Cancerous region detection in the prostate is performed by multiparametric magnetic resonance imaging using different imaging sequences. One of those modalities is dynamic contrast enhancement. The authors of this paper are testing possible modifications of workflow which use this modality for more accurate cancerous region detection in the prostate. The introduced changes are timestamp mapping in the segmentation step, proportionate Simple Linear Iterative Clustering region number to prostate region size in each slice, new definition of labels and new extracted features. Furthermore, experiments are performed for segmentation in a single timestamp only. The experiments test the effect of modification on curve classification by using XGBoost classification and flat neural network approaches. Lastly, the authors perform hyperparameter tuning of both approaches.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.