Automatic Segmentation of Kidneys and Kidney Tumors: The KiTS19 International Challenge

Abstract

Purpose: Clinicians rely on imaging features to calculate complexity of renal masses based on validated scoring systems. These scoring methods are labor-intensive and are subjected to interobserver variability. Artificial intelligence has been increasingly utilized by the medical community to solve such issues. However, developing reliable algorithms is usually time-consuming and costly. We created an international community-driven competition (KiTS19) to develop and identify the best system for automatic segmentation of kidneys and kidney tumors in contrast CT and report the results.Methods: A training and test set of CT scans that was manually annotated by trained individuals were generated from consecutive patients undergoing renal surgery for whom demographic, clinical and outcome data were available. The KiTS19 Challenge was a machine learning competition hosted on grand-challenge.org in conjunction with an international conference. Teams were given 3 months to develop their algorithm using a full-annotated training set of images and an unannotated test set was released for 2 weeks from which average Sørensen-Dice coefficient between kidney and tumor regions were calculated across all 90 test cases.Results: There were 100 valid submissions that were based on deep neural networks but there were differences in pre-processing strategies, architectural details, and training procedures. The winning team scored a 0.974 kidney Dice and a 0.851 tumor Dice resulting in 0.912 composite score. Automatic segmentation of the kidney by the participating teams performed comparably to expert manual segmentation but was less reliable when segmenting the tumor.Conclusion: Rapid advancement in automated semantic segmentation of kidney lesions is possible with relatively high accuracy when the data is released publicly, and participation is incentivized. We hope that our findings will encourage further research that would enable the potential of adopting AI into the medical field.

Point of care parenchymal volume analyses to estimate split renal function and predict functional outcomes after radical nephrectomy

“…The same software also led to more accurate estimations of NBGFR after RN in RCC patients when compared to subjective estimations of NBGFR made by expert urologic oncologists 6 . Recently, artificial intelligence algorithms have been developed to recognize distinct densities and morphological features of kidneys, kidney cysts, and kidney tumors on contrast-enhanced CT scans 28 . In polycystic kidney disease, such efforts have translated into very accurate automated estimations of total kidney volumes, which is virtually impossible using other methods 29 .…”

Integrating Artificial Intelligence Tools in the Clinical Research Setting: The Ovarian Cancer Use Case

“…In addition, segmentation models widely available today are generally limited to cancer types for which large and often public datasets are available (e.g., brain [25], liver [26], kidney [27]), which results in available models being of the more common cancers only. For other, more rare cancer types, solutions are far from being developed and/or commercialised, as access to images and high-quality annotations to train with is very limited.…”

Deep transfer learning from limited source for abdominal CT and MR image segmentation

“…The KITS (Kidney Tumor Segmentation Challenge) dataset was introduced as part of the grand challenge in 2019, with the objective of establishing a benchmark for automated kidney segmentation [26]. Eligibility for participation was extended to individuals who underwent surgery for a renal mass between January 2010 and July 2018 (n = 544).…”