DoğanayFatihAbutSerdarGökÖmer Faruk2026-10-232026-10-2320262147-79302149-6455https://doi.org/10.18586/msufbd.1877734https://search.trdizin.gov.tr/tr/yayin/detay/1440019https://hdl.handle.net/20.500.12604/11747Atelectasisthe partial or complete collapse of the lungrequires early diagnosis for effective treatment. This study investigates the effects of Region of Interest (ROI) cropping and transfer learning on the automatic classification of atelectasis from chest X-ray images. Four SqueezeNet-based models were developed using two image types (raw and cropped lung regions) and two training strategies (from-scratch and transfer learning). Models I and II were trained from scratch on raw and cropped imagesrespectivelywhile Models III and IV employed transfer learning on the same image sets. Performance was evaluated using standard classification metrics. Model IVcombining transfer learning and ROI-cropped imagesachieved the best performancewith 98.43% accuracy0.984 F1-scoreand 0.999 AUC. It also reached 0.9861 precision0.9828 sensitivityand 0.9859 specificity. ROI-based cropping consistently improved classification performance. Combining transfer learning with ROI-focused preprocessing significantly enhances atelectasis detection and shows promise as a reliable clinical decision-support approach.eninfo:eu-repo/semantics/openAccessSolunum SistemiRadyolojiNükleer TıpTıbbi GörüntülemeBilgisayar BilimleriYapay ZekaArticle1418494144001910.18586/msufbd.1877734