Computerized morphometric discrimination between normal and tumoral cells in oral smears

The oral exfoliative cytology allows a quick and fairly accurate assessment of suspicious lesions of the oral cavity. Within this context, our paper proposes a quantitative approach, focusing on the construction of a classifier for detecting the presence of the tumoral cells on oral smears. The design of the classifier relies on a detailed computerized analysis of the individual morphometric features exhibited by two large known populations of normal and tumoral cells, respectively; the digital image processing was performed in the Zeiss KS400 environment. The classifier was implemented as a neural network with step activation function, whose parameters were obtained from an adequate training, based on the nuclear and cytoplasmic areas of the cells belonging to the two populations. Our procedure based on this classifier was meant to operate by identifying the tumoral or normal nature of any cell randomly selected from a smear. To identify the nature of an arbitrary cell, its nuclear and cytoplasmic areas are presented at the input of the classifier. The classification procedure was tested on several smears, and the results coincided with the pathological diagnosis in all the considered cases. The performances of our approach are discussed in comparison with other analytical methods previously reported in oral exfoliative cytology. These discussions emphasize the role of numerical information exploited for the classifier design, concluding that the individual morphometric features are more meaningful than the global characterization of smears by mean values.     

Digital image analysis of AgNORs in cervical smears of women with premalignant and malignant lesions of the uterine cervix

We performed a hospital-based, unmatched case-control study to investigate the association between progressive stages of cervical neoplasia and digital analysis of cell proliferation by silver stained nucleolus organizer region associated proteins (AgNORs). We measured cell proliferation levels in the cervical epithelial cells of 10 women with low grade squamous intraepithelial lesions (LG-SIL), eight with high grade squamous intraepithelial lesions (HG-SIL), 11 with cervical cancer (CC) and eight with no cervical lesions (controls) using the AgNORs technique. Cell proliferation was measured by digital image analysis (DIA). DIA revealed increased total areas of AgNORs in HG-SIL and CC compared to LG-SIL and control patients. AgNORs with a kidney or cluster shape exhibited greater areas than those with a spherical or long shape. We propose a cut-off of 118 pixels to differentiate benign (control and LG-SIL) from malignant (HG-SIL and CC) lesions. DIA of AgNORs is a simple and inexpensive method for studying proliferation. The increased total area of AgNORs in malignant lesions provides information regarding cell behavior and may be related to cervical carcinogenesis; however, further validation studies are required to establish its usefulness in cytological analysis.     

Routine culturing of normal,dysplastic and malignant human mammary epithelial cells from small tissue samples

Summary We compared the growth and morphology of normal, dysplastic and malignant human mammary epithelial cells (HMEC) in medium containing 5% human serum, a serum-free medium (32) and serum-free medium with a low Ca⁺⁺ concentration. Tissues were dissociated and epithelial organoids or single cells were seeded onto collagen-coated dishes. The cells grew in serum-containing medium, but growth of fibroblasts was also stimulated. The serum-free medium consistently selected for and stimulated the growth of epithelial cells. There was little advantage in reducing the Ca⁺⁺ concentration to further increase cell yield. This serum-free primary culture system allows us to routinely prouce sufficient numbers of HMEC from small tissue samples for molecular biological investigations. Furthermore, the maintenance of cells in a defined medium can provide a system for evaluating the direct effects of factors on gene expression. This work was supported by a grant from the National Cancer Institute of Canada and funds contributed by Mr. B. T. Wharton in memory of his wife, Nadia.     

Canonical analysis of cells in normal and abnormal cervical smears

Over 4,000 cells from 105 normal and 96 abnormal uterine cervical scrapes were prepared according to the UCLA monolayer procedure, stained by a routine Papanicolaou method and visually classified by two cytopathologists and a technologist into seven classes: parabasal, metaplastic, mild dysplasia, moderate dysplasia, severe dysplasia, carcinoma in situ and invasive carcinoma. Canonical analysis was used to correlate effects-coded class membership variables with 23 cell features derived from digital image analysis. In general, nuclear texture measures derived from linear combinations of run-length correlations along with features derived from a Markov transitional probability matrix provided the best predictors of cell class. After cells were divided into benign (moderate dysplasia or less) and malignant (severe dysplasia or worse) groups, discriminant analysis correctly classified 84% of the benign cells and 91% of the malignant cells.     

Analysis of silver binding nucleolar organizer regions in exfoliative cytology smears of potentially malignant and malignant oral lesions

Nucleolar organizer regions are nucleolar components that contain proteins that are stained selectively by silver methods; they can be identified as black dots throughout the nucleolus and are known as silver binding nucleolar organizer regions (AgNOR). The number of AgNOR is related to the cell cycle and the proliferative activity of the cells. We investigated AgNOR using exfoliative cytology smears of potentially malignant oral lesions. Eighty individuals were divided into four equal groups: healthy controls, oral leukoplakia, oral submucous fibrosis and oral squamous cell carcinoma. The mean number of AgNOR in each study group gradually increased from control to oral leukoplakia to oral submucous fibrosis to oral squamous cell carcinoma. The proliferative index was increased in the oral premalignant and malignant patients compared to normal subjects. The mean AgNOR size gradually increased from control to oral leukoplakia to oral submucous fibrosis to oral squamous cell carcinoma. Spherical shaped AgNOR were most common in controls, whereas large, clustered and kidney shapes were most common in oral squamous cell carcinoma. Multiparameter analysis of AgNOR in oral exfoliative smears is a simple, sensitive and cost-effective method for differentiating premalignant from malignant lesions and can be used in conjunction with routine cytomorphological evaluation.     

