BD focalpoint slide profiler performance with atypical glandular cells on SurePath Papanicolaou smears

3cDe-Net: a cervical cancer cell detection network based on an improved backbone network and multiscale feature fusion

BACKGROUND

Cervical cancer cell detection is an essential means of cervical cancer screening. However, for thin-prep cytology test (TCT)-based images, the detection accuracies of traditional computer-aided detection algorithms are typically low due to the overlapping of cells with blurred cytoplasmic boundaries. Some typical deep learning-based detection methods, e.g., ResNets and Inception-V3, are not always efficient for cervical images due to the differences between cervical cancer cell images and natural images. As a result, these traditional networks are difficult to directly apply to the clinical practice of cervical cancer screening.

METHOD

We propose a cervical cancer cell detection network (3cDe-Net) based on an improved backbone network and multiscale feature fusion; the proposed network consists of the backbone network and a detection head. In the backbone network, a dilated convolution and a group convolution are introduced to improve the resolution and expression ability of the model. In the detection head, multiscale features are obtained based on a feature pyramid fusion network to ensure the accurate capture of small cells; then, based on the Faster region-based convolutional neural network (R-CNN), adaptive cervical cancer cell anchors are generated via unsupervised clustering. Furthermore, a new balanced L1-based loss function is defined, which reduces the unbalanced sample contribution loss.

RESULT

Baselines including ResNet-50, ResNet-101, Inception-v3, ResNet-152 and the feature concatenation network are used on two different datasets (the Data-T and Herlev datasets), and the final quantitative results show the effectiveness of the proposed dilated convolution ResNet (DC-ResNet) backbone network. Furthermore, experiments conducted on both datasets show that the proposed 3cDe-Net, based on the optimal anchors, the defined new loss function, and DC-ResNet, outperforms existing methods and achieves a mean average precision (mAP) of 50.4%. By performing a horizontal comparison of the cells on an image, the category and location information of cancer cells can be obtained concurrently.

CONCLUSION

The proposed 3cDe-Net can detect cancer cells and their locations on multicell pictures. The model directly processes and analyses samples at the picture level rather than at the cellular level, which is more efficient. In clinical settings, the mechanical workloads of doctors can be reduced, and their focus can be placed on higher-level review work.

USG guided fine needle aspiration cytology along with immunocytochemistry to diagnose primary malignant mixed mullerian tumors: A three-year study from a tertiary care center

Aims and Objectives:

To study the diagnostic utility of fine-needle aspiration cytology (FNAC) and immunocytochemistry in diagnosing primary malignant mixed Mullerian tumors (MMMT).

Materials and Methods:

A 3-year retrospective study carried out in a tertiary care hospital, which included all the gynecological patients who underwent USG-guided FNAC of their abdominopelvic masses.

Observations and Results:

Out of the 324 total cases, 05 (1.5%) were reported as primary malignant mixed Mullerian tumors. Out of these 05 cases, 03 were ovarian, 01 was uterine, and 01 involved both uterus and one-sided adnexa. The FNA smears from the masses revealed cytomorphological features of a biphasic neoplasm with elongated pleomorphic spindle cells and dispersed, focal attempted acinar pattern, thus indicating the possibility of MMMT. Immunocytochemistry was further carried out which showed both vimentin and cytokeratin positivity. The diagnosis was confirmed on subsequent biopsy and immunohistochemistry (without any histopathological-cytological discrepancy).

Conclusion:

Though the literature is replete in establishing a histo-pathological diagnosis of MMMT, the diagnosis on USG-guided FNAC has been rarely described. Emphasis should be made on the careful examination of small sarcomatous elements in smears. Utilization of cell block and immunocytochemistry with histopathological correlation should be done to avoid misdiagnosis.

False-negative Papanicolaou tests in women with biopsy-proven invasive endocervical adenocarcinoma/adenocarcinoma in situ: a retrospective analysis with assessment of interobserver agreement

INTRODUCTION

The objectives of our study were to identify factors contributing to false-negative Papanicolaou (Pap) tests in patients with endocervical adenocarcinoma (EA) or adenocarcinoma in situ (AIS), and to analyze the impact of educational instruction on interobserver agreement in these cases.

MATERIALS AND METHODS

False-negative Pap tests from patients with EA/AIS were reviewed by a consensus group and by 12 individual reviewers in 2 rounds, with an educational session on glandular neoplasia in Pap tests conducted between the 2 rounds.

RESULTS

Of 79 Pap tests from patients with EA/AIS, 57 (72.2%) were diagnosed as abnormal and 22 (27.8%) as negative. Of the 22 false-negative cases, 10 remained negative on consensus review, with false-negative diagnoses attributed to sampling variance. The other 12 cases were upgraded to epithelial abnormalities (including 8 to glandular lesions). The false-negative diagnoses were attributed to screening variance in 2 cases and interpretive variance in 10 cases. On individual review, abnormal cells were misinterpreted as reactive glandular cells or endometrial cells in 7 of 8 and 5 of 8 cases upgraded to glandular abnormalities, respectively. With education, the proportion of individual reviewers demonstrating at least moderate agreement with the consensus diagnosis (Cohen's kappa >0.4) increased from 33% (4 of 12) to 75% (9 of 12).

CONCLUSIONS

Sampling and interpretive variance each accounted for nearly one-half of the false-negative Pap tests, with underclassification as reactive glandular or endometrial cells the main source of the interpretive variances. Educational instruction significantly decreased the interpretive variance and interobserver variability in the diagnosis of glandular abnormalities.

