Many studies have examined the diagnostic concordance of whole slide imaging (WSI) and light microscopy (LM) for surgical pathology. In cytopathology, WSI use has been more limited, mainly because of ...technical issues. The aim of this study was to review the literature and determine the overall diagnostic concordance of WSI and LM in cytopathology. A systematic search of PubMed, Scopus, and the Cochrane Library was performed, with data extracted from the included articles. A quality assessment of studies was performed with a modified Quality Assessment of Diagnostic Accuracy Studies 2 tool. The primary outcome was concordance for the diagnoses rendered by WSI and LM as shown by the concordance rate with the original diagnosis, intra‐observer and interobserver concordance with the κ coefficient, or a percentage. Secondary outcomes included the time taken to reach a diagnosis and the quality and perception of WSI. A descriptive survey was provided. Among 1867 publications, a total of 19 studies (1%) were included. Overall, the concordance between WSI and the original diagnosis was 84.1%, the intra‐observer concordance between WSI and LM was 92.5% with a κ coefficient of 0.66, and the interobserver κ coefficient was 0.69. The time to reach a diagnosis was longer with WSI in all studies. The quality of WSI was good, but diagnostic confidence and cytologist preference were higher for LM. In conclusion, the concordance of WSI with LM is acceptable and in line with systematic reviews in surgical pathology. However, the time required for scanning and technical issues represent barriers to complete adoption. It is foreseeable that technical advances and rigorous validation study design will help to improve the diagnostic concordance of WSI with LM in cytopathology.
The diagnostic concordance of whole slide imaging with light microscopy in cytopathology as measured with the intra‐observer κ coefficient is comparable to that of surgical pathology. Some limitations due to the scan time and the time to render a diagnosis can be still barriers to the widespread adoption of whole slide imaging in cytopathology.
The mitotic count in breast carcinoma is an important prognostic marker. Unfortunately substantial inter- and intra-laboratory variation exists when pathologists manually count mitotic figures. ...Artificial intelligence (AI) coupled with whole slide imaging offers a potential solution to this problem. The aim of this study was to accordingly critique an AI tool developed to quantify mitotic figures in whole slide images of invasive breast ductal carcinoma.
A representative H&E slide from 320 breast invasive ductal carcinoma cases was scanned at 40x magnification. Ten expert pathologists from two academic medical centers labeled mitotic figures in whole slide images to train and validate an AI algorithm to detect and count mitoses. Thereafter, 24 readers of varying expertise were asked to count mitotic figures with and without AI support in 140 high-power fields derived from a separate dataset. Their accuracy and efficiency of performing these tasks were calculated and statistical comparisons performed.
For each experience level the accuracy, precision and sensitivity of counting mitoses by users improved with AI support. There were 21 readers (87.5%) that identified more mitoses using AI support and 13 reviewers (54.2%) that decreased the quantity of falsely flagged mitoses with AI. More time was spent on this task for most participants when not provided with AI support. AI assistance resulted in an overall time savings of 27.8%.
This study demonstrates that pathology end-users were more accurate and efficient at quantifying mitotic figures in digital images of invasive breast carcinoma with the aid of AI. Higher inter-pathologist agreement with AI assistance suggests that such algorithms can also help standardize practice. Not surprisingly, there is much enthusiasm in pathology regarding the prospect of using AI in routine practice to perform mundane tasks such as counting mitoses.
Limited studies on whole slide imaging (WSI) in surgical neuropathology reported a perceived limitation in the recognition of mitoses. This study analyzed and compared the inter- and intra-observer ...concordance for atypical meningioma, using glass slides and WSI. Two neuropathologists and two residents assessed the histopathological features of 35 meningiomas—originally diagnosed as atypical—in a representative glass slide and corresponding WSI. For each histological parameter and final diagnosis, we calculated the inter- and intra-observer concordance in the two viewing modes and the predictive accuracy on recurrence. The concordance rates for atypical meningioma on glass slides and on WSI were 54% and 60% among four observers and 63% and 74% between two neuropathologists. The inter-observer agreement was higher using WSI than with glass slides for all parameters, with the exception of high mitotic index. For all histological features, we found median intra-observer concordance of ≥ 79% and similar predictive accuracy for recurrence between the two viewing modes. The higher concordance for atypical meningioma using WSI than with glass slides and the similar predictive accuracy for recurrence in the two modalities suggest that atypical meningioma may be safely diagnosed using WSI.
