Journals / Fırat Üniversitesi Mühendislik Bilimleri Dergisi / 2026 / Cilt: 38 Sayı: 2

A Robust and Explainable Deep Learning Pipeline with Hybrid Image Enhancement for Sickle Cell Detection

Orak Hücre Tespiti için Hibrit Görüntü İyileştirme Tabanlı Sağlam ve Açıklanabilir Bir Derin Öğrenme Yaklaşımı

Pages
815–826
Publication Date
2026-09-30
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The results indicate that controlled contrast enhancement significantly improves discriminative performance; however, the selection of an appropriate operating point remains dependent on clinical priorities.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:In this study, a hybrid image enhancement and explainable deep learning framework is proposed for the automatic detection of morphological features associated with sickle cell disease in microscopic peripheral blood smear images.
📊
Key Findings & Data:The experimental dataset comprises 991 images and exhibits a pronounced class imbalance.
💡
Conclusion & Impact:The results indicate that controlled contrast enhancement significantly improves discriminative performance; however, the selection of an appropriate operating point remains dependent on clinical priorities.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Orak hücre hastalığı · mikroskopi görüntüleri · görüntü geliştirme

Abstract

In this study, a hybrid image enhancement and explainable deep learning framework is proposed for the automatic detection of morphological features associated with sickle cell disease in microscopic peripheral blood smear images. The experimental dataset comprises 991 images and exhibits a pronounced class imbalance. To address this limitation, data integrity was strengthened through systematic duplicate and near-duplicate removal, followed by the implementation of a leakage-safe data splitting strategy. Preprocessing strategies were comprehensively evaluated to mitigate variations in contrast, illumination, and staining conditions, with particular emphasis on a controlled contrast enhancement approach. In parallel, multiple methodological components, including augmentation ablation experiments, a three-phase dynamic learning strategy, validation-based threshold optimization, region-of-interest masking, and weighted ensemble techniques, were systematically examined. Model performance was assessed using standard evaluation metrics, including ROC-AUC, accuracy, sensitivity, specificity, and F1-score, and further analyzed through Grad-CAM-based explainability and quantitative attention analysis. The results indicate that controlled contrast enhancement significantly improves discriminative performance; however, the selection of an appropriate operating point remains dependent on clinical priorities. Overall, the proposed framework provides a more robust and clinically interpretable decision-support pipeline for binary classification of microscopic images, highlighting the importance of data leakage prevention, error distribution analysis, and explainability in biomedical image analysis.

Özet

Bu çalışmada, mikroskobik periferik kan yayma görüntülerinde orak hücre hastalığıyla ilişkili morfolojik özelliklerin otomatik olarak tespit edilmesi için, görüntü iyileştirme ve açıklanabilir derin öğrenmeyi birleştiren bir çerçeve önerilmiştir. Deney veri kümesi 991 görüntüden oluşmakta ve belirgin bir sınıf dengesizliği sergilemektedir. Bu sınırlamayı gidermek amacıyla, sistematik olarak yinelenen ve benzer verilerin temizlenmesi yoluyla veri bütünlüğü güçlendirilmiş, ardından sızıntıya karşı güvenli bir veri bölme stratejisi uygulanmıştır. Kontrast, aydınlatma ve boyama koşullarındaki farklılıkları azaltmak için ön işleme stratejileri kapsamlı bir şekilde değerlendirilmiş ve özellikle kontrollü kontrast geliştirme yaklaşımına ağırlık verilmiştir. Buna paralel olarak, veri artırma ablasyon deneyleri, üç aşamalı dinamik öğrenme stratejisi, doğrulama tabanlı eşik optimizasyonu, ilgi alanı maskeleme ve ağırlıklı topluluk teknikleri dahil olmak üzere çok sayıda metodolojik bileşen sistematik olarak incelenmiştir. Model performansı, ROC-AUC, doğruluk, duyarlılık, özgüllük ve F1 skoru dahil olmak üzere standart değerlendirme metrikleri kullanılarak değerlendirilmiş ve Grad-CAM tabanlı açıklanabilirlik ve kantitatif dikkat analizi yoluyla daha ayrıntılı olarak analiz edilmiştir. Sonuçlar, kontrollü kontrast geliştirmenin ayırt edici performansı önemli ölçüde iyileştirdiğini göstermektedir; ancak, uygun bir çalışma noktasının seçimi klinik önceliklere bağlı olmaya devam etmektedir. Genel olarak, önerilen çerçeve, mikroskobik görüntülerin ikili sınıflandırması için daha sağlam ve klinik olarak yorumlanabilir bir karar destek süreci sunmakta ve biyomedikal görüntü analizinde veri sızıntısının önlenmesi, hata dağılımı analizi ve açıklanabilirliğin önemini vurgulamaktadır.

