AN ENHANCED LUNG AND COLON CANCER DETECTION FRAMEWORK USING ADAPTIVE MULTI-DILATED EFFICIENTNET WITH SELF-ATTENTION NETWORK FROM HISTOPATHOLOGICAL IMAGES

The previous decades have been evolved with technical advancements in the medical and healthcare sector. During this time, the fundamental reasons for various deadly disorders have been addressed, and the powerful detection approaches have been implemented; thus, the usage of medical applications has been improved. However, there is still uncertainty in some of the disorder detection such as various types of cancer. Numerous experiments have been performed globally to decrease the mortality rate, to recognize it in early stages, or to detect the precautions. Among various kinds, the Lung and Colon Cancer (LCC) is very deadly and prevailing ones. These two cancers are accountable for 25% of other tumor cases globally. But, the timely recognition of the LCC can highly increase the survival rates. Presently, the deep learning has supported to detect the cancer cells automatically that is permitting the experts to examine numerous individuals in less amount of time without any assistance from specialists. Therefore, an “advanced deep learning model aided with an optimization algorithm” is implemented in this paper for precisely detecting the LCC using the histopathological images. Initially, the required histopathological images for the execution of the model are garnered from various standard data resources. The gathered images are then provided to the deep learning model called as Adaptive Multi-dilated EfficientNet with Self-Attention Network (AMEN-SAN) for disease detection. To enrich the detection functionality, the parameters in the AMEN-SAN are determined with the assistance of Enhanced Squid Game Optimizer (ESGO). Experimental evaluation is conducted to prove the efficiency of the suggested framework in recognizing the LCC more precisely.