DeepSwarmNet: A Lightweight Deep Learning Model for Kidney Cancer Detection using CT Scans

Kidney cancer ranks among the most popular of urological cancer types. Catching it very early can make a real difference in treatment for survival. However, current diagnostic tools often miss certain minor or low-contrast abnormalities, especially within settings where advanced medical infrastructure is limited. In this research, we present DeepSwarmNet, a streamlined and effective deep-learning model with the aim of identifying kidney irregularities from CT scans with great precision and minimal computational requirements. At the heart of DeepSwarmNet is MobileNet, a compact neural network known for its rapid speed and strong feature extraction. It gets paired alongside Darwinian Particle Swarm Optimization (DPSO), which fine-tunes the model's hyperparameters more effectively than customary methods. They form, all together, a strong system; this system can identify cysts, tumours, and kidney stones with an accuracy of 97.6 %. It also has an improved ability to find small or obscure lesionsareas missed by standard imaging sometimes. Although results seem promising, the model needs testing across diverse patient groups and clinical systems.