Year: 2013
Type: dissertation
Abstract: Systems pathology attempts to introduce more holistic approaches\ntowards pathology and attempts to integrate clinicopathological\ninformation with “-omics” technology. This doctorate researches\ntwo examples of a systems approach for pathology: (1) a\npersonalized patient output prediction for ovarian cancer and (2) an analytical\napproach differentiates between individual and collective tumour\ninvasion.\nDuring the personalized patient output prediction for ovarian\ncancer study, clinicopathological measurements and proteomic\nbiomarkers are analysed with a set of newly engineered bioinformatic\ntools. These tools are based upon feature selection, survival\nanalysis with Cox proportional hazards regression, and a novel\nMonte Carlo approach. Clinical and pathological data proves to\nhave highly significant information content, as expected; however,\nmolecular data has little information content alone, and is only significant\nwhen selected most-informative variables are placed in the\ncontext of the patient’s clinical and pathological measures. Furthermore,\nclassifiers based on support vector machines (SVMs) that\npredict one-year PFS and three-year OS with high accuracy, show\nhow the addition of carefully selected molecular measures to clinical\nand pathological knowledge can enable personalized prognosis\npredictions. Finally, the high-performance of these classifiers are\nvalidated on an additional data set.\nA second study, an analytical approach differentiates between\nindividual and collective tumour invasion, analyses a set of morphological\nmeasures. These morphological measurements are collected\nwith a newly developed process using automated imaging analysis\nfor data collection in combination with a Bayesian network analysis\nto probabilistically connect morphological variables with tumour invasion\nmodes. Between an individual and collective invasion mode,\ncell-cell contact is the most discriminating morphological feature.\nSmaller invading groups were typified by smoother cellular surfaces\nthan those invading collectively in larger groups. Interestingly,\nelongation was evident in all invading cell groups and was not a\nspecific feature of single cell invasion as a surrogate of epithelialmesenchymal\ntransition. In conclusion, the combination of automated\nimaging analysis and Bayesian network analysis provides\nan insight into morphological variables associated with transition\nof cancer cells between invasion modes. We show that only two\nmorphologically distinct modes of invasion exist.\nThe two studies performed in this thesis illustrate the potential of\na systems approach for pathology and illustrate the need of quantitative\napproaches in order to reveal the system behind pathology.
Author: Wim Verleyen
Language: English
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W1491338406