# Kidney Recognition in CT Using YOLOv3

**Authors:** Andréanne Lemay  
**arXiv:** [1910.01268](https://arxiv.org/abs/1910.01268)  
**Date:** 2019-10-03  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/1910.01268v1/

## Key points

- The paper evaluates the performance of the object detection model, YOLOv3, for kidney localization in both 2D and 3D from CT scans.
- YOLOv3 achieved a Dice score of 0.851 in 2D and 0.742 in 3D for kidney localization, indicating high accuracy.
- Another object detection model called SSD also demonstrated comparable scores on the test set.
- Both YOLOv3 and SSD can detect kidneys across a wide range of CT scans, including those from patients with various renal conditions.
- These models have potential applications in clinical settings where accurate kidney localization is crucial for diagnosis and treatment planning.
- The research provides valuable insights into the performance of YOLOv3 and SSD for kidney recognition in CT scans.
- The high accuracy achieved by these models highlights their potential as effective tools for organ localization tasks in medical imaging, contributing to improved patient care and diagnosis accuracy.

## Summary

The paper titled "Kidney Recognition in CT Using YOLOv3" addresses the challenges of organ localization in medical images due to their heterogeneity and the biological diversity among individuals. The main contribution of this study is to evaluate the performance of the object detection model, YOLOv3, for kidney localization in both 2D and 3D from CT scans. The results show that the YOLOv3 model achieved a Dice score of 0.851 in 2D and 0.742 in 3D for kidney localization. These scores indicate a high level of accuracy in detecting kidneys within medical images. Additionally, a similar state-of-the-art object detection model called SSD also demonstrated comparable scores on the test set. One significant finding of this study is that both YOLOv3 and SSD exhibited the ability to detect kidneys across a wide range of CT scans, including those from patients with various renal conditions. This suggests that these models have potential applications in clinical settings where accurate kidney localization is crucial for diagnosis and treatment planning. Overall, this research provides valuable insights into the performance of YOLOv3 and SSD for kidney recognition in CT scans. The high accuracy achieved by these models highlights their potential as effective tools for organ localization tasks in medical imaging, contributing to improved patient care and diagnosis accuracy.