Sriram Krishnan

Classification of Mycobacterium tuberculosis in Images of ZN-Stained Sputum Smears

Screening for tuberculosis (TB) in low- and middle-income countries is centered on the microscope. We present methods for the automated identification of Mycobacterium tuberculosis in images of Ziehl-Neelsen (ZN) stained sputum smears obtained using a bright-field microscope. We segment candidate bacillus objects using a combination of two-class pixel classifiers. The algorithm produces results that agree well with manual segmentations, as judged by the Hausdorff distance and the modified Williams index. The extraction of geometric-transformation-invariant features and optimization of the feature set by feature subset selection and Fisher transformation follow. Finally, different two-class object classifiers are compared. The sensitivity and specificity of all tested classifiers is above 95% for the identification of bacillus objects represented by Fisher-transformed features. Our results may be used to reduce technician involvement in screening for TB, and would be particularly useful in laboratories in countries with a high burden of TB, where, typically, ZN rather than auramine staining of sputum smears is the method of choice.

Automated focusing in bright-field microscopy for tuberculosis detection

Automated microscopy to detect Mycobacterium tuberculosis in sputum smear slides would enable laboratories in countries with a high tuberculosis burden to cope efficiently with large numbers of smears. Focusing is a core component of automated microscopy, and successful autofocusing depends on selection of an appropriate focus algorithm for a specific task. We examined autofocusing algorithms for bright‐field microscopy of Ziehl–Neelsen stained sputum smears. Six focus measures, defined in the spatial domain, were examined with respect to accuracy, execution time, range, full width at half maximum of the peak and the presence of local maxima. Curve fitting around an estimate of the focal plane was found to produce good results and is therefore an acceptable strategy to reduce the number of images captured for focusing and the processing time. Vollaths F4 measure performed best for full z‐stacks, with a mean difference of 0.27 μm between manually and automatically determined focal positions, whereas it is jointly ranked best with the Brenner gradient for curve fitting.

Automated detection of tuberculosis in Ziehl-Neelsen-stained sputum smears using two one-class classifiers.

Screening for tuberculosis in high‐prevalence countries relies on sputum smear microscopy. We present a method for the automated identification of Mycobacterium tuberculosis in images of Ziehl‐Neelsen‐stained sputum smears obtained using a bright‐field microscope. We use two stages of classification. The first comprises a one‐class pixel classifier for object segmentation. Geometric transformation invariant features are extracted for implementation of the second stage, namely one‐class object classification. Different classifiers are compared; the sensitivity of all tested classifiers is above 90% for the identification of a single bacillus object using all extracted features. The mixture of Gaussians classifier performed well in both stages of classification. This method may be used as a step in the automation of tuberculosis screening, in order to reduce technician involvement in the process.

Improved red blood cell counting in thin blood smears

Quantification of the extent of malaria parasite infection (parasitaemia) continues to rely on time-consuming manual microscopy of Giemsa-stained blood smears. We present an algorithm that counts red blood cells in thin blood smear images, the first step in the determination of malaria parasitaemia. Morphological methods and iterative thresholding are used for red blood cell segmentation, and boundary curvature calculations and Delaunay triangulation for red blood cell clump splitting. Our results compare well with those of published semi-automated methods, with an absolute error of 2.8% between manual and automatic counting of red blood cells.

Detection of tuberculosis in sputum smear images using two one-class classifiers

We present a method for the identification of Mycobacterium tuberculosis in images of Ziehl-Neelsen stained sputum smears obtained using a bright field microscope. We use two stages of classification; the first is a one-class pixel classifier, after which geometric transformation invariant features are extracted. The second stage is a one-class object classifier. Different classifiers are compared; the sensitivity of all tested classifiers is above 90% for the identification of a single bacillus object using all extracted features. Our results may be used to reduce technician involvement in screening for tuberculosis, and will be particularly useful in laboratories in countries with a high burden of tuberculosis.

Image fusion for autofocusing in fluorescence microscopy for tuberculosis screening

Automatic microscopy for screening of sputum smears for tuberculosis would reduce the reliance on technicians in heavily burdened laboratories in poorly-resourced countries. Autofocusing is a key component of automated microscopy. We investigate the use of wavelet-based image fusion for automatic focusing of sputum smear slides as a component of automated fluorescence microscopy to identify Mycobacterium tuberculosis. We use manual focusing as ground truth for assessing the performance of the algorithm. Image fusion produces images with a degree of focus comparable to that of a human operator.

Comparison of image fusion and focus function-based techniques for autofocusing in fluorescence microscopy for tuberculosis screening

Manual screening of sputum smear slides for tuberculosis using microscopy is a time consuming exercise that strains laboratory resources in regions with high prevalence of the disease. A system to automatically screen smears would benefit such areas by reducing the reliance on technicians. The first step of such a system would be to ensure that the slide is optimally focused. In this paper, we compare the use of a focus measure and wavelet-based image fusion for automatic focusing of sputum smear slides for fluorescence microscopy to detect tuberculosis. Our objective is to obtain the sharpest image for input into further processing stages in the identification of Mycobacterium tuberculosis. We use manual focusing as ground truth and assess the performance of the two methods by comparing segmented bacillus borders in the autofocused images with those in the manually focused images. Image fusion-based focusing performs marginally better than focus measure-based focusing.