Copyright ©The Author(s) 2019.
World J Gastroenterol. Feb 14, 2019; 25(6): 672-682
Published online Feb 14, 2019. doi: 10.3748/wjg.v25.i6.672
Table 1 Clinical application of artificial intelligence
1Detecting fatty liver disease and making risk stratificationDeep learning based on US100%100%100%[42]
2Detecting and distinguishing different focal liver lesionsDeep learning based on US97.2%98%95.7%[43]
3Evaluating liver steatosisDeep learning based on US96.3%100%88.2%[49]
4Evaluating chronic liver diseaseMachine learning algorithm based on SWE87.3%93.5%81.2%[12]
5Discriminating liver tumorsDCCA-MKL framework based on US90.41%93.56%86.89%[50]
6Predicting treatment responseMachine learning algorithm based on MRI78%62.5%82.1%[58]
Table 2 Liver leision detection
1Detecting liver new tumorsDeep learning based on CT86%[36]
2Predicting the primary origin of liver metastasisDeep learning based on CT56%[40]
3Detecting cirrhosis with liver capsulesDeep learning based on ultrasound96.8%[41]
4Detecting fatty liver disease and making risk stratificationDeep learning based on ultrasound100%[42]
5Detecting and distinguishing different focal liver lesions.Deep learning based on ultrasound97.2%[43]
6Detecting metastatic liver malignancyDeep learning based on PET/CT90.5%[44]
Table 3 Diffuse liver disease staging
1Deep learning based on MRIF4: 0.84; ≥ F3: 0.84; ≥ F2: 0.85[45]
2Deep learning based on CTF4: 0.73; ≥ F3: 0.76; ≥ F2: 0.74[46]
3Deep learning based on SWEF4: 0.97; ≥ F3: 0.98; ≥ F2: 0.85[32]