Fine needle aspiration (FNA) has diagnostic and therapeutic value in the management of salivary gland cysts. Rendering an accurate diagnosis from an aspirated salivary gland cyst is challenging ...because of the broad differential diagnosis, possibility of sampling error, frequent hypocellularity of specimens, morphologic heterogeneity, and overlapping cytomorphology of many cystic entities. To date, there have been no comprehensive review articles providing a practical diagnostic approach to FNA of cystic lesions of salivary glands. This article reviews the cytopathology of salivary gland cysts employing 2017 World Health Organization terminology, addresses the accuracy of FNA, and presents The Milan System approach for reporting in cystic salivary gland cases. The utility of separating FNA specimens from salivary gland cysts, based upon the presence of mucin and admixed lymphocytes in cyst fluid is demonstrated. A reliable approach to interpreting FNA specimens from patients with cystic salivary gland lesions is essential to accurately determine which of these patients may require subsequent surgery.
Cytology effusions are often the only material available for diagnosing malignant pleural mesothelioma (MPM). However, the cytomorphological features alone are not always diagnostic, and cytology ...samples preclude an assessment for pleural tissue invasion. Accordingly, immunohistochemical, soluble, and molecular biomarkers have been developed. The aim of this study is to provide quantitative evidence regarding the diagnostic performance of novel biomarkers. To that end, a systematic literature review was performed of articles dealing with a loss of BRCA1‐associated protein 1 (BAP1), methylthioadenosine (MTAP), 5‐hydroxymethylcitosine (5‐hmC), glucose transporter 1 (GLUT1), insulin like‐growth factor II messenger RNA–binding protein 3 (IMP3), enhanced zeste homologue 2 (EZH2) staining, cyclin‐dependent kinase inhibitor 2A (CDKN2A) homozygous deletion (HD) testing, soluble mesothelin, and microRNA quantification in cytological samples for the diagnosis of MPM versus reactive atypical mesothelial cells. Sensitivity and specificity were extracted, and a meta‐analysis was performed. The quality of the studies was assessed with Quality Assessment of Diagnostic Accuracy Studies 2, and the quality of the evidence was evaluated with the Grading of Recommendations Assessment, Development, and Evaluation approach. Seventy‐one studies were included. BAP1 loss showed a sensitivity of 0.65 (confidence interval CI, 0.59‐0.71) and a specificity of 0.99 (CI, 0.93‐1.00). MTAP loss and p16 HD showed 100% specificity with sensitivities of 0.47 (CI, 0.38‐0.57) and 0.62 (CI, 0.53‐0.71), respectively. BAP1 loss and CDKN2A HD combined showed maximal specificity and a sensitivity of 0.83 (CI, 0.78‐0.89). GLUT1 and IMP3 showed sensitivities of 0.82 (CI, 0.70‐0.90) and 0.65 (CI, 0.41‐0.90), respectively, with comparable specificity. Mesothelin showed a sensitivity of 0.73 (CI, 0.68‐0.77) and a specificity of 0.90 (CI, 0.84‐0.93). In conclusion, some of the recently emerging biomarkers are close to 1.00 specificity. Their moderate sensitivity on their own, however, can be significantly improved by the use of 2 biomarkers, such as a combination of BAP1 and CDKN2A with fluorescence in situ hybridization or a combination of BAP1 and MTAP immunohistochemistry.
The diagnostic biomarkers BAP1 and MTAP by immunohistochemistry and CDKN2A (p16) by fluorescence in situ hybridization (FISH) are reliable markers with superb specificity (close to 1.00) for distinguishing malignant mesothelial cells from reactive mesothelial cells, but none of them are sufficient when used alone because of only moderate sensitivity (0.40‐0.60). The diagnostic power is markedly improved when 2 of the contemporary biomarkers are used, such as a combination of BAP1 loss by immunohistochemistry and CDKN2A (p16) homozygous deletion by FISH. A combined analysis of BAP1 and MTAP loss by immunohistochemistry is also promising.
Artificial intelligence (AI) has made impressive strides recently in interpreting complex images, thanks to improvements in deep learning techniques and increasing computational power. Researchers ...have started applying these advanced techniques to pathology images, although most efforts have been focused on histopathology. Cytopathology, however, remains the original field of pathology for which AI models for clinical use were successfully commercialized, to assist with automating Papanicolaou test screening. Recent AI efforts have focused on whole slide images of both gynecologic and non-gynecologic cytopathology. This review summarizes the literature and commercial landscape of AI as applied to cytopathology.
