Hongming Xu

An efficient technique for nuclei segmentation based on ellipse descriptor analysis and improved seed detection algorithm.

In this paper, we propose an efficient method for segmenting cell nuclei in the skin histopathological images. The proposed technique consists of four modules. First, it separates the nuclei regions from the background with an adaptive threshold technique. Next, an elliptical descriptor is used to detect the isolated nuclei with elliptical shapes. This descriptor classifies the nuclei regions based on two ellipticity parameters. Nuclei clumps and nuclei with irregular shapes are then localized by an improved seed detection technique based on voting in the eroded nuclei regions. Finally, undivided nuclei regions are segmented by a marked watershed algorithm. Experimental results on 114 different image patches indicate that the proposed technique provides a superior performance in nuclei detection and segmentation.

Automatic Nuclei Detection Based on Generalized Laplacian of Gaussian Filters

Efficient and accurate detection of cell nuclei is an important step toward automatic analysis in histopathology. In this work, we present an automatic technique based on generalized Laplacian of Gaussian (gLoG) filter for nuclei detection in digitized histological images. The proposed technique first generates a bank of gLoG kernels with different scales and orientations and then performs convolution between directional gLoG kernels and the candidate image to obtain a set of response maps. The local maxima of response maps are detected and clustered into different groups by mean-shift algorithm based on their geometrical closeness. The point which has the maximum response in each group is finally selected as the nucleus seed. Experimental results on two datasets show that the proposed technique provides a superior performance in nuclei detection compared to existing techniques.

MULTI-PASS ADAPTIVE VOTING FOR NUCLEI DETECTION IN HISTOPATHOLOGICAL IMAGES

Nuclei detection is often a critical initial step in the development of computer aided diagnosis and prognosis schemes in the context of digital pathology images. While over the last few years, a number of nuclei detection methods have been proposed, most of these approaches make idealistic assumptions about the staining quality of the tissue. In this paper, we present a new Multi-Pass Adaptive Voting (MPAV) for nuclei detection which is specifically geared towards images with poor quality staining and noise on account of tissue preparation artifacts. The MPAV utilizes the symmetric property of nuclear boundary and adaptively selects gradient from edge fragments to perform voting for a potential nucleus location. The MPAV was evaluated in three cohorts with different staining methods: Hematoxylin & Eosin, CD31 & Hematoxylin, and Ki-67 and where most of the nuclei were unevenly and imprecisely stained. Across a total of 47 images and nearly 17,700 manually labeled nuclei serving as the ground truth, MPAV was able to achieve a superior performance, with an area under the precision-recall curve (AUC) of 0.73. Additionally, MPAV also outperformed three state-of-the-art nuclei detection methods, a single pass voting method, a multi-pass voting method, and a deep learning based method.

Automated segmentation of regions of interest in whole slide skin histopathological images

In the diagnosis of skin melanoma by analyzing histopathological images, the epidermis and epidermis-dermis junctional areas are regions of interest as they provide the most important histologic diagnosis features. This paper presents an automated technique for segmenting epidermis and dermis regions from whole slide skin histopathological images. The proposed technique first performs epidermis segmentation using a thresholding and thickness measurement based method. The dermis area is then segmented based on a predefined depth of segmentation from the epidermis outer boundary. Experimental results on 66 different skin images show that the proposed technique can robustly segment regions of interest as desired.

Automatic Nuclear Segmentation Using Multiscale Radial Line Scanning With Dynamic Programming

In the diagnosis of various cancers by analyzing histological images, automatic nuclear segmentation is an important step. However, nuclear segmentation is a difficult problem because of overlapping nuclei, inhomogeneous staining, and presence of noisy pixels and other tissue components. In this paper, we present an automatic technique for nuclear segmentation in skin histological images. The proposed technique first applies a bank of generalized Laplacian of Gaussian kernels to detect nuclear seeds. Based on the detected nuclear seeds, a multiscale radial line scanning method combined with dynamic programming is applied to extract a set of candidate nuclear boundaries. The gradient, intensity, and shape information are then integrated to determine the optimal boundary for each nucleus in the image. Nuclear overlap limitation is finally imposed based on a Dice coefficient measure such that the obtained nuclear contours do not severely intersect with each other. Experiments have been thoroughly performed on two datasets with H&E and Ki-67 stained images, which show that the proposed technique is superior to conventional schemes of nuclear segmentation.

