MRI IMAGE BASED DIAGNOSING AND CATERGORING CANCER USING DENSENET AND MOBILENET
Keywords:
Cancer, convolutional neural network (CNN), pretrained models, Bayesian optimization, transfer learning, learning without forgettin.Abstract
One out of six fatalities worldwide are brought about by malignant growth, making it the subsequent
driving reason for mortality universally. Regardless, the probability of endurance is significantly expanded by early
sickness distinguishing proof. We might have the option to look at additional cases quicker than expected assuming
we apply Artificial Intelligence (AI) to mechanize malignant growth recognizable proof. This study proposes the
utilization of artificial intelligence based deep learning models to arrange photographs of eight unique sorts of
malignant growth, including mind, cervical, lung, and bosom disease. Convolutional Neural Networks (CNN), one
of the deep learning models, was tried in this review against the characterization of photographs with harmful
attributes. MobileNet, VGGNet, and DenseNet are pre-prepared CNN adaptations that are utilized to move the data
they picked up utilizing the ImageNet dataset to recognize different sorts of disease cells. We decide the fitting
qualities for the hyperparameters by means of Bayesian advancement. In any case, models prepared on starting
datasets may lose their capacity to characterize because of move learning. In this way, we utilize Learning without
Forgetting (LwF), which protects the organization's innate abilities while preparing the organization only with new
undertaking information. The examination discoveries exhibit that the recommended move learning-based models
beat the cutting edge strategies right now concerning accuracy. Moreover, we exhibit that LwF performs better in
ordering recently prepared datasets as well as pristine datasets.
