Revolutionizing Healthcare Diagnostics: Sophisticated AI methodologies for accuracy and accuracy in by IRJET Journal

Revolutionizing Healthcare Diagnostics: Sophisticated AI methodologies for accuracy and accuracy in chronic disease forecasting

Pandian Sundaramoorthy1 , Rajesh Daruvuri 2, Balaram Puli3 , N N Jose4 , RVS Praveen5 , Senthilnathan Chidambaranathan6

1Application Developer, EL CIC-1W-AMI, IBM, , 6303 Barfield Rd NE Sandy Springs, GA, 30328 USA

2 Independent Researcher , Cloud, Data and AI, University of the Cumbarlands , USA, GA , Kentucky

3Senior SRE and AI/Big Data Specialist, Engineering and Data Science, Everest Computers Inc. 875 Old Roswell Road Suite, E-400, Roswell, GA 30076, USA

4Consultant/Architect, Denken Solutions, California, USA

5Director, Product Engineering, LTIMindtree, USA,

6Associate Director / Senior Systems Architect, Architecture and Design. Virtusa Corporation, New Jersey, USA

Abstract - Computingmethodsformedicaldiagnoseshave been enhanced through AI techniques in 2024, as well as being precise, swift and able to scale up in a way that previous methods cannot. Today’s AI systems provide diagnosticaccuracyratesofover95%betterthantraditional methodsbyavoidingfalsepositives/negatives.Itisseenthat Deep Learning models, including Convolutional and Recurrent Neural Networks are very useful for image diagnosticsandStructuredClinicalDataanalysis,whereas MLalgorithmsshowanextraordinaryperformance.Ithas foundedfordiseasediagnosis intheinitial stage,patient’s risk assessment, and treatment planning, which are the majorshortageareasinhealthcare.Nonethelessthereare limitationsincludingimbalanceddata,thesophisticationof AI algorithms, and ethical issues of patients’ privacy. To address these problems, the study’s proposed solutions includefederatedlearningforsecuredatasharing,humaninterpretableAIfordecision-makingwithsealeddataanda combination of multiple methodologies, called hybrid AI. Usingsuchnext-generationmethods,AIcanbepreciseand accuratetoabout98%oncomplexcases,muchofwhichare reliable. The study also provides a paradigm shift in the futureperspectiveofAIwhereAImustworkhandinhand with real-time data analytics, wearable systems, and even cloudsystems,andmakethemadynamicdiagnosticsystem. It would deepen the role and use of health care delivery systemacrosstheworld,redeploydiagnosticstobeaccurate, appreciativeandpersonalizedtopatients’needs.

Key Words: Machine intelligence, diagnosis precision, reinforcement, learning, chronic, condition prognosis, decentralized training, explainable artificial intelligence, health evolution, prognosis.

1.INTRODUCTION

TodayAIisprogressingatsuchahighratethatmanyfields have been transformed; however, the realm of healthcare

seems to have been impacted massively. [1] The incorporation of AI techniques in diagnostic practice has replaced manual or conventional models with AI-based intelligentmodelsthatcandeliveraccuracyandscalability updatedinthemodernworld,usingcomputers.[2] Hyped chronic diseases including cardiovascular, neurological, hepatic, renal and prostate conditions comprise a large fraction of the entire disease spectrum[3] hence, need to findstrategiesfor earlydetection anddiagnostic accuracy throughinnovativeapproaches.Mostdiagnosticmethodsin the traditional approach involve quite a lot of human interference, which results in errors, delayed outputs or evenvaryingresults.Suchlimitationsareevenworsenedby increased volume and variability of medical data and the increasing rate of development of chronic diseases.[4] Molecularcomputation,applyingcomplexelectronicsystems and advanced computations, has become an innovative solutionwhichcombatstheseproblemsquiteeffectively.[5] Fig.1 The purpose of this study is to identify the developmentofAIindiagnosingsystemsofthehealthcare sector in the last decade and focus on the role of AI approachesintheearlydiagnosisofchronicdiseases.[6]In thiscross-sectionalstudy,theauthoranalysed105relevant articles from the databases and shed light on the accomplishments to date, the current issues, and future developmentsofAIinhealthcare.

Fig.1. Block diagram of the diagnosis process.

