Deep learning identifies Acute Promyelocytic Leukemia in bone marrow smears

Abstract

Background Acute promyelocytic leukemia (APL) is considered a hematologic emergency due to high risk of bleeding and fatal hemorrhages being a major cause of death. Despite lower death rates reported from clinical trials, patient registry data suggest an early death rate of 20%, especially for elderly and frail patients. Therefore, reliable diagnosis is required as treatment with differentiation-inducing agents leads to cure in the majority of patients. However, diagnosis commonly relies on cytomorphology and genetic confirmation of the pathognomonic t(15;17). Yet, the latter is more time consuming and in some regions unavailable. Methods In recent years, deep learning (DL) has been evaluated for medical image recognition showing outstanding capabilities in analyzing large amounts of image data and provides reliable classification results. We developed a multi-stage DL platform that automatically reads images of bone marrow smears, accurately segments cells, and subsequently predicts APL using image data only. We retrospectively identified 51 APL patients from previous multicenter trials and compared them to 1048 non-APL acute myeloid leukemia (AML) patients and 236 healthy bone marrow donor samples, respectively. Results Our DL platform segments bone marrow cells with a mean average precision and a mean average recall of both 0.97. Further, it achieves high accuracy in detecting APL by distinguishing between APL and non-APL AML as well as APL and healthy donors with an area under the receiver operating characteristic of 0.8575 and 0.9585, respectively, using visual image data only. Conclusions Our study underlines not only the feasibility of DL to detect distinct morphologies that accompany a cytogenetic aberration like t(15;17) in APL, but also shows the capability of DL to abstract information from a small medical data set, i. e. 51 APL patients, and infer correct predictions. This demonstrates the suitability of DL to assist in the diagnosis of rare cancer entities. As our DL platform predicts APL from bone marrow smear images alone, this may be used to diagnose APL in regions were molecular or cytogenetic subtyping is not routinely available and raise attention to suspected cases of APL for expert evaluation.

The use and future perspective of Artificial Intelligence—A survey among German surgeons

“…Congruent with the response pattern from the previous answers, the category "Improved diagnostic confidence" received the greatest level of support. This is in line with the above-mentioned assumption that AI applications are already being used to some extent in the clinic, particularly for diagnostic procedures (2, 3,5,6,28) and are therefore much more widely accepted among surgeons than other AI tools. However, survey participants also see great potential in AI to facilitate "More precise and minimally invasive surgical techniques, " which was the second most selected use case.…”

“…In pathology, Eckhardt et al developed an AI program that could distinguish between AML samples and healthy controls based on bone marrow images with a very high sensitivity and specificity. Moreover, with the help of this AI application it was possible to predict with high probability the presence of one of the most common mutations in AML based on cytomorphological features alone (5).…”

Detection of acute promyelocytic leukemia in peripheral blood and bone marrow with annotation-free deep learning

“…genetics, cytochemistry, flow cytometry). An equally important drawback of previous fully supervised models 7,9,13,16,17 is that they require hundreds of thousands of object-level cell annotations provided by human experts 18 which are not only difficult to obtain at scale but are also susceptible to inconsistencies due to the subjectivity and cognitive fatigue of annotators. To overcome these limitations and, to provide a clinically…”

“…genetics, cytochemistry, flow cytometry). An equally important drawback of previous fully supervised models 7 , 9 , 13 , 16 , 17 is that they require hundreds of thousands of object-level cell annotations provided by human experts 18 which are not only difficult to obtain at scale but are also susceptible to inconsistencies due to the subjectivity and cognitive fatigue of annotators. To overcome these limitations and, to provide a clinically relevant system that could support, together with clinical evaluation and ancillary laboratory parameters, the prompt treatment in APL cases, we designed a Multiple Instance Learning for Leukocyte Identification (MILLIE) approach.…”

Toward Driverless AI: Automating Leukemia Detection and Classification using Hyperautomation, a Case Study

“…The testing findings demonstrated that the method proposed achieved a precision of (98.3%). Eckardt JN et al [20]suggested a multi-stage DL framework in another DL study to detect acute promyelocytic leukemia (APL) only from bone marrow smear pictures. They examined a small medical dataset of 51 APL patients and contrasted it with samples from 236 normal bone marrow donors and 1048 AML patients who did not have APL.…”