Early diagnosis of kidney stones can help identify very small stones, preventing them from growing larger or causing severe symptoms like intense pain, urinary tract infections, or kidney damage.
In the US, the process from initial suspicion to treatment for kidney stones, including imaging and consultation with a urologist, takes a few days to 2 weeks.
Clearly, there is a need for a computer vision-based system to accelerate the diagnosis process. As part of my recent research, the newly launched YOLOv10 model (by Ao Wang, Hui Chen, Lihao Liu, et al.) has been fine-tuned specifically using a data-centric approach. Reviewing several samples, I noticed that the size, shape, and location of the stones varied drastically.
This work also introduces the concept of Contextual ROI Sampling and explores techniques such as Random Salt/Pepper Noise, achieving a final mAP50 value of 94.1!
Here's the link to the article: https://lnkd.in/gBhsC336
#AI #ComputerVision #ObjectDetection #Innovation #DeepLearning #SmallObjectDetection #YOLO #MedicalDiagnosis #KidneyStoneDetection #Healthcare
Team Leader Manager
1mohttps://forms.gle/ELKf3ih174eSkZKXA