ResNet and ResNeXt-Powered Kidney Tumor Detection: A Robust Approach on a Subset of the KAUH Dataset

Abstract

Renal disorders, especially neoplasms, provide a global health risk. This study analyses 2,170 photographs from the 8,000-image KAUH dataset to identify kidney tumours. This study uses deep convolutional neural networks (CNNs) to identify kidney tumours by kind and stage. This project aims to create a reliable kidney tumour detection model.This study fuses two resilient CNN architectures, ResNet and ResNeXt, to improve the detection model. ResNeXt architecture solves gradient-related problems in deep learning models. The ResNeXt architecture allows multi-path networks with a cardinality choice. ResNeXt is ideal for feature extraction and learning due to its high scalability and gradient flow.Our research shows that our model is resilient, achieving 94% accuracy on a subset of 2,170 pictures from the KAUH dataset. This research uses ResNet-pretrained machine learning algorithms and ResNeXt’s gradient flow characteristics to improve healthcare outcomes.

Ähnliche Arbeiten