Analysis Of Fusion Gene Expression And Research Of The Diagnosis Of Leukaemia With AI

BackgroundGenetic abnormalities have been known to play an important role in the pathogenesis of acute myeloid leukemia(AML).Common recurrent fusion genes have become well-established molecular markers for disease classification,risk stratification,and targeted therapies.Different research groups around the world have conducted large-scale analyses of the spectrum and incidence of fusion genes in AML.Considering the likely differences in ethnic background among study populations,we conducted this study to analyze the panoramic incidence of common fusion genes in Chinese patients with de novo AML.ObjectiveTo analyze the genetic landscape of multiple fusion genes in patients with de novo acute myeloid leukemia(AML)and investigate the characteristics of other laboratory parameters.MethodsThe results of multiple fusion genes from 4192 patients with de novo acute myeloid leukemia(AML)were retrospectively analyzed from 2016 to 2020.In addition,the immunophenotypical data and the mutational results from high-through put method were statistically investigated and correlated as well.Results1.Among the 52 targets,29 different types of fusion genes were detected in 1948 patients(46.47%)with AML,which demonstrated an " exponential distribution”.2.As the age increased,the number of patients with fusion gene increased first and then decreased gradually.The total incidence rate of fusion genes and MLL rearrangment in children were significantly higher than those in adults(69.18%vs 44.76%,15.35%vs 8.36%).3.The mutations involving FLT3 and RAS signaling pathway contributed most in patients with MLL rearrangment.4.No specific immunophenotypic characteristics were found in AML patients with MLL or NUP98 rearrangements.ConclusionsOur findings delineate the frequencies and distributions of common fusion genes in Chinese AML cases and confirm different incidences between age groups.The incidence of fusion genes in AML patients is different and correlates with other laboratory results.This information may be used to help further improve fusion gene screening panels for clinical applications.BackgroundChronic myeloid leukaemia(CML)is a clonal proliferative disorder of granulocytic lineage with specific morphological and genetic features.The increased immature granulocytes,granulocyte ratio and ’dwarf megakaryocytes(MKs)can give pathologist cue to make a preliminary diagnosis.However,the morphological examination is time-consuming and subject to individual bias because it relaies on optical observe.A reliable ponautomated counter has yet to be explored given the intrinsic complexity of bone marrow coments.ObjectiveTo explore a novel multi-class myeloids segmentation model and select a series of statistical features about multi-class of myeloids,then perform clinical predictions on CML.MethodsThe pre-trained CMLcGAN model was used to segment the myeloid cells in the whole slide images(WSIs)of the CML group and the control group.After feature extraction and selection,eight kinds of classifiers were used to make the clinical prediction,and the best-performing binary classifier was selected by triple cross validation.Results1.BCR-ABL was positive in 58 patients and BCR-ABL negative in 31 controls.All these sections are qualified.2.The 5-dimensional features include the number of MK,the number of myeloid cells,MK’s density,and the maximum and minimum of MK size.The tissue of CML displays a conspicuous increase in the number of MKs,the number of myeloid cells,and MKs’ density,equivalent to the hypercellularity and increased megakaryocytes in the bone marrow.In addition,the maximum and minimum of MKs are more sizeable and smaller.3.The best-performing binary classifier was the linear SVM classifier with the achieved AUC40 of 84.93% ± 0.011.ConclusionsThe proposed CMLcGAN and diagnostic prediction feature set appeared to deliver satisfactory performance in CML diagnosis.