Applications of AI in hematopathology research
| Summary | Available website address | Reference |
|---|---|---|
| This research integrated previous studies and added a new analysis of macrophages, including three-dimensional (3D) rendering. The focus was on immuno-oncology markers. | https://doi.org/10.3390/cancers14215318 | [14] |
| This research predicted the prognosis of FL using 120 different and independent artificial neural networks. The random number generator was used to generate the different overall survival predictions. | https://doi.org/10.3390/biomedinformatics2020017 | [31] |
| The overall survival of MCL was predicted using two strategies. First, a dimensionality reduction was based on the previously identified genes. Second, on immuno-oncology panels. The results were correlated with the Lymphoma/Leukemia Molecular Profiling Project (LLMPP) MCL35 proliferation assay. | https://doi.org/10.3390/healthcare10010155 | [32] |
| The overall survival and cell-of-origin molecular subtypes of DLBCL were predicted using artificial neural networks and a pan-cancer immune-oncology panel of 730 genes. | https://doi.org/10.3390/cancers13246384 | [33] |
| A neural network predicted (classified) several non-HL subtypes, including FL, MCL, DLBCL, BL, and MZL. All the genes of the array were used and a cancer transcriptome panel. The survival of a pan-cancer series was also performed. | https://doi.org/10.3390/make3030036 | [15] |
| This research used immunohistochemical analysis and AI to predict the survival of DLBCL, with a focus on the protein and gene expression of the colony stimulating factor 1 receptor (CSF1R). | https://doi.org/10.3390/hemato2020011 | [34] |
| This research analyzed the predictive value of caspase-8 (CASP8) and related markers [cleaved CASP3, cleaved poly(ADP-ribose) polymerase 1 (PARP1), BCL2, TP53, MDM2, MYC, Ki67, E2F1, CDK6, MYB, LMO2, and tumor necrosis factor-alpha-induced protein 8 (TNFAIP8)] in DLBCL using immunohistochemical stainings and several machine learning and artificial neural networks. | https://doi.org/10.3390/biomedinformatics1010003 | [35] |
| In DLBCL, several AI techniques were used for multidimensionality reduction to predict the overall survival of the patients. As a result, two markers were highlighted, programmed cell death1 ligand 1 (PD-L1/CD274) and IKAROS, which were later tested by immunohistochemistry in an independent series of cases. | https://doi.org/10.3390/ai2010008 | [36] |
| Using a MLP and 25 genes, the overall survival of DLBCL was predicted. In the final model, the prognosis was predicted using MYC, BCL2, and enolase 3 (ENO3). | http://mj-med-u-tokai.com/pdf/450107.pdf | [37] |