Morphometry of pseudoepitheliomatous hyperplasia: objective comparison to normal and dysplastic oral mucosae

OBJECTIVE: To compare the architectural and morphometric features of pseudoepitheliomatous hyperplasia (PEH) associated with oral granular cell tumors (GCT), normal oral mucosa and oral epithelial dysplasia. STUDY DESIGN: Quantitative comparisons between the diagnostic entities were carried out at the tissue level by estimating the fractal complexity of the epithelial connective tissue interface and at the cellular level by analyzing the morphometric features of algorithmically segmented epithelial cell areas. RESULTS: Casewise multivariate analysis showed that the fractal properties produced a correct discrimination rate of 96.4% between PEH and normal mucosa. Cellular parameters gave a 100% correct discrimination rate between PEH and mild dysplasia. Combining the fractal and cellular properties also showed 100% discrimination between PEH and normal mucosa and between PEH and mild dysplasia. CONCLUSION: The results show that PEH associated with GCT displays quantifiable morphometric features that make it differentiable from normal oral mucosa and oral epithelial dysplasia.     

Masking effect of hormonal contraceptives on discriminating quantitative features of visually normal intermediate cells in positive and negative cervical smears

In quantitative studies of visually normal intermediate cells in smears from patients with cervical neoplasia, the contraceptive status of the patients has not previously been taken into account. In this study cervical smears from 151 patients with cervical intraepithelial neoplasia (CIN) III or invasive carcinoma and from 360 normal controls were grouped according to week of menstrual or pill cycle and mode of hormonal contraception. The nuclear-cytoplasmic (N/C) ratio of visually normal intermediate cells in smears from patients with neoplasia was significantly different from that of the normal controls (P less than .001). Based on nuclear area and N/C ratio, the percentages of intermediate cells correctly classified as having come from positive or negative smears were significantly better in women with ovulatory cycles (non-users) than in women using hormonal contraceptives (P less than .025). It is concluded that hormonal contraceptives can mask the salient quantitative features of visually normal intermediate cells from patients with CIN and the contraceptive status thus has to be taken into account in such studies.     

Numerical chromosomal aberrations of chromosome 1 and 7 in dysplastic cervical smears

Cervical smears with Papanicolaou's staining (PAP) reveal only morphological characteristics of epithelial cells of the cervix uteri. Since chromosomal aberrations are known to play a role in malignant transition, we analyzed cervical smears for numerical changes of the chromosomes 1 and 7 with fluorescence in-situ hybridization to probe for a diagnostic value of these chromosomes in the characterization of cervical dysplasia. Cervical smears were collected from 21 patients with suspect histology of curettage or biopsy specimen, 14 of them having been subsequently graded as cervical intraepithelial neoplasia (CIN) III and 5 as CIN II. Nineteen normal cervical smears (PAP I-II) served as controls. Smears were hybridized with chromosomal enumeration probes for chromosome 1 and 7. Disomic cells (2 copies of chromosome 1 and 7) were decreased in the CIN II (63%) and CIN III group (57%) with respect to the control group (77%). Cells with 3 signals for chromosome 7 were significantly more frequent in the CIN III and the CIN II group than in the control group (6.7, 6.4 and 0.7%, respectively). Only the CIN II group (10%), but not CIN II (6%), showed a significant trisomy for chromosome 1 as compared with the controls (3.8%). A close correlation between the incidence of trisomy 1 or 7 and PAP grading was observed. PAP III-IIID smears with high trisomy 1 counts corresponded to CIN III histology, while all CIN II patients were PAP III-IIID with low incidence of trisomy 1. We conclude that trisomy of chromosome 7 is a feature of cervical dysplasia and seems to be an early event in dysplastic transition. In contrast, trisomy of chromosome 1 is observed only in high grade dysplasia and may be a marker for pre-malignant lesions.     

Two-locus linkage analysis of cutaneous malignant melanoma/dysplastic nevi.

