Automatic segmentation of cell nuclei in Feulgen-stained histological sections of prostate cancer and quantitative evaluation of segmentation results

Digital image analysis of cell nuclei is useful to obtain quantitative information for the diagnosis and prognosis of cancer. However, the lack of a reliable automatic nuclear segmentation is a limiting factor for high-throughput nuclear image analysis. We have developed a method for automatic segmentation of nuclei in Feulgen-stained histological sections of prostate cancer. A local adaptive thresholding with an object perimeter gradient verification step detected the nuclei and was combined with an active contour model that featured an optimized initialization and worked within a restricted region to improve convergence of the segmentation of each nucleus. The method was tested on 30 randomly selected image frames from three cases, comparing the results from the automatic algorithm to a manual delineation of 924 nuclei. The automatic method segmented a few more nuclei compared to the manual method, and about 73% of the manually segmented nuclei were also segmented by the automatic method. For each nucleus segmented both manually and automatically, the accuracy (i.e., agreement with manual delineation) was estimated. The mean segmentation sensitivity/specificity were 95%/96%. The results from the automatic method were not significantly different from the ground truth provided by manual segmentation. This opens the possibility for large-scale nuclear analysis based on automatic segmentation of nuclei in Feulgen-stained histological sections.     

Automated cell nuclear segmentation in color images of hematoxylin and eosin-stained breast biopsy

OBJECTIVE: To develop an automated, reproducible epithelial cell nuclear segmentation method to quantify cytologic features quickly and accurately from breast biopsy. STUDY DESIGN: The method, based on fuzzy c-mean clustering of the hue-band of color images and the watershed transform, was applied to 39 images from 3 histologic types (typical hyperplasia, atypical hyperplasia, and ductal carcinoma in situ [cribriform and solid]). RESULTS: The performance of the segmentation algorithm was evaluated by visually determining the percentage of badly segmented nuclei (approximately 25% for all types), the percentage of nuclei that remained in clumps (4.5-16.7%) and the percentage of missed nuclei (0.4-1.5%) for each image. CONCLUSION: The segmentation algorithm was sensitive in that a small percentage of nuclei were missed. However, the percentage of badly segmented nuclei was on the order of 25%, and the percentage of nuclei that remained in clumps was on the order of 10% of the total number of nuclei in the duct. Even so, > 600 nuclei per duct, on average, were segmented correctly; that was a sufficient number by which to calculate accurate quantitative, cytologic, morphometric measurements of epithelial cell nuclei in stained tissue sections of breast biopsy.     

An edge relocation segmentation algorithm

An automated procedure that refines the nuclear contour of a previously segmented nucleus is described. The algorithm makes use of intensity information, edge magnitude information and both object and edge connectivity information. This automated procedure generates a closed contour precisely along the edge of the nucleus. The procedure was tested on a database of 3,680 red-green-blue images of thionin-SO2 and orange II-stained cervical cells obtained from normal and dysplastic samples. When used in conjunction with a simple threshold selection algorithm and an artifact removal routine, this edge relocation algorithm resulted in the correct segmentation of over 98% of the nuclei. Only 63 (1.7%) of all nuclei were incorrectly segmented.     

Cascade pattern recognition structure for improving quantitative assessment of estrogen receptor status in breast tissue carcinomas

OBJECTIVE: To develop and validate a computer-based approach for the quantitative assessment of estrogen receptor (ER) status in breast tissue specimens for breast cancer management. STUDY DESIGN: Microscopy images of 32 immunohistochemically (IHC) stained specimens of breast cancer biopsies were digitized and were primarily assessed for ER status (percentage of positively stained nuclei) by a histopathologist. A pattern recognition system was designed for automatically assessing the ER status of the IHC-stained specimens. Nuclei were automatically segmented from background by a pixel-based unsupervised clustering algorithm and were characterized as positively stained or unstained by a supervised classification algorithm. This cascade structure boosted the system's classification accuracy. RESULTS: System performance in correctly characterizing the nuclei was 95.48%. When specifying each case's ER status, system performance was statistically not significantly different to the physician's assessment (p = 0.13); when ranking each case to a particular 5-scale ER-scoring system (giving the chance of response to endocrine treatment), the system's score and the physician's score were in agreement in 29 of 32 cases. CONCLUSION: The need for reliable and operator independent ER-status estimation procedures may be served by the design of efficient pattern recognition systems to be employed as support opinion tools in clinical practice.     

