Prediction of cervical neoplasia diagnosis groups. Discriminant analysis on digitized cell images

The purpose of this study was to develop discriminant analysis models for predicting cervical dysplasia/neoplasia case diagnoses using cytometric features derived from the digital image analysis of cell monolayers. The data base consisted of 925 cells from 27 cases diagnosed either as moderate dysplasia (n = 10), severe dysplasia (n = 5), carcinoma in situ (n = 8) or invasive carcinoma (n = 4) on both tissue biopsy and monolayer preparations. Cell features examined were cell diameter, nuclear diameter, nuclear mean optical density (OD), nuclear integrated OD (IOD), nuclear OD standard deviation, normalized IOD, nuclear texture and nuclear-cytoplasmic ratio. Features derived from cells visually classified as moderate dysplasia correctly predicted the case diagnosis of moderate dysplasia versus more severe disease for 85% of the cells. Prediction models using summary measures (mean and variance) derived from all visually classified abnormal cells within each case correctly separated all cases into their respective diagnostic categories. These findings suggest that dysplastic cells in a cytologic sample have features that collectively reflect the tissue diagnosis, regardless of the visual differences among the cells. Such information has potential use for diagnosis and possibly for prognosis.     

Malignancy-associated changes in breast tissue detected by image cytometry.

In several tissues, nuclear differences have been described in normal-appearing cells from patients with invasive carcinomas compared to cases without invasive carcinoma, a phenomenon known as malignancy-associated changes (MACs). The aim of this study was to determine the presence of malignancy-associated changes in breast tissue. Image cytometry was performed on Feulgen stained tissue sections of patients with usual ductal hyperplasia with (n = 30) or without (n = 41) adjacent invasive breast carcinoma. Nuclear features of normal-appearing cells as well as of usual ductal hyperplastic cells were separately compared between the two groups. Many features of normal-appearing epithelial cells were significantly different between cases with and without invasive cancer. Significant differences were also found by measuring ductal hyperplastic nuclei instead of normal-appearing nuclei. Cases with or without cancer could be distinguished with a classification accuracy of 80% by discriminant analysis using 2 nuclear features derived from ductal hyperplastic cells. In conclusion, image cytometry on breast tissue sections shows that malignancy-associated changes can be found in normal as well as in usual ductal hyperplastic breast cells. This could be clinically relevant for the detection of occult breast cancer, for the prediction of risk in these lesions, and to monitor the effect of chemopreventive agents.     

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.     

Nuclear texture features for classifying benign vs. dysplastic or malignant squamous epithelium of the larynx.

OBJECTIVE: To search for nuclear features and feature combinations able to assess malignancy and premalignant changes on tissue sections of laryngeal squamous epithelium. STUDY DESIGN: A total of 139 lesions of benign changes (BC) (n = 44), epithelial dysplasias (ED) (n = 50) and invasive laryngeal cancer (LC) (n = 45) were retrieved from archival pathology specimens. The goal of this study was to identify the best features or feature combinations that discriminate BC from LC and also reflect the degree of ED. In order to verify the results on independent data, the groups were split into two separate subgroups, one for training and one for testing. RESULTS: On the test set of slides, the overall correct classification of BC vs. LC cases was 82% using only one feature, fractal2_area. This classification rate could be increased to 91% when a discriminant function based on 10 features was used. However, this gain was not significant. CONCLUSION: Fractal texture features can be used to assess malignancy on tissue sections as an alternative to DNA measurement. In this study feature combinations did not significantly improve classification rates.     

Karyometric marker features in fine needle aspirates of invasive follicular carcinoma of the thyroid.