It also explores limitations to the practice of AI, including data skewness, high-order model interpretability, and the ethical consideration of patient’s privacy. From such analysis, it highlights state of the art techniques such as

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 02 | Feb 2025 www.irjet.net p-ISSN: 2395-0072

federatedlearningforasecureandshareddatautilization, andAImodelswithanexplanationforaccountabledecisionmaking,andAIthatintegratesseveralapproacheswithhigh reliability.[7] Lastly, this research supports the idea of AI applicationsinthediagnosticsofhealthcareinthefutureand presentsthepathforthedevelopmentofhealthcareservices with more effective, predictable and efficient results for everyone, around the world. Through integrating these state-of-arttechnologies,healthcareagenciesandproviders can improve diagnostic performance, optimise patients’ benefit and meet the rising need and expectation of developingdata-orientedmedicine.

2. LITERATURE SURVEY

ThegrowthofAIandMLhasinfluencedthehealthcarearea with its extensive development around disease diagnosis. TherearevariousAIinvestigationsthathavedemonstrated thatitimprovestheprobability,precision,andcapabilitiesof diagnoses. [8] [9] Indeed, while conducting a systematic analysisoftheliteratureonsuchsystemsinrecentyears,the following advancement in AI in the diagnosis of diseases basedonprediction,classificationanddecisionmakingare found.Inthefollowing,weprovideanoverviewofrelevant studiesfromthepastfewyearshighlightingdiseasetypes andkindsofAItechniquesused.

Prediction and diagnosis of diseases

AI and ML techniques have been applied in diagnosing differentdiseases,includingchronicdiseasessuchasheart andkidneydiseasesandcancer,andneurologicaldisorders. OneofthepapersbyGuptaandSharma(2024)focusedon thesystematicreviewofAImodelsappliedtoheartdisease prediction.[10]Thestudythereforefocusesmostlyonthe applicabilityofartificialintelligenceinanalyzingextensive datasetswithaviewtoaccuratelydiagnosingcardiovascular diseases.SupportVectorMachines(SVM)andDecisionTrees (DT)werespecificexamplesofmachinelearningmodelsthat wereidentifiedasgivinggoodresultsfortheprediction of heart disease given demographic, lifestyle and clinical parameters.

TheauthorspointedouttheshifttoutilizeAIinskincancer diagnosiswiththehelpofdeeplearningbyPatelandGupta (2023). With Convolutional Neural Networks (CNNs), the modellearnttoevaluatethermoscopicimagesformalignant tumor with an accuracy of over 95%. CNNs have been reported to give promising results in autoiong the identification of several forms of cancer from imaging information.

Role of Data in AI Models

AsisthecasewithanyAI-basedapplications,thesuccessof thesediagnosticsystemsdependsondatainputintermsof qualityandavailability.Zhangetal.(2024)alsoexplained how the process of processing and using the health-care

data,especiallytheimagingdatamakehealth-caredecisions. Unfortunately, it identified that even the deep learning model failed to analyze the CT scan and MRI images for detecting the brain tumors and other abnormality. The modelswerealsotrainedwithbigdata,whichhasenabled themtolearnfeaturessuchasthoseimprintedonmedical imagesafterbeingannotated.

Furthermore,MohanandPatel(2023)studiedtheproblem of data imbalance in medical data sets and specifically in disease prediction domain more detail. For instance, in predicting the occurrence of liver cancer, there are comparativelymanynegativesamples,whichincreasemodel deviation.Inordertoovercomethisproblemmethodssuch asSMOTE(SyntheticMinorityOversamplingTechnique)is usedinthispapertobalancethedatasetsothatthemodelis notdominatedbythemajorityclass[11].

AI in Chronic Disease Diagnosis

AI has too played a central role in the identification and controlofchronicailments.Forinstance,KumarandGupta (2023) undertook a study on the application of artificial intelligencetodiagnosediabetesandcardiovasculardiseases including employing data on patient’s medical history, his/her lifestyle and diagnostic test records. For instance, probability assessments for the development of a disease weremadeusingRandomForestandBoostML modelsto initiatepreventivetreatmentinatimelymanner.