Karyometry of infiltrating breast lesions

OBJECTIVE: To characterize nuclei from well-differentiated, moderately differentiated and poorly differentiated lesions of invasive breast cancer by karyometry and to test the hypothesis that these diagnostic categories form homogeneous sets. STUDY DESIGN: Histopathologic sections from 6 cases of well-differentiated, 11 cases of moderately differentiated and 17 cases of poorly differentiated ductal carcinomas were digitally recorded. From each case 100 nuclei were segmented and analyzed by karyometry. A discriminant analysis was performed, and nuclear and lesion signatures were computed. The nonsupervised learning algorithm P-index was applied. A progression curve per diagnostic category based on mean nuclear abnormality and a discriminant function score was derived. RESULTS: The well-differentiated lesions formed a homogeneous set, but both the moderately and poorly differentiated lesions showed 2 significantly different subpopulations with nuclei of substantially different nuclear abnormality and progression. CONCLUSION: The visual histopathologic diagnostic assessment of these lesions was based on an evaluation of both tissue architectural criteria and nuclear criteria. Here, only the pattern of nuclear chromatin was evaluated. Cases belonging to the same diagnostic category as assessed by their differentiation may be further characterized by the extent to which the nuclei deviate from normal. There was substantial case-to-case heterogeneity in these invasive lesions.     

Morphometry of bladder carcinoma: definition of a new variable

In this study two measurements of nuclear size, the mean area and the size distribution curve of nuclear area, were used to differentiate between two polar groups: nuclei from non-neoplastic urothelium and nuclei from transitional cell carcinomas of bladder with a poor clinical outcome. Wide separation of these groups is necessary if a measurement is to be used to assess tumour grade where the morphometric differences are intermediate between such polar groupings. Separation between two groups was best achieved using a weighted distribution of nuclear size and this is a means of objective scoring of urothelial tumours.     

Confocal DNA cytometry: a contour-based segmentation algorithm for automated three-dimensional image segmentation

BACKGROUND: Confocal laser scanning microscopy (CLSM) presents the opportunity to perform three-dimensional (3D) DNA content measurements on intact cells in thick histological sections. So far, these measurements have been performed manually, which is quite time-consuming. METHODS: In this study, an intuitive contour-based segmentation algorithm for automatic 3D CLSM image cytometry of nuclei in thick histological sections is presented. To evaluate the segmentation algorithm, we measured the DNA content and volume of human liver and breast cancer nuclei in 3D CLSM images. RESULTS: A high percentage of nuclei could be segmented fully automatically (e.g., human liver, 92%). Comparison with (time-consuming) interactive measurements on the same CLSM images showed that the results were well correlated (liver, r = 1.00; breast, r = 0.92). CONCLUSIONS: Automatic 3D CLSM image cytometry enables measurement of volume and DNA content of large numbers of nuclei in thick histological sections within an acceptable time. This makes large-scale studies feasible, whereby the advantages of CLSM can be exploited fully. The intuitive modular segmentation algorithm presented in this study detects and separates overlapping objects, also in two-dimensional (2D) space. Therefore, this algorithm may also be suitable for other applications.     

c-erbB-2/neu protein expression, DNA ploidy and S phase in breast cancer

Abstract. DNA content and c-erbB2/neu protein (p185) expression were evaluated by flow-cytometry and ELISA, respectively, in 166 specimens of primary breast cancer. A non-diploid DNA content was found in 88 tumours (53%), with the DNA index ranging from 0.7-2.7. S phase fraction (SPF) evaluation, performed in 130 cases, showed significantly higher values in aneuploid than in diploid tumours (median values, 17.3% and 5.8%, respectively). Thirty-six tumours (21.6%) showed p185 overexpression, while 45 (27.1%) and 85 (51.3%) showed intermediate and low expression, respectively. A good correlation ( P =0.0023) was found between DNA content and p185 positivity. Tumours with high p185 values were mainly aneuploid, while tumours with intermediate or low expression had variable degrees of DNA content. Furthermore, p185 concentration was significantly higher in aneuploid than in diploid tumours ( P =0.009). The highest rate of p185 (+) cases and the highest p185 concentrations occurred in triploid (1.3相似文献   

a-Methylacyl coenzyme A racemase is highly expressed in the intestinal-type adenocarcinoma and high-grade dysplasia lesions of the stomach