Karyometric measurements were performed on fine needle aspirates of clearly identifiable tumor areas and adjacent normal-appearing areas in the surgical specimens from ten patients with invasive follicular carcinoma of the thyroid. Similar measurements were performed on aspirates from nine patients free of thyroid disease (controls). A total of 95 karyometric features were evaluated for each nucleus. Analysis of variance of optical density values indicated (1) a similarity between tumor and normal-appearing nuclei from carcinoma cases, (2) a significant difference between those nuclei and control nuclei and (3) that most of the differences were due to the differences of tissue origin. Stepwise linear discriminant analysis selected ten features that produced a statistically highly significant separation of tumor nuclei from control nuclei. A similar analysis selected six features that produced a statistically highly significant discrimination of normal-appearing nuclei from control nuclei; the validity of those karyometric features as markers of malignancy in normal-appearing nuclei from tissues adjacent to invasive follicular carcinomas of the thyroid was demonstrated by analysis in further training and control sets of nuclei. This analysis in thyroid aspirates identified more marker features than did a previous similar analysis using tissue sections.     

1H NMR metabolic profiling of human serum associated with benign and malignant gallstone diseases

Proton NMR based metabolic profile of serum associated with different gallbladder pathologies is presented. Quantitative and qualitative variations in the metabolic profile of serum in control samples and three different pathologies of gallbladder, chronic cholecystitis, xanthogranulomatous cholecystitis and carcinoma of gallbladder has been evaluated by use of ¹H NMR based metabonomics and multivariate chemometric methods. Multivariate partial least square discriminant analysis of ¹H NMR spectra showed a clear discrimination between control and diseased groups on the basis of quantitative and qualitative metabolic variations. Increased levels of lactate and pyruvate whereas decreased levels of glucose, some amino acids and low density lipoprotein/very low density lipoprotein (LDL/VLDL) were observed. These metabolites, responsible for class discrimination, from different metabolic pathways could be considered as the signatures of the carcinoma of gallbladder.     

Distinguishing cortical adrenal gland adenomas from carcinomas by their quantitative nuclear features

OBJECTIVE: To explore data from a set of cases of adrenal cortical adenomas with different endocrine syndromes and carcinomas to determine whether quantitative image analysis of nuclear features might be used to separate the groups. STUDY DESIGN: Fifteen adrenal cortical tumors in which clinical information and optimally preserved, paraffin-embedded tissue were available were used. There were 10 adenomas and 5 carcinomas. Among the adenomas, five were associated with primary hyperaldosteronism (Conn's syndrome) and five with Cushing's syndrome. Five-micrometer-thick sections were stained with hematoxylin and eosin. In each case 50 nuclei were measured, and a number of morphometric and densitometric features were extracted. The data were subjected to multivariate analysis. RESULTS: Quantitative analysis showed that nuclei from adrenal carcinomas are larger than those from adenomas. Total optical density (OD) had a near-diploid distribution in the adenomas, while it was clearly aneuploid in the carcinomas. The pixel OD histograms were almost identical for all groups. Differences in nuclear chromatin texture were found between adenomas and carcinomas and also between the two adenoma categories. Multivariate analysis showed good discrimination between carcinomas and adenomas (Wilks lambda = .628, P < .0001) and between adenomas. However, based on Bayesian decision boundaries, 20-25% of carcinoma nuclei could be expected to be in the range of adenoma, and about 12% of Cushing's adenoma nuclei and 15% of Conn's adenoma nuclei would be classified as carcinoma. CONCLUSION: Computer-assisted analysis of nuclear characteristics proved useful in identifying and describing differences between groups of tumors arising in the adrenal cortex.     

Discriminant analysis on cells from developing squamous cancer of the respiratory tract.

Cytologic preparations made from the tracheobronchial tree taken by the Schreiber catheter have been scanned by three color microphotometry. The digitized cell images were processed by the analytical cytodiagnostic programs of the TICAS system. Cells were sorted into two control groups and five groups of increasing atypia ranging from normal epithelium to invasive squamous cell carcinoma. Standard statistical tests, including Wilk's Lambda, Rao's V, and the Kruskal-Wallis tests are performed on these subsets of cell image features. This study demonstrates that discriminant analyses permit differentiation between normal cells and those from marked atypia or carcinoma and that the classification achieves a high degree of agreement with visual assignment.     

