Lectures and seminars CBB seminar: Using spatial information from medical images to predict outcomes
Centre for bioinformatics and biostatistics (CBB) presents a seminar with Dr. Henrike Häbel.
Title
Using spatial information from medical images to predict outcomes
Speaker
Dr. Henrike Häbel, Statistician at MedStat, Department of Learning, Informatics, Management and Ethics, KI Solna.
Description of the seminar
Analysis of stained tissue images is commonly performed in diagnosis and prognosis, where immunohistochemistry (IHC) is a valuable tool for characterizing cells or biomarkers of interest within the tissue. The extracted characteristics may include summaries of nuclei intensity, vessel diameter, and stained area fraction. Usually, there are several IHC markers available, and the question arises which of them may be associated with the outcome (for example overall survival or cancer-specific survival). For the investigation of their prognostic strength, image-assisted survival analysis can be conducted, where IHC marker specific characteristics are included in a regression model along with other prognostic factors such as age, sex, and disease specific characteristics. However, such IHC characteristics usually do not describe the spatial distribution of the point patterns of nuclei.
In this seminar, I will discuss challenges and benefits of including summaries of the spatial distribution of nuclei as prognostic factors. In particular, I will propose how spatial indices can be derived from the so-called L-function on the example of patients with stage III melanoma.
Before the seminar, a limited number of sandwiches and drinks will be served on a “First come – first served” basis. Show up early and network!
The seminar is held in English.