Enhancing Lymphoma Diagnosis: Transfer Learning with DenseNet201 for Subtype Classification

The classification of blood cancer cells is crucial for early diagnosis and effective treatment. This research study presents a methodology for classifying lymphoma subtypes of blood cancer using transfer learning with the DenseNet201 model. Accurate diagnosis and classification of lymphoma cells pose significant challenges due to their complex nature. This study proposes a comprehensive blood cancer type dataset comprising 13,500 images, categorized into three classes: lymph-mcl, lymph-fl, and lymph-cll. The dataset is systematically divided into training (10,800 images), validation (2,700 images), and testing sets (1,500 images). The DenseNet201 model is trained on this dataset to differentiate between the lymphoma subtypes. Upon training and evaluation, the proposed model demonstrates substantial performance metrics, achieving an accuracy of 98.53%, indicative of its robustness in accurately classifying lymphoma subtypes. Additionally, the model attains a high F1 score of 0.98, highlighting its balanced performance in terms of precision and recall. These metrics highlight the model’s effectiveness in distinguishing between lymphoma subtypes on the image dataset. The advancement in transfer learning-based classification as presented in this study holds significant potential for improving medical diagnosis and treatment planning in the field of oncology.