Diagnostic value of qualitative and quantitative variables in thyroid lesions

Two hundred and thirty-three thyroid lesions were studied by fine needle aspiration (FNA) cytology using standard cytologic criteria available in the literature. These included 114 cases of nodular colloid goitre (NCG), 47 cases of Hashimoto's thyroiditis (HT), 12 follicular adenomas (FAd), five cases of subacute thyroiditis and three cases of thyrotoxicosis among the benign lesions. The malignant lesions seen were 30 cases of papillary carcinoma (PCa), 16 follicular carcinomas (FCa), three cases with double lesions, e.g. papillary carcinoma with coexisting NCG, and three of papillary carcinoma with HT. Emphasis was given to eight qualitative and quantitative (morphometric) variables in these various thyroid lesions. Cell measurements were done using a Visopan Lux projection microscope. The three qualitative variables included type of nuclear membrane (regular/irregular), type of nuclear chromatin and the presence or absence of conspicuous nucleoli. The quantitative variables studied were nuclear diameter, nuclear area, cytoplasmic diameter, cytoplasmic area & N/C ratio. Statistical analysis was performed in order to know whether the standard cytologic criteria used at FNA cytology in the literature (increased cellularity, microfollicles, increased N/C ratio, absence of significant haemosiderin-laden macrophages and scanty colloid) could differentiate a follicular adenoma from a follicular carcinoma. A statistical analysis was also performed to establish the utility of the qualitative and quantitative variables. The results showed that none of the standard cytologic criteria applied could differentiate follicular adenoma from a follicular carcinoma. With regard to qualitative variables, irregularity of nuclear membrane and presence of conspicuous nucleoli were most significant in papillary carcinoma, followed by follicular carcinoma, then by follicular adenoma; these features being hardly evident in nodular colloid goitre and Hashimoto's thyroiditis. A coarse nuclear chromatin was most significant in follicular carcinomas followed by follicular adenomas. It was less obvious in the benign conditions, but more prominent in Hashimoto's thyroiditis compared with a goitre. It was also not obvious in a papillary carcinoma. Of the quantitative variables, all measurements were greatest in PCa > FCa > FAd > NCG = HT. When differentiating follicular adenoma from follicular carcinoma the qualitative variables of significance were the presence or absence of nucleoli, the chromatin pattern and regularity/irregularity of nuclear membrane; the nuclear area was the most important feature among quantitative variables.     

Validation of nuclear texture, density, morphometry and tissue syntactic structure analysis as prognosticators of cervical carcinoma

OBJECTIVE: To evaluate the performance of karyometry and histometry in the prediction of survival, recurrence and response of early-stage invasive cervical carcinoma. STUDY DESIGN: Nuclear morphometry, chromatin texture and tissue architecture (characterized by syntactic structure analysis) were measured using a semiautomated image analysis system on 46 cases of Feulgen-stained tissue sections. The performance of the features was compared to that of clinical features, reported to be the best prognosticators until now, such as age, lympho-vascular permeation, histologic type, stage and grade. A K nearest neighbor classifier was used for classification. RESULTS: In the prediction of three-year survival, recurrence and response, syntactic structure analysis proved to be the best performer. Classification rates were, respectively, 100%, 94.4% and 94.5%. In all classifications, karyometric and histometric features outperformed clinical features. In general, the best performing features described differences in second-order population statistics (standard deviations). CONCLUSION: The results show that a quantitative analysis based on nuclear morphology, chromatin texture and histology can be considered an excellent aid in the prognosis of invasive cervical carcinoma. The measurements are not hampered by the need to undertake complete resections and are suited to daily practice when implemented in a semiautomated image analysis system.     

