• Mr.Chitikela Srinivas
• Mr.Gudikandula Suresh
• Mr.Vuyyuru Chaitanya Kumar

Deep learning models in medical image analysis often rely on large annotated datasets, which are challenging to obtain. Self-supervised representation learning (SSRL) offers a promising solution by learning meaningful representations from unlabeled data. This paper explores SSRL techniques to improve robustness and generalization in medical image analysis tasks. We investigate various SSRL methods, including contrastive learning and reconstruction-based approaches. Our experiments on multiple medical imaging datasets demonstrate that SSRL significantly enhances model performance, particularly in low-data regimes. The learned representations also exhibit improved robustness to distribution shifts and adversarial attacks. Furthermore, we show that SSRL can be used to learn domain-invariant features, enabling effective transfer learning across different modalities and institutions. Our results highlight the potential of SSRL to advance medical image analysis by reducing annotation burden and improving model reliability. The proposed approach has implications for developing accurate and trustworthy AI systems in healthcare.

Self-supervised learning, Representation learning, Medical image analysis, Deep learning, Robustness, Generalization, Contrastive learning, Transfer learning, Domain adaptation

"SELF-SUPERVISED REPRESENTATION LEARNING IMPROVING ROBUSTNESS AND GENERALIZATION IN MEDICAL IMAGE ANALYSIS TASKS", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 1, page no.d139-d143, January-2026, Available :http://www.jetir.org/papers/JETIR2601316.pdf

"SELF-SUPERVISED REPRESENTATION LEARNING IMPROVING ROBUSTNESS AND GENERALIZATION IN MEDICAL IMAGE ANALYSIS TASKS", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 1, page no. ppd139-d143, January-2026, Available at : http://www.jetir.org/papers/JETIR2601316.pdf