Explainable CAD System for Classification of Acute Lymphoblastic Leukemia Based on a Robust White Blood Cell Segmentation

SIMPLE SUMMARY: Leukemia is a type of cancer that affects white blood cells and can lead to serious health problems and death. Diagnosing leukemia is currently performed through a combination of morphological and molecular criteria, which can be time-consuming and, in some cases, unreliable. Computer-aided diagnosis (CAD) systems based on deep-learning methods have shown promise in improving diagnosis efficiency and accuracy. However, these systems suffer from the “black box problem,” which can lead to incorrect classifications. This research proposes a novel deep-learning approach with visual explainability for ALL diagnoses based on robust white blood cell nuclei segmentation to provide a highly reliable and interpretable classification. The aim is to develop a CAD system that can assist physicians in diagnosing leukemia more efficiently, potentially improving patient outcomes. The findings of this research may impact the research community by providing a more reliable and explainable deep-learning-based approach to blood disorder diagnosis. ABSTRACT: Leukemia is a significant health challenge, with high incidence and mortality rates. Computer-aided diagnosis (CAD) has emerged as a promising approach. However, deep-learning methods suffer from the “black box problem”, leading to unreliable diagnoses. This research proposes an Explainable AI (XAI) Leukemia classification method that addresses this issue by incorporating a robust White Blood Cell (WBC) nuclei segmentation as a hard attention mechanism. The segmentation of WBC is achieved by combining image processing and U-Net techniques, resulting in improved overall performance. The segmented images are fed into modified ResNet-50 models, where the MLP classifier, activation functions, and training scheme have been tested for leukemia subtype classification. Additionally, we add visual explainability and feature space analysis techniques to offer an interpretable classification. Our segmentation algorithm achieves an Intersection over Union (IoU) of 0.91, in six databases. Furthermore, the deep-learning classifier achieves an accuracy of 99.9% on testing. The Grad CAM methods and clustering space analysis confirm improved network focus when classifying segmented images compared to non-segmented images. Overall, the proposed visual explainable CAD system has the potential to assist physicians in diagnosing leukemia and improving patient outcomes.