Effect of the special properties of monolayer cell preparations for automated cervical cytology on visual evaluation and classification. With an estimation of the number of cells required to be screened

Monolayer preparations used in cell image analysis show some peculiarities as compared with conventional cytologic smears, such as homogeneous distribution of cells, distinct appearance of cells and a reduced number of background elements. However, for use in gynecologic cytology, monolayer preparations must be accessible to visual examination and classification. To investigate the consequences of the special features of these preparations on the strategy of visual evaluation, we estimated the minimum number of cells needed for a diagnostic decision. Cell counts were made of gynecologic monolayer preparations from 50 women with no suspicion of malignancy and 50 women with invasive squamous-cell carcinoma of the uterine cervix and its precursors. Our results showed that the more serious the lesion, the lower the number of cells needed for a diagnostic decision. The highest mean values of numbers of cells needed for an effective diagnosis were estimated in cases of mild and moderate dysplasia (734 cells) and in non-suspicious cases (731 cells). The number of cells needed did not exceed 1,700 in any case. The false-negative and false-positive rates were 6% and 2%, respectively, including the cases of mild to moderate dysplasia.     

Marker features for malignancy in ectocervical cells. Statistical evaluation

Marker features for malignancy have recently been observed in ectocervical cells, even in cells that are visually normal in appearance. This study assessed the statistical significance of these marker features using a mixed-model nested-design analysis of variance (ANOVA). Features in blue intermediate cells from patients with normal cytology, moderate dysplasia, and severe dysplasia/carcinoma in situ, nonkeratinizing cells from patients with moderate dysplasia, severe dysplasia/carcinoma in situ, and invasive cancer, and dysplastic cells from areas of metaplasia from patients with moderate dysplasia, severe dysplasia/carcinoma in situ, and invasive cancer were tested. ANOVA clearly demonstrated that the marker features differentiate between cells of the same cell type originating from patients in different diagnostic categories. In every instance, the differences owing to the diagnostic category were statistically significantly greater than those caused by patient-to-patient variability. Although the discriminating marker features in the intermediate cells were almost exclusively spectral features reflecting staining differences, morphometric features were also marker features in the dysplastic cells.     

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.     

Clinical evaluation of the EA-rosette test in the early detection of cervical cancer

It was previously found that a negative EA-rosette test, showing EA-rosette-forming cells in a cervical cell suspension, excluded the presence of cells of invasive carcinoma (predictive value of 99.9%). This study on 2,462 patients confirmed the applicability of the EA-rosette test in screening for precancerous as well as cancerous lesions. In 98.6% of the cases of dysplasia, carcinoma in situ and invasive carcinoma, the cervical cell suspensions contained EA-rosette forming cells (the rosette test was positive). With a negative EA-rosette test, the probability of missing a specimen with class III cytology (mild/moderate dysplasia) was 1.4%, of missing one with class IV cytology (severe dysplasia/carcinoma in situ) was 0.8% and of missing one with class V cytology (invasive carcinoma) was 0.25%. The predictive value of a negative EA-rosette test was 98.6%. The false-negative rate for negative EA-rosette tests was 3.7% for invasive carcinoma, 17.5% for carcinoma in situ and severe dysplasia and 41.4% for mild to moderate dysplasia.     

Endocervical glandular neoplasia and its mimics in ThinPrep Pap tests. A descriptive study.

OBJECTIVE: To document the cytologic features of endocervical adenocarcinoma in situ (AIS) and invasive endocervical adenocarcinoma as observed in ThinPrep slides and to compare these features with those that have been described for conventional smears. STUDY DESIGN: Six cases of endocervical AIS and three cases of invasive adenocarcinoma were evaluated with respect to 3 low-power and 14 high-power features. All cases were biopsy proven. Glandular "look-alikes" (tubal metaplasia, n = 3; florid repair, n = 3; sampling of lower uterine segment, n = 1) were also examined. RESULTS: All cases of AIS contained dark groups and sheets at screening power. At higher power, nuclear detail was extremely well visualized. All cases had crowding, continuous depth of focus, variability of nuclear size and shape within groups, irregular nuclear membranes, uniformly stippled chromatin and at least occasional single atypical cells. Only one case lacked nucleoli. Traditional features (strips, feathering, rosettes and mitoses) were observed about as frequently as in conventional smears. Invasive lesions had many of the same features, with relatively more inflammation and lysed blood. Nonneoplastic look-alikes could be distinguished from neoplasms using traditional criteria. CONCLUSION: In this small study, AIS and invasive endocervical adenocarcinoma maintained the features previously described for conventional smears. Improved visualization of nuclear detail may allow the application of additional criteria, such as irregular nuclear membranes and the more consistent presence of nucleoli, for distinguishing glandular neoplasms from their look-alikes.     

