Review
Copyright ©The Author(s) 2021.
World J Gastroenterol. Jun 7, 2021; 27(21): 2681-2709
Published online Jun 7, 2021. doi: 10.3748/wjg.v27.i21.2681
Table 1 Terms commonly used to describe artificial neural network structures
Term of ANN
Specific meaning
SizeNumber of neurons in the whole model
WidthNumber of neurons in the one layer
DepthNumber of layers
FrameworkArrangement methods of layers and neurons
CapabilityThe reflection contents of reality by the specific model
Table 2 Comparisons between feedforward and feedback neural network
Category
Feedforward neural network
Feedback neural network
Signal directionUnidirectionalUnidirectional/bidirectional
Operation timeShortLong
Feedback by output signalNoYes
Structural complexitySimpleComplicated
Memory timeShort-term or noneLong-term
Applied ranges in medicineWideLimited
ApplicationPerceptron network, back propagation network, radial basis function networkRecurrent neural network, Hopfieid network, Boltzmann machine
Table 3 Summary of studies concerning artificial neural network translation of basic achievements
Ref.
Disease
Type of data
ANN technique
Application direction
Outcome
Bao et al[167], 2020CRCMicrosatellite instability from TCGA databaseMulti-layer perceptron networkPrognostic prediction100% accuracy
Coppedè et al[189], 2015CRCDNA methylationAutoCMIdentification of connections between DNA methylation and CRCA strong connection between the low methylation levels ofthe five CRC genes
Liu et al[190], 2004CRCGene signature from GEDatasetsMulti-layer networkIdentification of latent marker genes of CRC91.94% accuracy
Berishvili et al[164], 2019CRCApproximately 4000 complexes for which the data on the target binding constantsCNNScreening filter for compoundprioritization73% Spearman rank correlation coefficient
Bloom et al[159], 2007CRC and GCMSMulti-layer networkDifferentiation between 6 common tumor types87% accuracy
Dadkhah et al[120], 2019colorectal polypGut microbiomeANN developed by Orange data mining toolEarly screening using collected stool> 75% accuracy
Chang et al[166], 2011CRCmiRNA profileNot mentionedExploration of association between specific miRNAs and clinicopathological features100% accuracy of miRNA panel
Chen et al[191], 2004CRCMS of serum protein patternMulti-layer perceptron networkDifferentiation between CRC and healthy control91% sensitivity; 93% specificity; 0.967 AUC
He et al[121], 2020CRC and gastroesophageal cancerGene signature from TCGA databaseMulti-layer networkDifferentiation between types of cancerCRC: 98.06% sensitivity; 96.88% precision. Gastroesophageal cancer: 94.89% sensitivity; 96.33% precision
Hu et al[192], 2015CRCGene signature from database of NCBI NLM NIHS-Kohonen neural networkPrediction of recurrence using gene expressions91% accuracy
Kurokawa et al[128], 2005CRCGene signature of nodal metastasisBNNPrediction of metastatic potential of CRC at stage I88.0% sensitivity; 86.6% specificity; 0.904 AUC
Liu et al[160], 2019Cancer cellSynthetic microscopic images from two publicly datasetsCNNAutomated counting of cancer cells-
Ronen et al[193], 2019CRCGene signature from TCGA databaseBNNStratification of CRC subtypes-
Bilsland et al[194], 2015CRCA virtual library of compoundsPerceptron networkScreen of Benzimidazolone inhibitors for CRC treatmentCB-20903630 was selected out for further validation of CRC treatment
Maniruzzaman et al[195], 2019CRCGene signature from patientsFuzzy neural networkCRC classification99.84% sensitivity; 99.75% specific; 99.81% accuracy; 0.9995 AUC
Inglese et al[196], 2017CRC3D MSDeep neural network (unsupervised)Identification of metabolic heterogeneityUp to 0.6991 Pearson's correlation
Shi et al[197], 2020CRC with liver metastasisCTANNPrediction of KRAS, NRAS and BRAF status0.95 AUC
Jiang et al[198], 2020GCTwo drug datasetsdeep neural networkPrediction of drug-disease associations17 kinds of drugs that were screened out by ANN had been confirmed as anti-tumor drugs
Bidaut et al[158], 2009Stomach stem cellStemness signaturePerceptron networkCharacterization of stem cells-
Jing et al[168], 2019Calibration of laboratory markersCA-724Radial basis function neural networkThe effects of geographic factors on CA-724CA724 reference values show spatial autocorrelation and regional variation
Xiao et al[122], 2018GCRNA-seqProbabilistic neural networks (semi- supervised)Diagnosis of cancer96.23% accuracy; 99.08% precision
Hang et al[144], 2018GCMSIMulti-layer perceptron networkPrognostic prediction0.81 AUC
Xuan et al[161], 2019GCLncRNA profileCNNPrediction of GC0.977 AUC
Joo et al[163], 2019GCPotential drugs from databasesCNNExploration of new drugs targetingANN-based model accurately predicts drug responsiveness as models previously reported
Liu et al[165], 2010GCMS from GC patientsSupervised neural networkEarly screening100% sensitivity; 75% specificity
Que et al[199], 2019GCMS from GC patients and clinicopathological parametersSingle-layer neural networkPrediction of long-term survival0.82 AUC
Li et al[200], 2021GCGene Expression Omnibus databaseANNDifferentiation between GC and healthy tissues0.946 AUC
Table 4 Summary of existing studies of artificial neural networks applied in inflammatory bowel disease
Ref.
