Segmentation and classification of immunohistochemically stained samples based on NordiQC Pan-CK assessment
Translated title
Segmentering og klassifikation af immunohistokemisk farvede prøver baseret på NordiQC Pan-CK vurdering
Authors
Term
4. term
Publication year
2017
Submitted on
2017-06-07
Pages
75
Abstract
Immunohistokemisk farvning er et brugbart redskab for patologer når de ønsker at undersøge vævsprøver for tilstedeværelsen af cancer. Dog hersker der mulighed for variabilitet på laboratorie- men også klinikerniveau. For at øge kvaliteten af de immunohistokemiske farvninger fungerer NordiQC som eksternt QA for laboratorier på verdensplan. Kommercielle systemer findes og studier har vist at CAD systemer kan anvendes inden for det patologiske arbejdsfelt. Dette projekt søger at afklare om kvantificering af NordiQCs QA vurderinger. Datasættet bestod af 39 prøver som dækkede over 4 forskellige klasser. Resultaterne fra valideringen vist at det var muligt at opnå en gennemsnitlig nøjagtighed mellem 60-85%.
Immunohistochemical stains are useful when examining tissue samples for the presence of cancer. This tool is however subject to variability at different levels. To promote staining quality NordiQC acts as a QA for laboratories worldwide. Commercial systems exist and studies have found that CAD systems can be applied within the field of pathology. This study was aimed toward investigating the plausibility of quantifying quality assessments using CAD. Data consisted of 39 sample covering 4 different quality assessments. Results from the validation showed that using intensity-based and staining colour specific features, it was possible to achieved average accuracy between 60-85%.
Keywords
Documents