DL4ALL: Multi-Task Cross-Dataset Transfer Learning for Acute Lymphoblastic Leukemia Detection

Methods for the detection of Acute Lymphoblastic (or Lymphocytic) Leukemia (ALL) are increasingly considering Deep Learning (DL) due to its high accuracy in several fields, including medical imaging. In most cases, such methods use transfer learning techniques compensate limited availability labeled data. However, current ALL traditional learning, which requires models be fully trained on source domain, then fine-tuned target with drawback possibly overfitting domain and reducing generalization capability domain. To overcome this increase classification that can obtained using paper we propose our method named “Deep Leukemia” (DL4ALL), a novel multi-task DL model detection, cross-dataset approach. The adapts an existing into problem, trains it procedures consider both databases at same time, interleaving batches from two domains even when they significantly different. proposed DL4ALL represents first work literature procedure detection. Results publicly-available database confirm validity approach, achieves higher detecting respect methods, not manual labels