← Back to Publications List

A Deep Learning Approach for Pneumonia Detection Using Chest X-Ray Images

Students & Supervisors

Student Authors
Fahmida Rashid
Master of Science in Computer Science, FST
Labib Hasan Bayzid
Bachelor of Science in Computer Science & Engineering, FST
Supervisors
Firoz Ahmed
Professor, Faculty, FST
Tonny Shekha Kar
Lecturer, Faculty, FST

Abstract

Pneumonia,which isanacute respiratory infectiousdisease thataffectsoneor bothlungs inhumans, iscausedbybacteriacalledStreptococcuspneumoniae (alsoknownaspneumococcus). It isoneof themaincausesofdeathglobally, and Pneumonianeedstobediagnosedimmediatelyandcorrectly.Priorstudieshaveex ploreddeeplearningmodelssuchasCheXNet,VGG16,Xception, andResNet-RS. However,manyoftheseapproachesdonotsimultaneouslycapturebothspatialand contextual features,andoftenprovide limitedinterpretabilitythroughvisualization methodssuchasGrad-CAM.Toaddressthesechallenges,weproposeahybriddeep learningmodelthatcombinesResNet-18andtheSwinTransformerisusedtoattain higheraccuracyandreducetimedelay.ResNet-18captures important featuresand model sequenceswell fromdatasets. Italsoemploys thecomputational capacityof SwinTransformerinclassificationcomparedtothosefeatures,whichismoreflexible thanotherdeep learning structures.Gradient-weightedClassActivationMapping (Grad-CAM)isalsobeingusedtointerpretthemodels.Theproposedapproachgot anaccuracyof93.75%.Theabilitytofocusonlungregionsbyemphasizingmodel attentionhelpsmodeldecision-makinginterpretability.Thisresearchcontributesto thegrowingbodyofworkinAI-baseddiagnosticsolutions,emphasizinginterpretabil ityandclinical relevance.Thismodel isalightweightdeeplearningmodel thatwill mitigatetheselimitationsandcreateanefficientPneumoniadetectionsystem

Keywords

Deep Learning Pneumonia Detection

Publication Details

  • Type of Publication:
  • Conference Name: 3rd International Conference on Big Data, IoT and Machine Learning (BIM 2025)
  • Date of Conference: 25/09/2025 - 25/09/2025
  • Venue: Dhaka International University, Bangladesh