Artificial Neural Networks as Decision Support Tools in Cytopathology: Past, Present, and Future

Abstract

OBJECTIVEThis study aims to analyze the role of artificial neural networks (ANNs) in cytopathology. More specifically, it aims to highlight the importance of employing ANNs in existing and future applications and in identifying unexplored or poorly explored research topics.STUDY DESIGNA systematic search was conducted in scientific databases for articles related to cytopathology and ANNs with respect to anatomical places of the human body where cytopathology is performed. For each anatomic system/organ, the major outcomes described in the scientific literature are presented and the most important aspects are highlighted.RESULTSThe vast majority of ANN applications are related to cervical cytopathology, specifically for the ANN-based, semiautomated commercial diagnostic system PAPNET. For cervical cytopathology, there is a plethora of studies relevant to the diagnostic accuracy; in addition, there are also efforts evaluating cost-effectiveness and applications on primary, secondary, or hybrid screening. For the rest of the anatomical sites, such as the gastrointestinal system, thyroid gland, urinary tract, and breast, there are significantly less efforts relevant to the application of ANNs. Additionally, there are still anatomical systems for which ANNs have never been applied on their cytological material.CONCLUSIONSCytopathology is an ideal discipline to apply ANNs. In general, diagnosis is performed by experts via the light microscope. However, this approach introduces subjectivity, because this is not a universal and objective measurement process. This has resulted in the existence of a gray zone between normal and pathological cases. From the analysis of related articles, it is obvious that there is a need to perform more thorough analyses, using extensive number of cases and particularly for the nonexplored organs. Efforts to apply such systems within the laboratory test environment are required for their future uptake.

Automating the Paris System for urine cytopathology—A hybrid deep‐learning and morphometric approach

“…Other research groups have performed morphometric and machine-learning analyses of urine cytology specimens supporting the validity of the NC ratio as a marker of atypia and demonstrating the utility and accuracy of neural networks in relation to urinary cytology images. [15][16][17][18][19][20][21] These studies mainly demonstrate the diagnostic and prognostic utility of morphometry and machine learning in isolation.…”

Image analysis and multi‐layer perceptron artificial neural networks for the discrimination between benign and malignant endometrial lesions

“…32,33 To achieve this, various techniques have been tested so far, involving either classical statistical models or more advanced techniques, such as neural networks. 23,[34][35][36][37] The present results are encouraging in the context of supplementing the performance of cytology in the diagnosis of endometrial cancer. [38][39][40][41] Nuclear morphological characteristics seem to play a central role in endometrial cytology, a finding which has been also observed in our previous studies.…”

“…Its implementation in endometrial cytological analysis appears to help overcoming the subjective nature of the standard evaluation, which is related to the lack of consensus in the measuring process, resulting in poor reproducibility, and significant interobserver and intraobserver variability . To achieve this, various techniques have been tested so far, involving either classical statistical models or more advanced techniques, such as neural networks …”

Application of Immunocytochemistry on Cell Block Sections for the Investigation of Thyroid Lesions

“…There are numerous mechanical learning techniques that have been applied to solve cytopathology problems [8,9,10]. For this study, we selected Classification and Regression Trees (CARTs) [8,11] due to their ability to produce relatively simple, understandable rules.…”

“…For this study, we selected Classification and Regression Trees (CARTs) [8,11] due to their ability to produce relatively simple, understandable rules. Such rules can be applied as a guideline in everyday practice, and thus help cytopathologists to choose the required immunocytochemistry staining for ancillary techniques during diagnosis and to be aware of the uncertainty of diagnosis in each step.…”