Can abundant colloid exclude oncocytic (Hürthle cell) carcinoma in thyroid fine needle aspiration? Cytohistological correlation of 127 oncocytic (Hürthle cell) lesions

Objective:  The objective of the present study was to find out whether the presence or absence of certain cytological features can exclude oncocytic (Hürthle cell) carcinoma in thyroid fine needle aspiration (FNA) to minimize unnecessary surgery. Methods:  Over a 17‐year period, 127 hypercellular, oncocyte‐exclusive, lymphocyte‐absent aspirates obtained via ultrasound‐guided FNA with on‐site assessment had histology slides for review. The presence or absence of six cytological features (microfollicular arrangement, discohesive single cells, small cell dysplasia, large cell dysplasia, transgressing blood vessels and colloid) and one histological feature (macrofollicular component) were determined for each case independently by two cytopathologists. Results:  Histology showed 12 (9.4%) cases of Hashimoto thyroiditis, 23 (18.1%) oncocytic adenomatoid nodules in nodular goitre, 66 (52.0%) oncocytic adenomas and 26 (20.5%) oncocytic carcinomas (13 minimally invasive without angioinvasion, six minimally invasive with angioinvasion, seven widely invasive). Histologically, a macrofollicular component was present in seven of 26 (26.9%) oncocytic carcinomas, including one case with abundant thin colloid. A microfollicular arrangement, discohesive single cells, small cell dysplasia, large cell dysplasia and transgressing vessels were present in oncocytic carcinoma, oncocytic adenoma and oncocytic adenomatoid nodules in nodular goitre. Conclusions:  A macrofollicular component is frequently present in oncocytic carcinoma, oncocytic adenoma and oncocytic adenomatoid nodules in nodular goitre. None of the cytological features studied, including abundant colloid, can exclude oncocytic carcinoma. Oncocytic carcinoma can only be excluded by thorough histological examination of thyroidectomy specimens. A molecular marker is needed to triage oncocytic lesions in thyroid FNA.     

A Pilot Study Combining a GC-Sensor Device with a Statistical Model for the Identification of Bladder Cancer from Urine Headspace

There is a need to reduce the number of cystoscopies on patients with haematuria. Presently there are no reliable biomarkers to screen for bladder cancer. In this paper, we evaluate a new simple in–house fabricated, GC-sensor device in the diagnosis of bladder cancer based on volatiles. Sensor outputs from 98 urine samples were used to build and test diagnostic models. Samples were taken from 24 patients with transitional (urothelial) cell carcinoma (age 27-91 years, median 71 years) and 74 controls presenting with urological symptoms, but without a urological malignancy (age 29-86 years, median 64 years); results were analysed using two statistical approaches to assess the robustness of the methodology. A two-group linear discriminant analysis method using a total of 9 time points (which equates to 9 biomarkers) correctly assigned 24/24 (100%) of cancer cases and 70/74 (94.6%) controls. Under leave-one-out cross-validation 23/24 (95.8%) of cancer cases were correctly predicted with 69/74 (93.2%) of controls. For partial least squares discriminant analysis, the correct leave-one-out cross-validation prediction values were 95.8% (cancer cases) and 94.6% (controls). These data are an improvement on those reported by other groups studying headspace gases and also superior to current clinical techniques. This new device shows potential for the diagnosis of bladder cancer, but the data must be reproduced in a larger study.