, unstained bacteria, resulting in improved definition of cell boundaries and sub-cellular features. These developments demonstrate practical application of standard restoration methods to improve imaging of phase objects such as cells in transmitted light microscopy.

Show MeSH
Related in: MedlinePlus