Recent Advancement in Pancreatic Cancer Diagnosis Using Machine Learning-Based Methods: A Systematic Review

Painuli D., Bhardwaj S., KÖSE U.

International Conference on Advances and Applications of Artificial Intelligence and Machine Learning, ICAAAIML 2022, Noida, Hindistan, 16 - 17 Eylül 2022, cilt.1078 LNEE, ss.685-701, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası: 1078 LNEE
  • Doi Numarası: 10.1007/978-981-99-5974-7_55
  • Basıldığı Şehir: Noida
  • Basıldığı Ülke: Hindistan
  • Sayfa Sayıları: ss.685-701
  • Anahtar Kelimeler: Classification, Feature extraction, Machine learning, Pancreas cancer diagnosis, Segmentation
  • Süleyman Demirel Üniversitesi Adresli: Evet

Özet

Pancreatic cancer is a leading cause of mortality worldwide, which usually occurs due to atypical growth of some cells in pancreas or its nearby organ. Moreover, due to its location in abdomen and smaller size, it is quite difficult to detect and diagnose this cancer till tumor reached to certain size. Computer assistive methods like machine learning techniques may help in early diagnosis of pancreatic cancer due to its high computation efficiency and accuracy. Various machine learning-based techniques have been employed in different cancer type diagnosis, i.e., brain, breast, lung, liver cancer, etc. in past studies by researchers, however very limited work has been observed, dedicated to pancreas cancer detection. Therefore, it’s a need of hour to deliver a viable framework with established cancer avoidance, assessment and provision for initial stage detection of pancreas cancer for upkeeping of patients against mortality. The aim of this study is to explore, analyze, classify and discuss the current technological advancement in diagnosis of pancreas cancer using various machine learning methods. Several utilized machine learning techniques are grouped into alike groups and results are scrutinized using key performance metrics, i.e., accuracy, sensitivity, dice score and area under the curve and evaluated on benchmark datasets utilized by previous studies and concluded with witnessed challenges.