Research Article
Classification of Kidney Tumor Grading on Preoperative Computed Tomography Scans
@INPROCEEDINGS{10.1007/978-3-031-34586-9_6, author={Maryamalsadat Mahootiha and Hemin Ali Qadir and Jacob Bergsland and Ilangko Balasingham}, title={Classification of Kidney Tumor Grading on Preoperative Computed Tomography Scans}, proceedings={Pervasive Computing Technologies for Healthcare. 16th EAI International Conference, PervasiveHealth 2022, Thessaloniki, Greece, December 12-14, 2022, Proceedings}, proceedings_a={PERVASIVEHEALTH}, year={2023}, month={6}, keywords={Kidney cancer Renal cancer Deep neural networks Tumor grading Classification CT scan}, doi={10.1007/978-3-031-34586-9_6} }- Maryamalsadat Mahootiha
Hemin Ali Qadir
Jacob Bergsland
Ilangko Balasingham
Year: 2023
Classification of Kidney Tumor Grading on Preoperative Computed Tomography Scans
PERVASIVEHEALTH
Springer
DOI: 10.1007/978-3-031-34586-9_6
Abstract
Deep learning (DL) has proven itself as a powerful tool to capture patterns that human eyes may not be able to perceive when looking at high-dimensional data such as radiological data (volumetric data). For example, the classification or grading of kidney tumors in computed tomography (CT) volumes based on distinguishable patterns is a challenging task. Kidney tumor classification or grading is clinically useful information for patient management and better informing treatment decisions. In this paper, we propose a novel DL-based framework to automate the classification of kidney tumors based on the International Society of Urological Pathology (ISUP) renal tumor grading system in CT volumes. The framework comprises several pre-processing techniques and a three-dimensional (3D) DL-based classifier model. The classifier model is forced to pay particular attention to the tumor regions in the CT volumes so that it can better interpret the surface patterns of the tumor regions to attain performance improvement. The proposed framework achieves the following results on a public dataset of CT volumes of kidney cancer: sensitivity 85%, precision 84%. Code used in this publication is freely available at:https://github.com/Balasingham-AI-Group/Classification-Kidney-Tumor-ISUP-Grade.