Previous linkage analyses of 19 cutaneous malignant melanoma/dysplastic nevi (CMM/DN) kindreds showed significant evidence of linkage and heterogeneity to both chromosomes 1p and 9p. Five kindreds also showed evidence of linkage (Z>0.7) to both regions. To further examine these findings, we conducted two-trait-locus, two-marker-locus linkage analysis. We examined one homogeneity and one heterogeneity single-locus model (SL-Hom and SL-Het), and two-locus (2L) models: an epistatic model (Ep), in which CMM was treated as a genuine 2L disease, and a heterogeneity model (Het), in which CMM could result from disease alleles at either locus. Both loci were modeled as autosomal dominant. The LOD scores for CMM alone were highest using the SL-Het model (Z = 8.48, theta = .0). There was much stronger evidence of linkage to chromosome 9p than to 1p for CMM alone; the LOD scores were approximately two times greater on 9p than on 1p. The change in LOD scores from an evaluation of CMM alone to CMM/DN suggested that a chromosome 1p locus (or loci) contributed to both CMM and CMM/DN, whereas a 9p locus contributed more to CMM alone. For both 2L models, the LOD scores from 1p were greater for CMM/DN than for CMM alone (Ep: Z=4.63 vs. 3.83; Het: 4.94 vs. 3.80, respectively). In contrast, for 9p, the LOD scores were substantially lower with CMM/DN than with CMM alone (Ep: 4.64 vs. 7.06; Het: 5.38 vs. 7.99, respectively). After conditioning on linkage to the other locus, only the 9p locus consistently showed significant evidence for linkage to CMM alone. Thus, the application of 2L models may be useful to help unravel the complexities of familial melanoma.     

Detecting infection-related changes in peripheral blood smears with image analysis techniques

High-resolution image analysis has the potential to flag subtle changes in white blood cell morphology that may indicate the presence of certain diseases. A study was made of the feasibility of identifying patients with hematologic bacterial infections (sepsis) using measurements on Wright-Giemsa-stained peripheral blood smears. Neutrophils and lymphocytes from a group of patients with sepsis and from a control group were digitized, and parameters quantifying geometry, color, texture and shape were extracted. While color parameters differed the most between the infected and control samples, substantial differences in geometric, texture and shape parameters also were observed. Analysis of the data showed that individual neutrophils and lymphocytes from patients with sepsis were distinguishable from those of the control group with better than 84% accuracy. When average parameters were calculated from all cells of one type for each specimen, 100% accurate classification was obtained. These studies demonstrate that the image-analysis techniques used are sensitive enough to detect disease-related changes in cell morphology that are generally too subtle for reliable detection by the human eye. Future experiments will determine the specificity of this test for bacterial infections and will explore the possibility of using image analysis techniques on peripheral blood to detect and monitor a wide variety of diseases.     

Automated nuclear image morphometry on fine needle aspiration smears of malignant round cell tumors

OBJECTIVE: To analyze nuclear image morphometry in fine needle aspiration cytology smears of different groups of malignant round cell tumors (MRCTs) to evaluate its diagnostic role. STUDY DESIGN: In this study there were 55 cases of MRCT, consisting of 18 Ewing's sarcoma (EW), 10 neuroblastoma (NB), 5 non-Hodgkin's lymphoma (NHL), 6 rhabdomyosarcoma (RMS), 4 peripheral neuroectodermal tumor (PNET), 8 Wilm's tumor (WT), 2 retinoblastoma (RB) and 2 undifferentiated round cell tumor (URCT). A Leica image cytometer with Quantimet 600 software (Leica, Cambridge, U.K) was used to measure nuclear area, nuclear diameter, nuclear perimeter, nuclear convex perimeter (CP), nuclear roundness and nuclear convex area on hematoxylin and eosin-stained cytologic smears. At least 100 cells were studied in each case. RESULTS: The RB group of tumors showed the highest mean nuclear area (NA), convex area (CA), CP, diameter (D), perimeter (P) and roundness (R). RMS had the highest mean CA, and URCT had the highest mean roundness. ANOVA was performed on the tumors and showed significant differences for all the variables in all the groups (P < .000). All the morphometric data (except roundness) were significantly different in RMS versus all other MRCTs except RB. Similarly, morphometric data on WT were also significantly different from that on NHL. Most of the morphometric data (except CA and R) showed significant differences between RB and all other MRCTs except RMS. PNET, EW and NB could not be differentiated with those variables. CONCLUSION: RMS and RB could successfully be differentiated from all other MRCTs with the help of morphometry. It was not possible to differentiate RMS and RB by image cytometry (ICM) since the ICM data overlapped in those two groups. It was possible to differentiate WT and NHL with ICM. Nuclear ICM was not significantly different in the NB, PNET and ES groups, and probably ICM would not be very helpful to differentiate these groups of MRCT.     

Automated identification of diploid reference cells in cervical smears using image analysis

BACKGROUND: Acquisition of DNA ploidy histograms by image analysis may yield important information regarding the behavior of premalignant cervical lesions. Accurate selection of nuclei for DNA measurement is an important prerequisite for obtaining reliable data. Traditionally, manual selection of nuclei of diagnostic and reference cells is performed by an experienced cytotechnologist. In the present study, a method for the fully automated identification of nuclei of diploid epithelial reference cells in Feulgen- restained Papanicolaou (PAP) smears is described. METHODS: The automated procedure consists of a decision tree implemented on the measurement device, containing nodes with feature threshold values and multivariate discriminant functions. Nodes were constructed to recognize debris and inflammatory cells, as well as diploid and nondiploid epithelial cells of the uterine cervix. Evaluation of the classifier was performed by comparing resulting diploid integrated optical densities with those from manually selected reference cells. RESULTS AND CONCLUSION: On average, automatically acquired values deviated 2.4% from manually acquired values, indicating that the method described in this paper may be useful in cytometric practice.