Myelodysplastic syndromes (MDS) are a collection of hematologic illnesses with varying clinical presentations that are all linked by impaired blood cell formation. Decisions on managing and treating children with MDS depend critically on early and correct diagnosis. This paper explores the integration of deep learning techniques with fluorescent in situ hybridization (FISH), flow cytometry, and karyotype panel analysis to improve the detection and diagnosis of MDS in pediatric patients. We discuss the potential benefits, challenges, and prospects of utilizing deep learning algorithms in conjunction with these diagnostic modalities. In this paper we are proposing DeepMDS: Deep Learning Model for Myelodysplastic Syndromes Diagnosis. The proposed DeepMDS model, leveraging advanced deep learning techniques, exhibits superior performance in accurately detecting and diagnosing myelodysplastic syndromes (MDS) in children, surpassing alternative approaches. Its high accuracy and reliability make DeepMDS a promising tool that could revolutionize MDS diagnosis and improve patient outcomes in pediatric settings.