To study a-Methylacyl coenzyme A racemase (AMACR) expression in gastric intestinal-type adenocarcinoma and its precursors, we performed an immunohistochemical assay (using an avidin-biotin-peroxidase complex method) on 106 paraffin-embedded gastric mucosal biopsy samples and 25 gastrectomy samples (37 negative for dysplasia; 30 indefinite for dysplasia; 22 low-grade dysplasia; 25 high-grade dysplasia; and 34 invasive intestinal adenocarcinoma). The results showed that AMACR staining was uniformly negative in the groups negative for dysplasia and indefinite for dysplasia. Only 1 of 22 (4.5%) low-grade dysplasia showed weak staining for AMACR. In the groups of high-grade dysplasia and invasive intestinal-type adenocarcinoma, however, 19 of 25 (76%) and 18 of 34 (52.9%) were positive for AMACR respectively. Expression of AMACR was not correlated with location, H. Pylori infection or intestinal metaplasia. These results suggested that AMACR may play a role in the intermediate stage of gastric carcinogenesis. The high level expression of AMACR in high-grade dysplasia and carcinoma suggests that it may be a useful biomarker in distinguishing high-grade dysplasia and carcinoma from low-grade dysplasia.     

Tissue architectural features for the grading of prostatic carcinoma

In research for the development of a computer-aided workstation for the objective grading of prostatic carcinoma, tissue architectural (histometric) features were analyzed in ten cases each of well-differentiated, moderately differentiated and poorly differentiated carcinoma (as subjectively graded by the consensus of a panel of experts). Sections were cut at 4 microns, stained by the Feulgen reaction and digitized by two different video-based photometric systems. Some images were interactively segmented, considering the histometric clues to be studied; others were automatically segmented by an expert system-guided technique. The latter procedure produced good results, with over 90% of the nuclei judged to be correctly segmented in 64% of the fields studied and over 80% in another 24% of the fields. While the number of nuclei per field provided some separation of well-differentiated from other lesions, the number of nuclei per gland distinguished between well-differentiated and moderately differentiated lesions. Simplicial decomposition of the images also provided a measure of the degree of differentiation, as did the "texture" of the nuclear placement, based on two run-length statistics. Combination of the run-length features distinguished the three categories of lesions with statistical significance. The results of this study provided insights into the problems (such as the effect of field boundaries) faced in the design of an computer-aided grading system. They also showed the value of expert system-guided scene segmentation and of such histometric features as the field cellularity and the number of nuclei per gland for the discrimination between lesions of different grades of differentiation.     

Automatic image segmentation of nuclear stained breast tissue sections using color active contour model and an improved watershed method

Automatic image segmentation of immunohistologically stained breast tissue sections helps pathologists to discover the cancer disease earlier. The detection of the real number of cancer nuclei in the image is a very tedious and time consuming task. Segmentation of cancer nuclei, especially touching nuclei, presents many difficulties to separate them by traditional segmentation algorithms. This paper presents a new automatic scheme to perform both classification of breast stained nuclei and segmentation of touching nuclei in order to get the total number of cancer nuclei in each class. Firstly, a modified geometric active contour model is used for multiple contour detection of positive and negative nuclear staining in the microscopic image. Secondly, a touching nuclei method based on watershed algorithm and concave vertex graph is proposed to perform accurate quantification of the different stains. Finally, benign nuclei are identified by their morphological features and they are removed automatically from the segmented image for positive cancer nuclei assessment. The proposed classification and segmentation schemes are tested on two datasets of breast cancer cell images containing different level of malignancy. The experimental results show the superiority of the proposed methods when compared with other existing classification and segmentation methods. On the complete image database, the segmentation accuracy in term of cancer nuclei number is over than 97%, reaching an improvement of 3–4% over earlier methods.     

Segmentation of color images from serous cytology for automated cell classification

OBJECTIVE: To design an automated system for the classification of cells based on analysis of serous cytology, with the aim of segmenting both cytoplasm and nucleus using color information from the images as the main characteristic of the cells. STUDY DESIGN: The segmentation strategy uses color information coupled with mathematical morphology tools, such as watersheds. Cytoplasm and nuclei of all diagnostic cells are retained; erythrocytes and debris are eliminated. Special techniques are used for the separation of clustered cells. RESULTS: A large set of cells was assessed by experts to score the segmentation success rate. All cells were segmented whatever their spatial configurations. The average success rate was 92.5% for nuclei and 91.1% for cytoplasm. CONCLUSION: This color information-based segmentation of images of serous cells is accurate and provides a useful tool. This segmentation strategy will improve the automated classification of cells.     

DNA quantification of squamous cell carcinoma of the oesophagus by flow cytometry and cytophotometric image analysis using formalin fixed paraffin embedded tissue.