HrHPV testing vs liquid-based cytology in cervical cancer screening among women aged 50 and older: a prospective study

Introduction

Evidence supports high-risk human papilloma virus (HPV) testing as the primary cervical cancer screening tool. However, benefits and harms should be carefully considered before replacing liquid-based cytology. In women age 50 and older, we evaluated how a commercially available DNA amplification HPV test compares with routine liquid-based cytology.

Methods

This prospective study included 4043 patients who had a cervical sample analyzed from September 2011 to September 2012. Patients were followed between 64 and 76 months (median: 70 months). Samples were analyzed using both liquid-based cytology and the Cobas 4800 HPV DNA test. We calculated the diagnostic efficacy of liquid-based cytology and HPV, with or without the opposite test as triage, using cervical intraepithelial neoplasia (CIN2+/CIN3+) as reference.

Results

The patients had a median age of 58 years, (range; 50–90). At baseline, HPV prevalence was 8.0%: a total of 3.7% of patients had atypical squamous cells of undetermined significance or worse (ASCUS+). Positive test results were 1.9% for liquid-based cytology with HPV triage and 3.0% for HPV with liquid-based cytology triage. The cumulative incidence of CIN3+ was 1.0% (40/4043). Sensitivities for CIN3+ were: liquid-based cytology 47.5% (31.5%–63.9%); liquid-based cytology with HPV triage 45.0% (29.3%–61.5%); HPV 90.0% (76.3%–97.2%); and HPV with liquid-based cytology triage 67.5% (50.9%–81.4%). Corresponding specificities were: liquid-based cytology 96.6% (96.0%–97.2%); liquid-based cytology with HPV triage 98.5% (98.0%–98.8%); HPV 92.8% (92.0%–93.6%); and HPV with liquid-based cytology triage 97.7% (97.2%–98.1%). At baseline, HPV testing overlooked five cases of gynecological cancer other than cervical cancer. Five cervical cancers were detected, two had been overlooked at baseline by liquid-based cytology and two by HPV testing

Conclusion

HPV screening using DNA amplification is a promising alternative to liquid-based cytology in women age 50 and older, but evaluation of alternative triage methods is warranted. The risk of overlooking cancers needs consideration when replacing liquid-based cytology with HPV testing as a method for primary screening.

Cytologic parameters predicting neoplasia in Papanicolaou smears with atypical glandular cells and histologic follow-up: a single-institution experience

INTRODUCTION

Studies investigating histologic follow-up of The 2001 Bethesda System diagnosis of atypical glandular cells (AGC) have focused on various screening methods, patient populations, and Papanicolaou preparations. Our aim was to report the histologic follow-up of AGC diagnoses from ThinPrep slides and evaluate specific cytologic features predicting benign or malignant follow-up results.

MATERIALS AND METHODS

A retrospective search identified liquid-based cervical cytology results interpreted as AGC. AGC diagnoses were stratified into four groups: atypical endometrial cells (AGC-EM); atypical endocervical cells (AGC-EC); AGC, favor neoplastic (AGC-FN); and AGC not otherwise specified (AGC-NOS). Evidence of disease was based on histologic follow-up (biopsy or resection specimen) with a diagnosis of cancer, complex endometrial hyperplasia, or high-grade squamous dysplasia. Available slides were blindly reviewed for specific cytologic features. Statistical analysis compared cytologic factors that would predict benign or malignant follow-up.

RESULTS

We interpreted 264 samples as AGC from 2005 through 2009. Of the 246 (93.2%) with follow-up histologic material, 60 (24.4%) were AGC-EM; 36 (14.6%) were AGC-EC; 28 (11.4%) were AGC-FN; and 122 (49.6%) were AGC-NOS. Neoplasia was diagnosed in 80 (32.5%). Neoplastic cases showed significantly increased numbers of single cells, cells in 3-dimensional clusters, engulfed neutrophils, nuclear enlargement, increased nuclear-to-cytoplasmic ratio, irregular nuclear borders, reniform nuclei, loss of polarity, and macronucleoli.

CONCLUSIONS

Cytologic parameters can be used to predict benign from neoplastic glandular lesions. Biopsy follow-up is necessary to correlate cytologic findings when AGC is diagnosed on a Papanicolaou smear.

Automated screening of conventional gynecological cytology smears: feasible and reliable

OBJECTIVES

We tested the ability of automated screening in processing conventional gynecological cytology smears and its efficacy in assessing sample adequacy and stratifying cases for risk of malignancy.

STUDY DESIGN

Cases were retrospectively selected, including unsatisfactory samples and slides with various sorts of artifacts. Automated screening was performed using the FocalPoint GS Imaging System (Becton Dickinson, Franklin Lakes, N.J., USA), with classification into five quintiles. For agreement purposes, cases were grouped into high risk for malignancy (quintiles 1 and 2) and low risk for malignancy (quintiles 3, 4 and 5).

RESULTS

A total of 120 cases (median age 37.5 years, range 18-85) were included in the study. Eighty-three cases (69.2%) could be successfully classified into quintiles. When divided by risk, 31 cases were placed in the high-risk and 52 in the low-risk group. The overall sensitivity and specificity of the automated analysis was 100 and 70.3%, respectively.

CONCLUSIONS

Automated analysis could analyze the majority of conventional smears, including one case previously screened as unsatisfactory. All malignant and high-grade lesions were correctly classified into the high-risk group. Broad use of this automation system could potentially decrease screening time and augment the efficacy in detecting precursor neoplastic changes in cervical cytology smears.