🔖

Cite This Article

Aşar, T., Doğan, Y., *, T. A. (2026). A Robust and Explainable Deep Learning Pipeline with Hybrid Image Enhancement for Sickle Cell Detection. Fırat Üniversitesi Mühendislik Bilimleri Dergisi, Cilt: 38 Sayı: 2, 815–826. https://doi.org/10.35234/fumbd.1932194

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (31)

  1. National Heart, Lung, and Blood Institute. Sickle cell disease. NHLBI, 2025.
  2. Centers for Disease Control and Prevention. About sickle cell disease. 2025.
  3. Piel FB, Steinberg MH, Rees DC. Sickle cell disease. N Engl J Med. 2017;376(16):1561-1573.
  4. World Health Organization. Sickle-cell anaemia: Report by the Secretariat. World Health Organization, 2006.
  5. Chase ML, Drews R, Zumberg MS, Ellis LR, Reid EG, Gerds AT, Lee AI, Hobbs GS ve diğerleri. Consensus recommendations on peripheral blood smear review: defining curricular standards and fellow competency. Blood Adv. 2023;7(13):3244-3252.
  6. N KT, Prasad K, Singh BMK. Analysis of red blood cells from peripheral blood smear images for anemia detection: a methodological review. Med Biol Eng Comput. 2022;60:2445-2462.
  7. Patel J, Muralikrishna H, Chadaga K, Thalengala A, Sampathila N. Sickle cell disease detection in low-resource conditions using transfer-learning and contrastive-learning coupled with XAI. Sci Rep. 2026.
  8. Tomasi C, Manduchi R. Bilateral filtering for gray and color images. Sixth International Conference on Computer Vision; 1998. p. 839-846.
  9. Pizer SM, Amburn EP, Austin JD, Cromartie R, Geselowitz A, Greer T, Yankaskas BC, Muller KE ve diğerleri. Adaptive histogram equalization and its variations. Comput Vis Graph Image Process. 1987;39(3):355-368.
  10. Zuiderveld K. Contrast Limited Adaptive Histogram Equalization. Graphics Gems. Heckbert PS, editör. San Diego: Academic Press; 1994. p. 474-485.
  11. He K, Sun J, Tang X. Guided image filtering. IEEE Trans Pattern Anal Mach Intell. 2013;35(6):1397-1409.
  12. Gonzalez RC, Woods RE. Digital Image Processing. 3. baskı. Upper Saddle River: Pearson/Prentice Hall; 2008.
  13. Stacke K, Eilertsen G, Unger J, Lundstrom C. Measuring domain shift for deep learning in histopathology. IEEE J Biomed Health Inform. 2021;25(2):325-336.
  14. Tellez D, Litjens G, Bándi P, Bulten W, Bokhorst JM, Ciompi F, van der Laak J. Quantifying the effects of data augmentation and stain color normalization in convolutional neural networks for computational pathology. Med Image Anal. 2019;58:101544.
  15. Xu C, Sun Y, Zhang Y, Liu T, Wang X, Hu D, Li G ve diğerleri. Stain normalization of histopathological images based on deep learning: A review. Diagnostics. 2025;15(8):1032.
  16. Zech JR, Badgeley MA, Liu M, Costa AB, Titano JJ, Oermann EK. Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: a cross-sectional study. PLoS Med. 2018;15(11):e1002683.
  17. Geirhos R, Jacobsen JH, Michaelis C, Zemel R, Brendel W, Bethge M, Wichmann FA. Shortcut learning in deep neural networks. Nat Mach Intell. 2020;2(11):665-673.
  18. Kılıç, E. İ., & Kılıç, Ş. (2026). DeepInsight-Net: a CBAM-enhanced ResNet50 framework with focal loss for robust cervical cancer classification on multi-center datasets. Frontiers in medicine, 13, 1783634.
  19. Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D. Grad-CAM: Visual explanations from deep networks via gradient-based localization. Proceedings of the IEEE International Conference on Computer Vision; 2017. p. 618-626.
  20. Adebayo J, Gilmer J, Muelly M, Goodfellow I, Hardt M, Kim B. Sanity checks for saliency maps. Adv Neural Inf Process Syst. 2018;31.
  21. Tushabe, F., 2020. Sickle Cell Disease Dataset [online], Kaggle, Web adresi: https://www.kaggle.com/datasets/florencetushabe/sickle-cell-disease-dataset.
  22. Schisterman EF, Faraggi D, Reiser B, Hu J. Youden index and the optimal threshold for markers with mass at zero. Stat Med. 2008;27(2):297-315.
  23. Hassanzad M, Hajian-Tilaki K. Methods of determining optimal cut-point of diagnostic biomarkers with application of clinical data in ROC analysis: an update review. BMC Med Res Methodol. 2024;24(1):84.
  24. Tekin, H., Kılıç, Ş., & Doğan, Y. (2025). DiagNeXt: A Two-Stage Attention-Guided ConvNeXt Framework for Kidney Pathology Segmentation and Classification. Journal of Imaging, 11(12), 433.
  25. Perez L, Wang J. The effectiveness of data augmentation in image classification using deep learning. arXiv:1712.04621, 2017.
  26. Zhang H, Cisse M, Dauphin Y, Lopez-Paz D. mixup: Beyond empirical risk management. International Conference on Learning Representations; 2018.
  27. Qin J, Fang J, Zhang Q, Liu W, Wang X, Wang X. ResizeMix: Mixing data with preserved object information and true labels. arXiv:2012.11101, 2020.
  28. Sönmez, D. P., & Kılıç, Ş. (2026). PlantPathNet: a novel network based on cross-layer feature integration for plant disease classification. Frontiers in plant science, 17, 1783576.
  29. Goodfellow I, Bengio Y, Courville A. Deep Learning. Cambridge: MIT Press; 2016.
  30. He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition; 2016. p. 770-778.
  31. Kılıç Ş, Doğan Y (2026) A pruned and parameter-efficient Xception framework for skin cancer classification. PLOS ONE 21(3): e0341227.