A simple "expert system" for morphometric evaluation of cells in pleural effusions

Our plain-English summary of A simple "expert system" for morphometric evaluation of cells in pleural effusions, by M. Oberholzer, G. Feichter, P. Dalquen, et al (Pathology, Research and Practice, 1989).

The study

• Authors: M. Oberholzer, G. Feichter, P. Dalquen, R. Ettlin, H. Christen, M. Buser
• Journal: Pathology, Research and Practice
• Published: 1989;185(5):647-51
• Cited by: 7 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

Quantitation in diagnostic pathology serves to improve both diagnostic reliability and the prognosis of various malignant tumors. Unfortunately, flexible, user-friendly, reasonably priced and compatible morphometric systems are hardly available on the market. A modular Apple II PC-based system was therefore developed in-house. It records stereologic, planimetric and digital image analysis data and calculates secondary parameters. Detailed statistical analyses can be performed, after data transfer, on larger computers. A typical application of diagnostic morphometry is to determine whether, and with which probability a patient with an unclear histopathologic finding can be assigned to one or more groups of patients with known diseases and prognoses. To this end morphometric data of the patient under investigation are compared with other patient data stored in an expert system. This paper describes the concept of an expert system carried out on the Apple II PC system mentioned above and outlines the evaluation procedures. Invariant moments were used to describe nucleus textures. Results obtained from the cytological analysis of pleural effusions demonstrate that with this approach it is possible to differentiate between normal mesothelium, mesothelioma and metastases of adenocarcinomas.

In plain English

This paper, from the Department of Pathology of the University of Basel, Switzerland, described a simple computer system built in-house on an Apple II personal computer. The system measured cell nuclei in microscope images and compared one patient's measurements with stored data from patients with known diagnoses; the authors called it an "expert system". It was a methods paper and did not discuss asbestos exposure. As a worked example, the authors examined stained smears of pleural fluid from 11 control subjects, 7 patients with mesotheliomas, 13 patients with metastases of adenocarcinomas and 3 patients without a clear diagnosis. They measured 15-20 nuclei per person, describing nuclear texture with seven "invariant moments" (PHI 1 to PHI 7). For one patient with suspected mesothelioma, the values of PHI 1 to PHI 4 fell outside the control group's confidence limits, and the case best matched the mesotheliomas. For a patient labelled "4 h", the system gave an 85% probability of belonging to the mesothelioma group. The authors concluded that the approach could differentiate normal mesothelium, mesothelioma and adenocarcinoma metastases. They noted that an incorrect diagnosis is always possible for statistical reasons.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Methods paper describing a computer-based morphometric (image measurement) "expert system", illustrated with example cases from pleural effusion cytology
• Who was studied: Papanicolaou-stained smears of pleural aspirations from 11 control subjects, 7 patients with mesotheliomas, 13 patients with metastases of adenocarcinomas, and 3 patients without a clear pathological diagnosis; 15-20 cell nuclei evaluated per person (Department of Pathology, Basel, Switzerland; dates not given)
• Main results:
  • In a univariate comparison, a patient with suspected mesothelioma had values of PHI 1 to PHI 4 outside the confidence limits of the 11 control subjects (for example, PHI 1 of 0.85300 against control limits of 1.06758 to 1.40661). The case's values best matched those of malignant mesotheliomas.
  • In a multivariate comparison using PHI 1, PHI 3 and PHI 4, the system assigned patient "4 h" (suspected mesothelioma) to the mesothelioma group with a probability of 85%.
  • For patient "4 h", adding parameter PHI 2 did not improve the classification result.
• The authors' conclusion: The authors concluded that results from the cytological analysis of pleural effusions showed this approach could differentiate between normal mesothelium, mesothelioma and metastases of adenocarcinomas.
• Limitations the authors state:
  • The authors stated that results of such comparisons cannot be absolutely certain, and an incorrect diagnosis is always possible for statistical reasons.
  • The invariant moments PHI 5 and PHI 7 varied considerably and were regarded as unreliable.

Topics

The National Library of Medicine indexes this study under subjects our site covers: Pleural plaques and thickening.

Medical terms in this study, explained in our Lexicon: Pleural Effusion, Pathology, Prognosis, Mesothelium, Mesothelioma, Adenocarcinoma.

Related studies

• Pleural effusions and pneumothorax: Beyond simple plumbing: Expert opinions on knowledge gaps and essential next steps — R. Thomas, N.M. Rahman, et al. (2020)
• The distinction of adenocarcinoma from malignant mesothelioma in cell blocks of effusions: the role of routine mucin histochemistry and immunohistochemical assessment of carcinoembryonic antigen, keratin proteins, epithelial membrane antigen, and milk fat globule-derived antigen — E.S. Cibas, J.M. Corson, et al. (1987)
• Calretinin. A selective marker of normal and neoplastic mesothelial cells in serous effusions — M.C. Barberis, M. Faleri, et al. (1997)
• The PD-L1/PD-1 pathway promotes dysfunction, but not "exhaustion", in tumor-responding T cells from pleural effusions in lung cancer patients — H. Prado-Garcia, S. Romero-Garcia, et al. (2017)
• Quality control of immunocytochemical staining of effusions using a standardized method of cell processing — V. Kuenen-Boumeester, P. van Loenen, et al. (1996)
• Evaluation of pleural CYFRA 21-1 and carcinoembryonic antigen in the diagnosis of malignant pleural effusions — G. Salama, M. Miédougé, et al. (1998)

Browse more research in Pathology, Research and Practice, or return to medical research.