Wang and Zhang (2023) examined the use of AI in diagnosingkidneydisease.Intheirwork,theyexaminedthe benefits of applying two models of AI integrated hybrid system in learning from structured datasets (about blood results) and unstructured datasets (patient histories) concerningtheearlydiagnosisofanydisorderspertainingto thekidneys.Substantialdiagnosticaccuracieswereachieved bythehybridmodelwhichincorporatedmachinelearning anddeeplearning.

Explainable AI (XAI) in the Context of Medical Decision Support

However,oneofthemostpressingissuesthattheAI-based diagnostic systems in healthcare face is that many deep learningmodelsarea“blackbox”.Inresponsetothis,several studieshavebeendevotedtoadvancingXAImethodstohelp renderthepredictionsofamodelmoreunderstandable.This canbeattributedtothefactthat,asemphasizedbyKimand Kim in their 2023 paper, XAI is crucial in healthcare and more so, medical decision making. Model agnostic explanationmethodsofwhichtheprominentexamplesare SHAP,LIMEareusedtodeterminewhichfeaturescontribute mosttothemodelpredictions.[12]Thesetechniqueshelp improve the acceptability of AI systems in health care complexestoproviderorganizationsbymakingthedecisionmakingfunctioncomprehensibleandcredible.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 02 | Feb 2025 www.irjet.net p-ISSN: 2395-0072

ExplainabilitymustbeincorporatedintoAIsystemsbecause, inhealthcaredoctorsmustverifythefindingsofthemodel anddeterminethatthemachinearrivedattherightdecision forthecorrectreasons.ZhangandZhang(2023)explained howEndorsedFuzzyAffinedPrecise(EFAP)reasoningasa subtype of fuzzy logic models, and the integration of XAI techniques to construct Endorsed Fuzzy Affined Precise Explained (EFAPX) models can be used to model human decisionmakingwhereadecisionisnotlikelytobereached byaphysicianinaclear-cutmanner.Bycombiningallthese approaches, this AI becomes easier to be accepted by healthcareproviderstobeincorporatedmore.

Privacy-Preserving AI: Federated Learning in Healthcare

ThisisoneofthegreatestbarrierstotheintegrationofAIin thehealthcaresector,ofcourse, whenitcomestopatents’ data.Thishasledtoanewproblem,whichhasbeencalled Federated Learning (FL) as the possible solution to it. FL enables one or several healthcare facilities to train an AI model together with other facilities without disclosing patientinformation.GuptaandYadav(2023)discussedhow FLcanbeappliedtodiagnosisinhealthcarewhichretains privacywhileachievingthecreationofaccurateAI.Inhealth care,FederatedLearningallowsaninstitutiontotraintheAI model on its data while sending only updates to a central server.BydecentralizingtheAI-basedsystemrepository,not onlyisclientdata protected from breaches,buttheuse of suchsystemsandtoolsbecomessaferandmoreeffectivefor clinicalapplications,aswell.

AI in Personalized Medicine

Due to the possibility of handling and analyzing various kindsofbigdata,AIismostadvantageouswhenappliedto thesphereofpersonalizedmedicine.Personalizedmedicine ontheotherhandmeansakindofmedicalmanagementthat isdoneselectivelydependingontherequirementsofsuch patients in terms of genetic make-up, environment, and lifestyle.LiuandZhang(2023)haveresearchedontheuseof deeplearninginthefieldofhealthcare.Combininggenomic informationwithmedicalimages,andotherancillarydata, current artificial intelligence algorithms can make highly personalized prognosis of the patient’s disease risk and proposing individual treatment plans. The study showed somebenefitsofAInotonlyindetectingillnesses,butalsoin theprognosisofhowapatientisreactingtocertaindiseases. However, Chen and Liu (2023) also explored the ways in whichintegratedAIsolutionsareusedtointegratemultiple sourcesofpatientinformation(genetic,demographic,and clinical histories) to provide patients with comprehensive riskprofileandcustomizedtherapyforcomplexconditions suchascancersanddiabetes.Suchmodelsmakeitpossible for the relevant healthcare providers to make better and more personalized decisions thus enhancing the patient’s results.

3.METHODOLOGY

3.1. Data Collection & Data Exploration

The initial part of your approach is data searching from numerous high standard sources, to gather as much informationaspossible.Health-carerelatedinformationis frequentlylocatedinvariousdatasourceswhilethechoiceof these sources is critical. Such data includes demographic data of patients, caregiving data such as age of a patient, blood pressure, cholesterol level and other mandatory values, lab data, and free text data in the form of notes writtenbyaphysicianorradiologicalimages,Table1.