Several urine cytology classifications have been published in the literature. However, global acceptance in the field of urine cytology was only gained in 2016 after The Paris System for reporting ...urinary cytology was published. Despite this Paris System and its shifted focus toward the detection of high-grade urothelial carcinoma, the perceived weakness of low sensitivity and indeterminate diagnoses when screening with urine cytology remains unresolved. To overcome these shortcomings, investigators have studied a variety of emerging ancillary tests to augment urine cytology (UroVysion, ImmunoCyt/uCyte+, BTA-stat/TRAK, NMP22, SCD-A7, URO17, CellDetect, UroMark, UroSEEK). Furthermore, with the advent of digital cytology, the creation of artificial intelligence tools has created innovative opportunities to aid with urine cytology. This review article discusses the lessons learned in the evolution of reporting systems, explores the merit and challenges of ancillary tests, and calls attention to potential utility of applying artificial intelligence in urine cytology.
The expression of programmed death-ligand 1 (PD-L1) is an established prerequisite for the administration of checkpoint inhibitor therapy and is of prognostic value in several cancer types. Data ...concerning the potential effect of PD-L1 on the prognosis of thyroid carcinoma are limited. Therefore, this study aimed to provide a systematic review of the published data on this topic. The literature was reviewed to gather and quantify evidence on the prognostic role of PD-L1 in follicular epithelial derived thyroid carcinomas and determine its association with clinicopathological parameters. A meta-analysis was performed using the DerSimonian-Laird random-effects model. The quality of studies was evaluated with the Newcastle-Ottawa Scale and a modified GRADE approach used to rate the quality of evidence. Out of 445 papers, 18 were included and 15 provided adequate data for meta-analysis. The quality of evidence ranged from low to high. PD-L1 expression was significantly associated with a reduced disease-free survival (DFS) (RR 1.63, CI 1.04–2.56,
p
= 0.03, I
2
68%, τ
2
0.19 and HR 1.90, CI 1.33–2.70,
p
< 0.001, I
2
0%, τ
2
0.00); however, no association was found with the overall survival (OS). Furthermore, a significant association was found with respect to underlying chronic lymphocytic thyroiditis and
BRAFV600E
mutation status in papillary thyroid carcinomas. In the subgroup analysis, the association of PD-L1 and DFS remained strong in papillary thyroid carcinoma when compared with dedifferentiated thyroid carcinomas (anaplastic and poorly differentiated thyroid carcinomas) that failed to demonstrate a significant association with respect to PD-L1. These findings underscore the role of PD-L1 immunohistochemistry as a potential prognostic biomarker of disease recurrence in patients with papillary thyroid carcinoma.
One of the most relevant prognostic factors in cancer staging is the presence of lymph node (LN) metastasis. Evaluating lymph nodes for the presence of metastatic cancerous cells can be a lengthy, ...monotonous, and error-prone process. Owing to digital pathology, artificial intelligence (AI) applied to whole slide images (WSIs) of lymph nodes can be exploited for the automatic detection of metastatic tissue. The aim of this study was to review the literature regarding the implementation of AI as a tool for the detection of metastases in LNs in WSIs. A systematic literature search was conducted in PubMed and Embase databases. Studies involving the application of AI techniques to automatically analyze LN status were included. Of 4584 retrieved articles, 23 were included. Relevant articles were labeled into three categories based upon the accuracy of AI in evaluating LNs. Published data overall indicate that the application of AI in detecting LN metastases is promising and can be proficiently employed in daily pathology practice.
Certain features are helpful in the identification of gunshot entrance and exit wounds, such as the presence of muzzle imprints, peripheral tears, stippling, bone beveling, and wound border ...irregularity. Some cases are less straightforward and wounds can thus pose challenges to an emergency room doctor or forensic pathologist. In recent years, deep learning has shown promise in various automated medical image classification tasks. This study explores the feasibility of using a deep learning model to classify entry and exit gunshot wounds in digital color images. A collection of 2418 images of entrance and exit gunshot wounds were procured. Of these, 2028 entrance and 1314 exit wounds were cropped, focusing on the area around each gunshot wound. A ConvNext Tiny deep learning model was trained using the Fastai deep learning library, with a train/validation split ratio of 70/30, until a maximum validation accuracy of 92.6% was achieved. An additional 415 entrance and 293 exit wound images were collected for the test (holdout) set. The model achieved an accuracy of 87.99%, precision of 83.99%, recall of 87.71%, and F1-score 85.81% on the holdout set. Correctly classified were 88.19% of entrance wounds and 87.71% of exit wounds. The results are comparable to what a forensic pathologist can achieve without other morphologic cues. This study represents one of the first applications of artificial intelligence to the field of forensic pathology. This work demonstrates that deep learning models can discern entrance and exit gunshot wounds in digital images with high accuracy.