Efficient segmentation of skin epidermis in whole slide histopathological images.

Segmentation of epidermis areas is an important step towards automatic analysis of skin histopathological images. This paper presents a robust technique for epidermis segmentation in whole slide skin histopathological images. The proposed technique first performs a coarse epidermis segmentation using global thresholding and shape analysis. The epidermis thickness is then estimated by a series of line segments perpendicular to the main axis of the initially segmented epidermis mask. If the segmented epidermis mask has a thickness greater than a predefined threshold, the segmentation is suspected to be inaccurate. A second pass of fine segmentation using k-means algorithm is then carried out over these coarsely segmented result to enhance the performance. Experimental results on 64 different skin histopathological images show that the proposed technique provides a superior performance compared to the existing techniques.

Automated analysis and classification of melanocytic tumor on skin whole slide images

This paper presents a computer-aided technique for automated analysis and classification of melanocytic tumor on skin whole slide biopsy images. The proposed technique consists of four main modules. First, skin epidermis and dermis regions are segmented by a multi-resolution framework. Next, epidermis analysis is performed, where a set of epidermis features reflecting nuclear morphologies and spatial distributions is computed. In parallel with epidermis analysis, dermis analysis is also performed, where dermal cell nuclei are segmented and a set of textural and cytological features are computed. Finally, the skin melanocytic image is classified into different categories such as melanoma, nevus or normal tissue by using a multi-class support vector machine (mSVM) with extracted epidermis and dermis features. Experimental results on 66 skin whole slide images indicate that the proposed technique achieves more than 95% classification accuracy, which suggests that the technique has the potential to be used for assisting pathologists on skin biopsy image analysis and classification.

Automated detection of cavernous malformations in brain MRI images

Cavernous malformation or cavernoma is a kind of brain vessel abnormality that can cause serious symptoms such as seizures, intracerebral hemorrhage and various neurological deficits. It is one of the most common epileptogenic lesions that can be identified by physicians based on magnetic resonance imaging (MRI) of the brain. However, visual detection of cavernomas in a large set of brain MRI slices is a time-consuming task. This paper proposes a computer aided cavernomas detection method based on T2-weighted MRI analysis. The proposed method includes the following steps: template matching to find suspected cavernoma regions and classification based on support vector machines (SVMs) to remove most of the false positives. The performance of the proposed technique is evaluated and a sensitivity of 0.96 is obtained after testing.

Computerized measurement of melanoma depth of invasion in skin biopsy images

Melanoma is the deadliest form of skin cancer, and its depth of invasion (DoI) is an important factor used by pathologist for grading the severity of skin disease. In this paper, we propose an automated technique for measuring melanoma DoI in MART1 stained skin histopathological images. The proposed technique first segments skin melanoma areas based on image color features. The skin epidermis is then segmented by a multi-thresholding method. After that, the skin granular layer is identified based on Bayesian classification of segmented epidermis pixels. Finally, the melanoma DoI is computed using a Hausdorff distance measure. Experiments on 28 skin biopsy images show that the proposed technique provides a superior performance in measuring the melanoma DoI than two closely related works.

Automatic measurement of melanoma depth of invasion in skin histopathological images

Measurement of melanoma depth of invasion (DoI) in skin tissues is of great significance in grading the severity of skin disease and planning patients treatment. However, accurate and automatic measurement of melanocytic tumor depth is a challenging problem mainly due to the difficulty of skin granular identification and melanoma detection. In this paper, we propose a technique for measuring melanoma DoI in microscopic images digitized from MART1 (i.e., meleanoma-associated antigen recognized by T cells) stained skin histopathological sections. The technique consists of four modules. First, skin melanoma areas are detected by combining color features with the Mahalanobis distance measure. Next, skin epidermis is segmented by a multi-thresholding method. The skin granular layer is then identified based on Bayesian classification of segmented skin epidermis pixels. Finally, the melanoma DoI is computed using a multi-resolution approach with Hausdorff distance measurement. Experimental results show that the proposed technique provides a superior performance in measuring the melanoma DoI than two closely related techniques.