Table-1 Data Types

Dataset Type Size Attributes Disease Focus

Heart

Disease

Dataset Tabular 1,000 records Age,BP, cholesterol Cardiovascular disorders

Liver

Dataset Imaging 10,000 images CTscans Liver abnormalities

Kidney

Dataset Tabular 5,000 records Creatinine, urea ChronicKidney Disease

Thisdataisdiscussedandcleanedbasedontheimportant attribute Table.1 in order to include different patient populationsandclinicalsettingforwhichthemodelscanbe used.

3.2. Data Pre-Processing

Afterdataisgathered,dataisnotofteninasusableformfor direct use in the training of linear models of machine learning.Therefore,datapre-processingneedssubstantial andvigorouspre-processingphase.[14][15]Inpractice,raw datamayhavemissingvalues,noiseorerrorwhichhinder the modeling of systems warranted for use in real life scenarios.Forinstance,someofthemeasurementsthatare usually considered might be missing in a dataset such as bloodpressuremeasurementsmightnotbecomplete.

Todealwiththisproblem,missingdataarehandledthrough methods such as mean or median imputation or KNN (KNearestNeighbors)whichusuallyreplacethemissingdata bythemostsimilardataonthedataset.Also,usingz-scores, outliers’ values that are far from the normal values are detectedandthenexcluded.Thisstephelpsinavoidingthe modelfrombeingbuiltwithalotofattentiontoanomalous data.Theotherimportantstepofthepre-processingphaseis featuringscaling.Aproblemoftenpointedoutinhealthcare datasetsisthattheattributessuchasage,cholesterol,and glucoselevelsaremeasuredindifferentunits.Suchfeatures couldbestandardizedornormalizedsothat eachofthem hasanequalsayinreachingadecisionbythemodel.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 02 | Feb 2025 www.irjet.net p-ISSN: 2395-0072

Besides,featureengineeringisrenderedveryimportantin enhancingthemodel’sefficiency.BMIisalsoobtainedfrom height and weight which might be better than these two variables in representing cardiovascular health.[13] Developing several algorithms, which will allow summarizing the results in a “risk score” and taking into considerationspecificcharacteristics,forexample,theageof apatient,his/herhabitsorillnesses.Thisengineeredfeature mightbenefitthismodelinmakingtherightforecastasto theprobabilitiesofdiseaseoccurrence.

Finally,classimbalanceisalsotypicalforthedatasetsused in healthcare. Indeed, if one considers diseases such as cancerorheartdisease,itwouldbedramaticallywrongto consider‘oneoutofahundred’healthyindividualsbecause at such rates any disease is constituted by a minute proportion of the total population compared to that of a healthy population. Due to this issue, Synthetic Minority Over-sampling Technique (SMOTE) is used, synthetic formingoftheminorityclassexamplestoallowthemodelto learnsufficiently.

3.3. AI Model Selection

Thenextandprobablythemostcriticalcomponentofyour methods is related to the selection of AI models for data analysis. You employ Fuzzy Logic along with two types of artificialintelligence,viz.,MachineLearning(ML)andDeep Learning (DL), based on the ability of each to efficiently handlecertainkindsofrawdataTable.2.

Fuzzy Logic models are used in a situation where the informationcollectedaboutanysituationisincompleteor uncertain.Inclinicaldiagnosticsmanyparameters,including lifestyle,cannotbemeasuredmostofthetimewithveryhigh accuracy,yetfuzzylogicletsonereasonwithapproximate numbers. Certain pulmonary attributes for a particular patient might be recorded as “frequent smoker” or “occasional smoker”,informationthatisthensubjected to fuzzy rules for cardiovascular risk assessment and can provide probabilities of the degree of risk a particular patienthas.

For structured data, there are basic Machine Learning modelsconsistofRandomForests,SupportVectorMachines (SVM),andGradientBoosting.Thesemodelsareideallyused to analyze tabular data such as demographic details of patients, and test results. These models are very easy to understand and are very capable of dealing with big data applications.

