Computer Aided Diagnosis for Gastrointestinal Cancer Classification Using Hybrid Rice Optimization With Deep Learning

• Published in: IEEE access Vol. 11; pp. 76321 - 76329
• Main Authors: Mirza, Olfat M., Alsobhi, Aisha, Hasanin, Tawfiq, Ishak, Mohamad Khairi, Karim, Faten Khalid, Mostafa, Samih M.
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Artificial intelligence; Artificial neural networks; Biomedical imaging; Cancer; Classification; Classification algorithms; Clinical outcomes; Deep learning; Design optimization; Diagnosis; Digestive system; Endoscopy; Feature extraction; gastric cancer classification; Gastrointestinal diseases; Gastrointestinal tract; hybrid rice optimization; Hyperparameter optimization; hyperparameter tuning; Image classification; Image quality; Image reconstruction; Intestine; Machine learning; Medical diagnostic imaging; Medical imaging; Optimization; Signs and symptoms; Tuning
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2023.3297441

A gastrointestinal disease is a group of cancers which mainly affects the digestive system, along with the stomach, small intestine, oesophagus, rectum, and colon. Accurate classification and earlier diagnosis of this cancer are crucial for better patient outcomes. Deep learning (DL) algorithm, especially convolutional neural network (CNN), is trained to categorize endoscopic images of gastrointestinal tissue as either benign or malignant. Gastrointestinal cancer (GC) classification with DL is the process of using artificial intelligence (AI), especially the DL algorithm, to categorize endoscopic images of gastric tissue as benign or malignant. It could help clinicians to identify the earliest symptoms of cancer and make treatment decisions, resulting in improved patient outcomes. The study designs a new gastrointestinal disease Detection and Classification using Hybrid Rice Optimization with Deep Learning (GDDC-HRODL) model. The presented GDDC-HRODL model intends to classify the medical images for GC. To achieve this, the GDDC-HRODL technique initially preprocesses the input data to improve image quality. In addition, the presented GDDC-HRODL algorithm employs the HybridNet model to produce feature vectors and the hyperparameter tuning process takes place using the HRO algorithm. For GC classification purposes, the GDDC-HRODL technique uses an attention-based long short-term memory (ALSTM) model and its hyperparameters can be selected by the ant lion optimization (ALO) algorithm. The design of hyperparameter tuning processes helps to accomplish enhanced GC classification performance. The experimental analysis of the GDDC-HRODL algorithm on the medical dataset demonstrates its betterment in the GC classification process.