Limits...
Empirical comparison of color normalization methods for epithelial-stromal classification in H and E images.

Sethi A, Sha L, Vahadane AR, Deaton RJ, Kumar N, Macias V, Gann PH - J Pathol Inform (2016)

Bottom Line: For the main MRS method, which relied on classification of super-pixels, the number of variables used was reduced using backward elimination without compromising accuracy, and test - area under the curves (AUCs) were compared for original and normalized images.Khan method reduced color saturation while Vahadane reduced hue variance.Color normalization can give a small incremental benefit when a super-pixel-based classification method is used with features that perform implicit color normalization while the gain is higher for patch-based classification methods for classifying epithelium versus stroma.

View Article: PubMed Central - PubMed

Affiliation: Department of Electronics and Electrical Engineering, Indian Institute of Technology Guwahati, Guwahati, Assam, India; Department of Pathology, University of Illinois, Chicago, IL, USA.

ABSTRACT

Context: Color normalization techniques for histology have not been empirically tested for their utility for computational pathology pipelines.

Aims: We compared two contemporary techniques for achieving a common intermediate goal - epithelial-stromal classification.

Settings and design: Expert-annotated regions of epithelium and stroma were treated as ground truth for comparing classifiers on original and color-normalized images.

Materials and methods: Epithelial and stromal regions were annotated on thirty diverse-appearing H and E stained prostate cancer tissue microarray cores. Corresponding sets of thirty images each were generated using the two color normalization techniques. Color metrics were compared for original and color-normalized images. Separate epithelial-stromal classifiers were trained and compared on test images. Main analyses were conducted using a multiresolution segmentation (MRS) approach; comparative analyses using two other classification approaches (convolutional neural network [CNN], Wndchrm) were also performed.

Statistical analysis: For the main MRS method, which relied on classification of super-pixels, the number of variables used was reduced using backward elimination without compromising accuracy, and test - area under the curves (AUCs) were compared for original and normalized images. For CNN and Wndchrm, pixel classification test-AUCs were compared.

Results: Khan method reduced color saturation while Vahadane reduced hue variance. Super-pixel-level test-AUC for MRS was 0.010-0.025 (95% confidence interval limits ± 0.004) higher for the two normalized image sets compared to the original in the 10-80 variable range. Improvement in pixel classification accuracy was also observed for CNN and Wndchrm for color-normalized images.

Conclusions: Color normalization can give a small incremental benefit when a super-pixel-based classification method is used with features that perform implicit color normalization while the gain is higher for patch-based classification methods for classifying epithelium versus stroma.

No MeSH data available.


Related in: MedlinePlus

Receiver operating characteristic curves for pixel-level accuracy for twenty-feature models for ten test images for the three models for thresholds 0.75, 0.5, 0.25, 0.15, and 0.1 on the logistic regression output
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4837797&req=5

Figure 8: Receiver operating characteristic curves for pixel-level accuracy for twenty-feature models for ten test images for the three models for thresholds 0.75, 0.5, 0.25, 0.15, and 0.1 on the logistic regression output

Mentions: Although the performance of models trained on the three sets of images was similar beyond eighty variables, computation of class labels for super-pixels becomes slower and thus less efficient with a large number of variables. Therefore, for pixel-level classification, models with twenty variables, each was selected for the three image sets. Classifier performance on the two sets of normalized images was nearly constant for twenty or more variables while no such “elbow” was evident for original images. Pixel classification sensitivity and specificity on the ten test images for different thresholds are shown in Figure 8 for the three sets, which form partial ROC curves. Relative performance at pixel-level was similar to super-pixel-level performance in that Vahadane and Khan were very close and performed marginally better than the original set of images.


Empirical comparison of color normalization methods for epithelial-stromal classification in H and E images.

Sethi A, Sha L, Vahadane AR, Deaton RJ, Kumar N, Macias V, Gann PH - J Pathol Inform (2016)

Receiver operating characteristic curves for pixel-level accuracy for twenty-feature models for ten test images for the three models for thresholds 0.75, 0.5, 0.25, 0.15, and 0.1 on the logistic regression output
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4837797&req=5

Figure 8: Receiver operating characteristic curves for pixel-level accuracy for twenty-feature models for ten test images for the three models for thresholds 0.75, 0.5, 0.25, 0.15, and 0.1 on the logistic regression output
Mentions: Although the performance of models trained on the three sets of images was similar beyond eighty variables, computation of class labels for super-pixels becomes slower and thus less efficient with a large number of variables. Therefore, for pixel-level classification, models with twenty variables, each was selected for the three image sets. Classifier performance on the two sets of normalized images was nearly constant for twenty or more variables while no such “elbow” was evident for original images. Pixel classification sensitivity and specificity on the ten test images for different thresholds are shown in Figure 8 for the three sets, which form partial ROC curves. Relative performance at pixel-level was similar to super-pixel-level performance in that Vahadane and Khan were very close and performed marginally better than the original set of images.

Bottom Line: For the main MRS method, which relied on classification of super-pixels, the number of variables used was reduced using backward elimination without compromising accuracy, and test - area under the curves (AUCs) were compared for original and normalized images.Khan method reduced color saturation while Vahadane reduced hue variance.Color normalization can give a small incremental benefit when a super-pixel-based classification method is used with features that perform implicit color normalization while the gain is higher for patch-based classification methods for classifying epithelium versus stroma.

View Article: PubMed Central - PubMed

Affiliation: Department of Electronics and Electrical Engineering, Indian Institute of Technology Guwahati, Guwahati, Assam, India; Department of Pathology, University of Illinois, Chicago, IL, USA.

ABSTRACT

Context: Color normalization techniques for histology have not been empirically tested for their utility for computational pathology pipelines.

Aims: We compared two contemporary techniques for achieving a common intermediate goal - epithelial-stromal classification.

Settings and design: Expert-annotated regions of epithelium and stroma were treated as ground truth for comparing classifiers on original and color-normalized images.

Materials and methods: Epithelial and stromal regions were annotated on thirty diverse-appearing H and E stained prostate cancer tissue microarray cores. Corresponding sets of thirty images each were generated using the two color normalization techniques. Color metrics were compared for original and color-normalized images. Separate epithelial-stromal classifiers were trained and compared on test images. Main analyses were conducted using a multiresolution segmentation (MRS) approach; comparative analyses using two other classification approaches (convolutional neural network [CNN], Wndchrm) were also performed.

Statistical analysis: For the main MRS method, which relied on classification of super-pixels, the number of variables used was reduced using backward elimination without compromising accuracy, and test - area under the curves (AUCs) were compared for original and normalized images. For CNN and Wndchrm, pixel classification test-AUCs were compared.

Results: Khan method reduced color saturation while Vahadane reduced hue variance. Super-pixel-level test-AUC for MRS was 0.010-0.025 (95% confidence interval limits ± 0.004) higher for the two normalized image sets compared to the original in the 10-80 variable range. Improvement in pixel classification accuracy was also observed for CNN and Wndchrm for color-normalized images.

Conclusions: Color normalization can give a small incremental benefit when a super-pixel-based classification method is used with features that perform implicit color normalization while the gain is higher for patch-based classification methods for classifying epithelium versus stroma.

No MeSH data available.


Related in: MedlinePlus