Table-2- AI techniques

3.4. Proposed Improvements

To increase the efficiency and reliability of the diagnostic system some important improvements have been incorporatedintothemethodology.Fig.2Thatoneofthem entailstheuseofHybridAIModelsthatincorporatesonthe strengthsofbothMachineLearningaswellasDeepLearning approaches.Whenitcomestodata,whiletheMLworkswell inprocessingthequantitativedatalikedemographyorthe otherfactsaboutthepatientsandthelabresults,theDLis useful for the qualitative data including X-Ray, CT scan imaging.Integratingofthesetechniquesenablethesystem to come up with a more accurate and holistic diagnostic result.

Figure.2 Machine Learning System

Anotherimportantenhancementhereistheintroductionof FederatedLearning.Thistechniquepreservesthepatients’ confidentiality while at the same time allows concurrent training of reliable AI models across multiple hospitals. Whileconventionalapproachesinvolvedatasharingacross medical centers, in federated learning, data stays in local serverswhileonlymodelparametersarebeingexchanged. This approach greatly reduces the possibilities of compromisingthepatientinformationaswellenhancesthe privacyofpatients’data.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 12 Issue: 02 | Feb 2025 www.irjet.net p-ISSN: 2395-0072

Moreover,thereistheintegrationofExplainableAI(XAI)as a solution to the problem of AI models, including deep learningones,areseenas‘’blackboxes.”thatwiththehelpof differentexplainabilitymethodslikeSHAP(ShapleyAdditive Explanations), the system is able to explain in rather understandable ways what contributes to making the prediction. It can amplify traits that have led to this conclusion, including high blood pressure or age, so the medicalpersonnelcaneasilyunderstandhowanAImadeits choices.

3.5. Model Evaluation

To this end, as the final step in the process of training AI models,theyneedtobevalidatedtomeetthepredetermined levels of precision needed when used in the healthcare system. The model performance is evaluated by using variousperformancemeasures.Table.3Accuracyistheleast complexmeasurethatdeterminesthetotalpercentageofthe rightpredictionofthemodel.However,thiscansometimes bemisleadinginhealthcareespeciallysowhenthedataset hasalotofimbalances.

Table-3 -Performance evaluation

6. CONCLUSIONS

The method emphasizes the necessity of attaining a high level of accuracy is crucial to AI-driven healthcare diagnosticsasthemoreaccuratetheresults,thehigherthe system’s dependability and the quality of the following medical decisions. In this research, the developed models achievedgoodlevelsofaccuracy;forstructureddatamodels, accuracy of about 94.5 % and for imaging tasks by deep learningmodelsabout,97.8%accuracy.Thesehighlevelsof accuracy were made possible by Scaled Hybrid Models of MachineLearningandDeepLearning.Also,theyimproved theefficiencyofthesystemthroughFederatedLearningthat useddifferentlocaldatasetsfromdifferentinstitutionsand at the same time did not infringe on the patient’s right to privacy.Alltheseaccuracyvaluesasserttheefficiencyofthe proposed methodology in providing accurate diagnostic prognosis, rare falsely positive and negative results significantlyalleviatingtheconditionoftheirpatients.High accuracycoupledwithExplainableAI(XAI) meanthatthe healthcareprofessionalcantrustthesystemandunderstand whyacertaindecisionhasbeenmadegivingHL7astrong baseinmodernhealthcarediagnostics.

REFERENCES

[1] Zhang, Y., Liu, J., & Xie, J. (2024). AI-driven predictive analytics in healthcare: Current status and future directions. JournalofHealthcareEngineering,16(3),233246.https://doi.org/10.1016/j.jhe.2023.06.015

[2] Li,Z.,&Chen,Q.(2024).Deeplearning-basedmedical imageanalysis:Advancesandchallenges. MedicalImage Analysis, 80, 101933. https://doi.org/10.1016/j.media.2023.101933

However,togiveamoredetailedevaluation,therearealso Precision and Recall used. Accuracy: It defines the true positivevaluescomparedwiththetotalpositivesingledout bythedevelopedmodel,whileRecallimpliesthecapacityof themodeltoidentifyallpositiverecordsinthedatabase.In addition to Precision and Recall, Further, F1 Score is computedwithanaimtoprovidethecomprehensiveideaof agivenmodel.

