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

Abstract

Background The Paris System for Urine Cytopathology (the Paris System) has succeeded in making the analysis of liquid‐based urine preparations more reproducible. Any algorithm seeking to automate this system must accurately estimate the nuclear‐to‐cytoplasmic (N:C) ratio and produce a qualitative “atypia score.” The authors propose a hybrid deep‐learning and morphometric model that reliably automates the Paris System. Methods Whole‐slide images (WSI) of liquid‐based urine cytology specimens were extracted from 51 negative, 60 atypical, 52 suspicious, and 54 positive cases. Morphometric algorithms were applied to decompose images to their component parts; and statistics, including the NC ratio, were tabulated using segmentation algorithms to create organized data structures, dubbed rich information matrices (RIMs). These RIM objects were enhanced using deep‐learning algorithms to include qualitative measures. The augmented RIM objects were then used to reconstruct WSIs with filtering criteria and to generate pancellular statistical information. Results The described system was used to calculate the N:C ratio for all cells, generate object classifications (atypical urothelial cell, squamous cell, crystal, etc), filter the original WSI to remove unwanted objects, rearrange the WSI to an efficient, condensed‐grid format, and generate pancellular statistics containing quantitative/qualitative data for every cell in a WSI. In addition to developing novel techniques for managing WSIs, a system capable of automatically tabulating the Paris System criteria also was generated. Conclusions A hybrid deep‐learning and morphometric algorithm was developed for the analysis of urine cytology specimens that could reliably automate the Paris System and provide many avenues for increasing the efficiency of digital screening for urine WSIs and other cytology preparations.

Performance of an artificial intelligence algorithm for reporting urine cytopathology

“…Indeed, Pantazopoulos et al reported a sensitivity of 94.5% and a specificity of 100% for detecting urothelial carcinoma in urine cytology specimens . Vaickus et al developed an equivalent hybrid deep learning and morphometric algorithm with which to analyze urine cytology specimens . However, they used a much smaller sample size (217 cases) of carefully selected cases, compared with the 2405 cases that were scanned for the current study.…”

“…To our knowledge to date, significant work applying artificial intelligence (AI) to digital pathology has largely involved histopathology . Several groups have started applying similar promising techniques to computational cytology, including urine cytology …”

Innovations: Brave new worlds

“…). His work explored the development of a hybrid deep‐learning and morphometric algorithm for the analysis of urine cytology specimens . This breakthrough article presents the framework for a system that can automate the Paris System for Urine Cytopathology reliably, with the potential to increase the efficiency of digital screening for urine WSI.…”

Mankind and the machine: A relationship of symbiosis or conflict?

“…Sanghvi et al used whole slide imaging (WSI) and then extracted subimages via open source image processing software and passed these data through a deep learning algorithm to detect urothelial cells. In their study, the authors believe they analyzed more cells in more specimens than all previous studies combined . In fact, a total of 1.9 million urothelial cells were analyzed from 2405 ThinPrep glass slides.…”

“…Although other imaging studies have produced higher sensitivities and specificities, the study by Sanghvi et al stands out because of its design. After the development data set (1615 cases), Sanghvi et al used a validation set of 790 consecutive urine specimens .…”