This is the first comparative study of DNA quantification of oesophageal squamous cell carcinoma by flow cytometry (FCM) and image cytometry (ICM) using formalin fixed paraffin embedded tissue. The potential advantages of ICM include the identification of a reliable control cell population; avoidance of non-tumour stromal and inflammatory cell nuclei, nuclear fragments, degenerate cell nuclei and doublets, triplets etc., which are not possible with FCM using archival tissue. Twenty-eight cases, all of the same stage (stage 2a) and similar grade (well or moderately differentiated) were analysed. The cases were separated into two groups, those that had succumbed to tumour in less than 18 months (group A) and those that were tumour free at least 18 months post-resection (group B). Using ICM all 28 tumours yielded interpretable histograms by comparison to 25 of 28 using FCM. Aneuploidy was identified in 100% of cases in group A using ICM (in comparison to 73% by FCM) and in 73% of group B using ICM (in comparison to 44% by FCM). Any tumour aneuploid by FCM was also aneuploid by ICM. Nine cases aneuploid by ICM were euploid by FCM. The mean 5C exceeding rate (% of cells whose nuclei contain a DNA mass equivalent to > 5 sets of 23 chromosomes) was 21% in group A and 14% in group B (P < 0.01). Euploidy was confined to tumours of those patients disease free for more than 18 months. The conclusions of this study are that: firstly, ICM is superior in its yield of interpretable histograms to FCM using formalin fixed paraffin embedded tissue; secondly, ICM is more sensitive in the identification of aneuploid stemlines than FCM; and thirdly, euploid tumours (as detected by ICM) appear to have a better prognosis than aneuploid tumours of similar stage and grade.     

Benign phyllodes tumour vs fibroadenoma: FNA cytological differentiation

The differential diagnosis of fibroadenomas vs phyllodes tumours by fine needle aspiration (FNA) cytology is not possible in the majority of cases. The present study aims to look at common and dissimilar features to allow differentiation, if possible. We reviewed the FNA findings of 18 histologically proven phyllodes tumours and 18 fibroadenomas, checking in each case the epithelial features, the stromal features, and any atypia. Using a semi-quantitative score assessed by two observers we were able in most cases to distinguish a phyllodes tumour from a fibroadenoma. The most important criteria were larger stromal fragments, numerous plump stromal bare nuclei, and the higher ratio of stromal bare nuclei to epithelial bare nuclei in phyllodes tumours. In the present study, an original diagnosis of phyllodes tumour was made in 7/18 (38.9%) cases but with our criteria this could be improved to 15/18 (83.3%) cases. Therefore, the presence of specific stromal features in a dimorphic cellular pattern should suggest the correct diagnosis and differentiate its appearance from a cellular fibroadenoma.     

Quantitative study of KI-67 antibody staining in non-Hodgkin's lymphomas using image analysis.

In sections from 32 B malignant lymphomas (ML), the total KI-67 stained area was compared to the number of KI-67 positive cells in order to demonstrate the reliability of using image analysis to quantify the proliferative activity. The total KI-67 area percentage correlated highly with the number of KI-67 positive cellular profiles (r = .93). Significant differences were found between low- and high-grade ML according to the Kiel classification (mean values +/- SD, respectively, of 7.7 +/- 3.81% and 16.6 +/- 6.23%), and between low-, or intermediate- and high-grade ML only, according to the International Working Formulation. Within the Working Formulation, the statistical analysis grouped the diffuse large cell subtype of intermediate grade with the immunoblastic high-grade subtype. A wide range of KI-67 area percentage values was noted, particularly in follicular ML; for these follicular ML, considering follicular areas only, values were comparable to high-grade ML (14.8 +/- 6.60%). In conclusion, the KI-67 area percentage is a reliable alternative method to manual cell counting, and image analysis allows quicker measurements appropriate to large and strictly lymphomatous areas, using a greater number of cells than in manual cell counting.     

A comparison of some quick and simple threshold selection methods for stained cells

The segmentation accuracy of four fast and simple gray-scale threshold selection methods were compared using a data base of stained cervical cell images. Some postprocessing was applied to the segmented images to increase the accuracy of the nuclear segmentation. The most accurate method correctly segmented the cytoplasm of 81% of the cell images and the nuclei of 78% of the cell images in the data base.     

Prognostic value of flow cytometrically determined DNA-ploidy, intracellular pH and esterase activity of non-small cell lung carcinomas.