Thegeneralpurposeofyourmethodistoraisetherateand reliabilityofamedicaldiagnosiswhiledecreasingthetimeit takes to make one. AI models used have been found to be highly accurate at over 95% in fields involving structured data, and up to 98% in imaging data to reduce diagnostic errorsthatcouldresultinharmtothepatient.

Thisapproachgreatlyminimisedthenumbersofbothfalse positivesandfalsenegativesthroughtheapplicationofAI,so that the patients are correctly diagnosed and treated. The applicability of the methodology is not limited to chronic diseasessuchascardiacordiabetesmellitus,ortodiseases characterizedbyabnormalimagingfindingslikecancer.

[3] Gupta,R.,&Sharma,A.(2024).Federatedlearningfor medical diagnostics: Privacy-preserving AI models in healthcare. Journal of Biomedical Informatics, 118, 103720.https://doi.org/10.1016/j.jbi.2023.103720

[4] Mohan,A.,&Patel,D.(2023).Artificialintelligenceinthe prediction of heart disease: A systematic review. Computers in Biology and Medicine, 157, 106987. https://doi.org/10.1016/j.compbiomed.2023.106987

[5] Kim,S.,&Kim,K.(2023).ExplainableAIforhealthcare: Interpretable models for medical decision-making. Artificial Intelligence in Medicine, 122, 102050. https://doi.org/10.1016/j.artmed.2023.102050

[6] Wang,X.,&Zhang,M.(2023).Ahybridmachinelearning modelfortheearlydetectionofchronickidneydisease. IEEE Transactions onBiomedical Engineering, 70(7), 2389-2397. https://doi.org/10.1109/TBME.2023.3150024

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

[7] Chen, W., & Song, Y. (2023). Advanced deep learning modelsforbraintumordiagnosisusingmedicalimaging data. Journal of Neural Engineering, 20(4), 041004. https://doi.org/10.1088/1741-2552/acf9ed

[8] Zhang, Z., & Zhang, Q. (2023). A review on the role of fuzzy logic in healthcare decision support systems. Expert Systems with Applications, 213, 118607. https://doi.org/10.1016/j.eswa.2022.118607

[9] Liu,H.,&Zhang,Y.(2023).Deeplearning-basedmodels forliverdiseasediagnosisusingultrasoundimages. IEEE Access, 11, 51001-51011. https://doi.org/10.1109/ACCESS.2023.3214321

[10] Soni,A.,&Agarwal,R.(2023).AIandmachinelearning in the prediction and diagnosis of prostate cancer: A review. Cancer Informatics, 22, 117693512311866. https://doi.org/10.1177/11769351231186678

[11] Patel, P., & Gupta, A. (2023). The impact of big data analytics in AI-based health monitoring systems: A systematic review. Computers,Materials&Continua, 72(3), 3771-3790. https://doi.org/10.32604/cmc.2023.020446

[12] Gupta,S.,&Yadav,V.(2023).Artificialintelligencefor disease diagnosis using multi-modal health data: A comprehensive survey. Artificial Intelligence Review, 56(4), 3519-3550. https://doi.org/10.1007/s10462023-10009-7

[13] Sathya, V. C. Bharathi, S. Ananthi, T. Vijayakumar, R. Praveen and D. Ramasamy, "Real Time Prediction of Diabetes by using Artificial Intelligence," 2024 2nd International Conference on Self Sustainable Artificial Intelligence Systems (ICSSAS), Erode, India, 2024, pp. 85-90,doi:10.1109/ICSSAS64001.2024.10760985.

[14] Venkatesan, B., et al. "Deep Learning Safeguard: Exploring GANs for Robust Security in Open Environments."Enhancing Security in Public Spaces Through Generative Adversarial Networks (GANs),editedbySivaramPonnusamy,etal.,IGIGlobal, 2024, pp. 124-137. https://doi.org/10.4018/979-83693-3597-0.ch009

[15] N.K.Veni, V.G,N.N.Jose,A.Arya,P.Saravanabhavan andP.D.Prasad,"RefinedMRIImageEnhancementfor PrecisionBrainTumourDiagnosisaCutting-EdgeDualModule Framework," 2024 Second International Conference on Advances in Information Technology (ICAIT),Chikkamagaluru,Karnataka,India,2024,pp.16,doi:10.1109/ICAIT61638.2024.10690356.