Beyond the Baseline: Investigating Failed Segmentations within the Universal Lesion Segmentation Challenge ‘23
Keywords
Loading...
Authors
Issue Date
2024-07-21
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
Radiologists are already under significant pressure, which is expected to increase even further
in coming years due to the predicted rise in the global cancer burden [20]. Cancer patients
have to undergo many scans during treatment to track the progress of the cancer, which all
need to be analysed by radiologists [18]. Additionally, radiologists have to manually measure
longitudinal measurements of the tumours to follow clinical guidelines such as the Response
Evaluation Criteria In Solid Tumors (RECIST) [10]. Manually measuring these long- and
short-axis lesion measurements in CT scans is not only time-consuming but also has a large
inter-observer variability [2]. Automatic segmentation models can help relieve some of the
pressure by automatically segmenting the lesions from CT scans, which in turn can be used to
automatically compute the longitudinal axis measurements. By standardizing the segmentation
process, these models ensure consistency in measurements, thereby reducing inter-observer
variability. Consequently, automatic lesion segmentation models can alleviate the workload on
radiologists and aid in decreasing inter-observer variability [16].
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen