Performance Analysis of Transfer Learning Framework for the Detection of Polyps in Colorectal Cancer
Year: 2024
Type: article
Abstract: Colorectal cancer (CRC) begins in the colon or rectum, gastrointestinal tract organs. It is a common cancer that causes many cancer deaths worldwide. CRC usually starts with a polyp, a benign growth that can become cancerous. CRC prevention, treatment, and control require early detection and treatment. In this study, we reviewed various, pertinent research based on CRC diagnostic techniques, colonoscopy, and the use of AI screening. We performed various quantitative and qualitative comparative analyses of diagnostic techniques based on numerous features. Colonoscopy and sigmoidoscopy allow doctors to examine the colon and rectum for abnormalities. Deep learning (DL) techniques in medical imaging and Artificial Intelligence (AI) have improved CRC diagnosis, particularly polyp detection. We discussed the present and possible use of AI, DL in CRC diagnosis. A sigmoidoscopy, a minimally invasive procedure, shows the potential in terms of reducing the number of incidences and mortality. Colonoscopy was the most invasive technique and possesses the risk of morbidity. The Markov model demonstrated that cost per life can be saved for a colonoscopy performed once in 10 years. Thus, colonoscopy certainly proves to be a golden standard with highest sensitivity with the capability of biopsy during diagnosis. The proposed pre-trained VGG19 model confirmed 97% accuracy in polyp detection when applied with the approach of Transfer Learning (TL). The model is not overfitting and is proven to be more accurate than the recommended Adenoma Detection Rate (ADR).
Institutions: Navrachana University, Florida Atlantic University, Broward Health
Language: English
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W7216976308
