Enhancing diagnostic accuracy of multiple myeloma through ML-driven analysis of hematological slides: new dataset and identification model to support hematologists

Abstract

Multiple Myeloma (MM) is a hematological malignancy characterized by the clonal proliferation of plasma cells within the bone marrow. Diagnosing MM presents considerable challenges, involving the identification of plasma cells in cytology examinations on hematological slides. At present, this is still a time-consuming manual task and has high labor costs. These challenges have adverse implications, which rely heavily on medical professionals' expertise and experience. To tackle these challenges, we present an investigation using Artificial Intelligence, specifically a Machine Learning analysis of hematological slides with a Deep Neural Network (DNN), to support specialists during the process of diagnosing MM. In this sense, the contribution of this study is twofold: in addition to the trained model to diagnose MM, we also make available to the community a fully-curated hematological slide dataset with thousands of images of plasma cells. Taken together, the setup we established here is a framework that researchers and hospitals with limited resources can promptly use. Our contributions provide practical results that have been directly applied in the public health system in Brazil. Given the open-source nature of the project, we anticipate it will be used and extended to diagnose other malignancies.

Figure 1

(a) Bone marrow aspirate procedure; (b) Wright–Giemsa stained bone marrow aspirate smear slides from MM patients, analyzed by the oncohematology and immunophenotyping service of the Laboratory of Immunology and Molecular Biology; (c) Observation of stained slides in visible light optical microscopes and image capture by smartphone device; (d) Identification and labeling of detected cells.

Figure 2

Example of expected (ground truth) and predicted bounding boxes on plasma cells. Images (a), (c), and (e) show the regions of interest drawn by specialists. On the other hand, Images (b), (d), and (f) illustrates the bounding boxes defined by our AI-based approach.

Figure 3

Precision-recall curve: the approximation on the right-top corner emphasizes the important performance of our approach.

Figure 4

Bounding boxes illustrating a Region of Interest (ROI) defined by specialists (ground truth) in blue and a classification output in red. The black-hatched region constitutes the intersection area between both bounding boxes. The union area stands for all parts combining the colored boxes.

Figure 5

Results obtained by executing the grid search strategy to find (a) the best IOU threshold and (b) model confidence. The best results are represented by red dots.