Disease
Aim
Number of samples
ANN technique
Included variables
Outcome
Ahmed et al[169], 2017CDDiagnosis144 CD patients; 243 HC individualsBPNN103 variablesAccuracy 97.67%; sensitivity 96.07%; specificity 100%
Ananthakrishnan et al[154], 2017UC and CDPredicting treatment response to vedolizumab43 UC patients; 42 CD patientsvedoNetGut microbiomeAUC of CD 88.1%; AUC of UC 85.3%
Anekboon et al[201], 2014CDPredicting single nucleotide polymorphisms144 CD patients; 243 HC individualsMulti-layer perceptron network103 SNPsAccuracy 90.4%; sensitivity 87.5%; specificity 92.2%
Dong et al[173], 2019CDPredicting the risk of surgical intervention in Chinese patients83 patients with surgery; 83 patients without surgeryANN131 variablesAccuracy 90.89%; precision 46.83%; F1 score 0.5757
Fioravanti et al[202], 2018IBDClassification of metagenomics data222 IBD patients; 38 HC individualsCNNGut microbiota-
Hardalaç et al[203], 2015IBDPredicting the effect of azathioprine on mucosal healing129 IBD patientsBPNNAge, age at diagnosis, usage of other medications prior to azathioprine use, smoking, sex, UC-CDAccuracy 79.1%
Kirchberger-Tolstik et al[170], 2020UCDiagnosis227 Raman maps with 567500 spectraCNNImages of Raman spectroscopysensitivity of 78%; specificity 93%
Klein et al[204], 2017CDPredicting the clinical phenotype47 B1 patients; 19 B2 patients; 39 B3 patientsTwo-layer FNNH&EB1 vs B2 phenotype: sensitivity 81%, specificity 74%, accuracy 75%, AUC 0.74; B1 vs B3 phenotype: sensitivity 69%, specificity 76%, accuracy 70.5%, AUC 0.78; B2 vs B3 phenotype: sensitivity 67%, specificity 72.5%, accuracy 69%, AUC 0.72
Lamash et al[71], 2019CDVisualization and quantitative estimation of CD23 pediatric CD patientsCNNMRIDSCs of 75 ± 18%, 81 ± 8%, and 97 ± 2% for the lumen, wall, and background, respectively
Le et al[174], 2020IBDPredicting IBD and treatment status68 CD patients; 53 UC patients; 34 HC individualsNeural encoder-decoder (NED) networkGut microbiotaCD vs HC: 95.2% AUC; UC vs HC: 92.5% AUC; CD vs UC: 81.8% AUC
Morilla et al[175], 2019UCPredicting treatment responses to infliximab for patients with acute severe UC47 patients with acute severe ulcerative colitisDeep neural networkMicroRNA profiles84% accuracy; 0.82 AUC
Ozawa et al[112], 2019UCIdentification of endoscopic inflammation severity841 patientsCNN (GoogLeNet)Colonoscopy images0.86 AUC of Mayo 0; 0.98 AUC of Mayo 0-1
Peng et al[205], 2015IBDPredicting the frequency of relapse569 UC patients; 332 CD patientsANNMeteorological dataHigh accuracy in predicting the frequency of relapse of IBD (MSE = 0.009, MAPE = 17.1 %)
Shepherd et al[171], 2014IBDDifferential diagnosis between IBD and IBS59 UC patients; 42 CD patients; 34 IBS patients; 46 HC individualsMulti-layer perceptron neural networkGas chromatograph coupled to a metal oxide sensor in stool samples76% sensitivity, 88% specificity, 76% accuracy
Takayama et al[132], 2015UCPredicting treatment response to cytoapheresis90 UC patientsMulti-layer perceptron neural network13 clinical variables96% sensitivity; 97% sensitivity
Tong et al[172], 2020CD, UC and ITBDifferential diagnosis between CD, UC and ITB5128 UC patients; 875 CD patients; ITB 396 patientsCNNDifferential features of endoscopic images between UC, CD and ITBThe precisions/recalls of UC-CD-ITB when employing the CNN were 0.99/0.97, 0.87/0.83, and 0.52/0.81, respectively