30 surgical specimens of patients with non-small cell lung carcinomas (NSCLC) were investigated. Significant increases of intracellular pH values in epithelial and inflammatory cells, in the percentage of dead epithelial and inflammatory cells and in the cell volume of vital inflammatory cells in cancerous lung tissue were encountered. Furthermore, decreases of the esterase activity of vital epithelial cells and of the percentage of free cell nuclei were observed. The DNA aneuploidy in 36.6% of the tumours was frequently associated with non-squamous cell carcinomas and stage II, III, IV tumours. Patients with DNA aneuploid tumours had a significantly shorter survival rate than those with DNA euploid tumours. Within the different tumour stages a similar tendency was observed which was, however, only significant in stage III tumour patients. Stage III tumours constitute therefore a heterogeneous entity with a worse prognosis for DNA aneuploid tumour patients. The intracellular pH values and esterase activity as well as the cell volume, the percentage of free cell nuclei and dead inflammatory or epithelial cells contained no significant prognostic information.     

Methylation of HOXA9 and ISL1 Predicts Patient Outcome in High-Grade Non-Invasive Bladder Cancer

Introduction

Inappropriate DNA methylation is frequently associated with human tumour development, and in specific cases, is associated with clinical outcomes. Previous reports of DNA methylation in low/intermediate grade non-muscle invasive bladder cancer (NMIBC) have suggested that specific patterns of DNA methylation may have a role as diagnostic or prognostic biomarkers. In view of the aggressive and clinically unpredictable nature of high-grade (HG) NMIBC, and the current shortage of the preferred treatment option (Bacillus:Calmette-Guerin), novel methylation analyses may similarly reveal biomarkers of disease outcome that could risk-stratify patients and guide clinical management at initial diagnosis.

Methods

Promoter-associated CpG island methylation was determined in primary tumour tissue of 36 initial presentation high-grade NMIBCs, 12 low/intermediate-grade NMIBCs and 3 normal bladder controls. The genes HOXA9, ISL1, NKX6-2, SPAG6, ZIC1 and ZNF154 were selected for investigation on the basis of previous reports and/or prognostic utility in low/intermediate-grade NMIBC. Methylation was determined by Pyrosequencing of sodium-bisulphite converted DNA, and then correlated with gene expression using RT-qPCR. Methylation was additionally correlated with tumour behaviour, including tumour recurrence and progression to muscle invasive bladder cancer or metastases.

Results

The ISL1 genes’ promoter-associated island was more frequently methylated in recurrent and progressive high-grade tumours than their non-recurrent counterparts (60.0% vs. 18.2%, p = 0.008). ISL1 and HOXA9 showed significantly higher mean methylation in recurrent and progressive tumours compared to non-recurrent tumours (43.3% vs. 20.9%, p = 0.016 and 34.5% vs 17.6%, p = 0.017, respectively). Concurrent ISL1/HOXA9 methylation in HG-NMIBC reliably predicted tumour recurrence and progression within one year (Positive Predictive Value 91.7%), and was associated with disease-specific mortality (DSM).

Conclusions

In this study we report methylation differences and similarities between clinical sub-types of high-grade NMIBC. We report the potential ability of methylation biomarkers, at initial diagnosis, to predict tumour recurrence and progression within one year of diagnosis. We found that specific biomarkers reliably predict disease outcome and therefore may help guide patient treatment despite the unpredictable clinical course and heterogeneity of high-grade NMIBC. Further investigation is required, including validation in a larger patient cohort, to confirm the clinical utility of methylation biomarkers in high-grade NMIBC.     

Histometric features for the grading of prostatic carcinoma

Histometric features for the objective grading of prostatic adenocarcinoma in histologic specimens were analyzed in five cases each of well, moderately and poorly differentiated lesions. Tissue sections from the selected cases were stained by the Feulgen method and digitized by a video-based microphotometer. Twenty total fields were recorded for each grade: ten at high resolution (an image sampling of 0.5 micron per pixel) and ten at low resolution (0.8 micron per pixel), with two fields per case recorded at each resolution. The images were segmented by an automated expert system-guided scene segmentation procedure. The performance of that procedure was measured by comparing the automated counts of nuclei in the segmented fields to the visual counts made by a pathologist in the same fields. For well, moderately and poorly differentiated cases, respectively, the nuclear counts made by the expert system at high resolution were 2.7%, 4.2% and 4.7% higher than the visual counts (as estimated from a total of 6,628 nuclei), but 1.2%, 2.5% and 1.1% lower at low resolution (10,329 nuclei). High-resolution features and tissue textural features were computed for each case. The high-resolution features showed good separation between the three groups of cases. The tissue textural features showed consistent separation between well and moderately differentiated cases. The relaxation of the spatial resolution (to 0.8 micron/pixel spacing) did not affect the selection of features, but led to less separation between the data from different grades. In conclusion, the automated system performed satisfactorily in distinguishing sections of prostatic tumors of varying degrees of differentiation.