Routine bone marrow examination in the management of acute lymphoblastic leukaemia of childhood

Multi-Method Diagnosis of Blood Microscopic Sample for Early Detection of Acute Lymphoblastic Leukemia Based on Deep Learning and Hybrid Techniques

Leukemia is one of the most dangerous types of malignancies affecting the bone marrow or blood in all age groups, both in children and adults. The most dangerous and deadly type of leukemia is acute lymphoblastic leukemia (ALL). It is diagnosed by hematologists and experts in blood and bone marrow samples using a high-quality microscope with a magnifying lens. Manual diagnosis, however, is considered slow and is limited by the differing opinions of experts and other factors. Thus, this work aimed to develop diagnostic systems for two Acute Lymphoblastic Leukemia Image Databases (ALL_IDB1 and ALL_IDB2) for the early detection of leukemia. All images were optimized before being introduced to the systems by two overlapping filters: the average and Laplacian filters. This study consists of three proposed systems as follows: the first consists of the artificial neural network (ANN), feed forward neural network (FFNN), and support vector machine (SVM), all of which are based on hybrid features extracted using Local Binary Pattern (LBP), Gray Level Co-occurrence Matrix (GLCM) and Fuzzy Color Histogram (FCH) methods. Both ANN and FFNN reached an accuracy of 100%, while SVM reached an accuracy of 98.11%. The second proposed system consists of the convolutional neural network (CNN) models: AlexNet, GoogleNet, and ResNet-18, based on the transfer learning method, in which deep feature maps were extracted and classified with high accuracy. All the models obtained promising results for the early detection of leukemia in both datasets, with an accuracy of 100% for the AlexNet, GoogleNet, and ResNet-18 models. The third proposed system consists of hybrid CNN–SVM technologies, consisting of two blocks: CNN models for extracting feature maps and the SVM algorithm for classifying feature maps. All the hybrid systems achieved promising results, with AlexNet + SVM achieving 100% accuracy, Goog-LeNet + SVM achieving 98.1% accuracy, and ResNet-18 + SVM achieving 100% accuracy.

Towards the Segmentation and Classification of White Blood Cell Cancer Using Hybrid Mask-Recurrent Neural Network and Transfer Learning

Inside the bone marrow, plasma cells are created, and they are a type of white blood cells. They are made from B lymphocytes. Antigens are produced by plasma cells to combat bacteria and viruses and prevent inflammation and illness. Multiple myeloma is a plasma cell cancer that starts in the bone marrow and causes the formation of abnormal plasma cells. Multiple myeloma is firmly identified by examining bone marrow samples under a microscope for myeloma cells. To diagnose myeloma cells, pathologists have to be very selective. Furthermore, because the ultimate decision is based on human sight and opinion, there is a possibility of error in the result. The nobility of this research is that it provides a computer-assisted technique for recognizing and detecting myeloma cells in bone marrow smears. For recognizing purposes, we have used Mask-Recurrent Convolutional Neural Network, and for detection purposes, Efficient Net B3 has been used. There are already many studies on white blood cell cancer, but very few with both segmentation and classification. We have designed two models. One is for recognizing myeloma cells, and the other is for differentiating them from nonmyeloma cells. Also, a new data set has been made from the multiple myeloma data sets, which has been used in our classification model. This research focuses on hybrid segmentation models and increases the accuracy level of the classification model. Both of our models are trained pretty well, where the Mask-RCNN model gives a mean average precision (mAP) of 93% and the Efficient Net B3 model gives 94.68% accuracy. The result of this research indicates that the Mask-RCNN model can recognize multiple myeloma and Efficient Net B3 can distinguish between myeloma and nonmyeloma cells and beats most of the state of the art in myeloma recognition and detection.

Recognition of acute lymphoblastic leukemia and lymphocytes cell subtypes in microscopic images using random forest classifier

Acute lymphoblastic leukemia (ALL) is the most frequently leukemia and categorized into three morphological subtypes named L1, L2 and L3. Early diagnosis of ALL plays a key role in treatment procedure especially in the case of children. Several similarities between morphology of three subtypes ALL (L1, L2, L3) and lymphocyte subtypes (normal, reactive and atypical) as noncancerous cells have remained a high challenge. Diagnosis of ALL and lymphocyte subtypes are done by microscopic viewing examination of cells in the peripheral blood samples by hematologists. Since this exam is time-consuming, boring and dependent on the skill of the hematologists, automatic systems are desired to overcome these limitations. In this study, 312 microscopic images including 958 cells are obtained from blood samples of 7 normal subjects and 14 patients. The first step of proposed system is image enhancement to decreases the effects of various luminosity situations with transformation from RGB to HSV color space and then applying histogram equalization on V channel for equalizing the grey level of image lightness. Nuclei segmentation from the blood cell images is the second step and performed using fuzzy c-means (FCM) clustering. After identify cluster of nuclei, we performed opening and closing process in morphological operation binary in order to remove extra noises and fill some minor holes in the nuclei. Moreover, to discrete the link between nuclei, watershed transform was applied. Then, a set of quantitative features (five geometric features about the size and figure of a cell and 36 statistical features about the spatial arrangement of intensities of nuclei image) are extracted to characterize the properties of these nuclei. In the next step, due to high number of features, the best features are selected by exhaustive search of all of the subsets of features and 13 features are selected. The final step is the classification of L1, L2, L3, normal, reactive and atypical cells by applying Random Forest (RF) classifier and result in 98% accuracy. We compared RF classifier with two other commonly classifiers named: MultiLayer Perceptron (MLP), and multi-SVM classifier with more success especially for recognition of L1, normal and reactive cells. So, this system can be used as an assistant diagnostic tool for hematologists to recognize subtypes of ALL and lymphocyte.

Automatic Recognition of Acute Myelogenous Leukemia in Blood Microscopic Images Using K-means Clustering and Support Vector Machine

Acute myelogenous leukemia (AML) is a subtype of acute leukemia, which is characterized by the accumulation of myeloid blasts in the bone marrow. Careful microscopic examination of stained blood smear or bone marrow aspirate is still the most significant diagnostic methodology for initial AML screening and considered as the first step toward diagnosis. It is time-consuming and due to the elusive nature of the signs and symptoms of AML; wrong diagnosis may occur by pathologists. Therefore, the need for automation of leukemia detection has arisen. In this paper, an automatic technique for identification and detection of AML and its prevalent subtypes, i.e., M2-M5 is presented. At first, microscopic images are acquired from blood smears of patients with AML and normal cases. After applying image preprocessing, color segmentation strategy is applied for segmenting white blood cells from other blood components and then discriminative features, i.e., irregularity, nucleus-cytoplasm ratio, Hausdorff dimension, shape, color, and texture features are extracted from the entire nucleus in the whole images containing multiple nuclei. Images are classified to cancerous and noncancerous images by binary support vector machine (SVM) classifier with 10-fold cross validation technique. Classifier performance is evaluated by three parameters, i.e., sensitivity, specificity, and accuracy. Cancerous images are also classified into their prevalent subtypes by multi-SVM classifier. The results show that the proposed algorithm has achieved an acceptable performance for diagnosis of AML and its common subtypes. Therefore, it can be used as an assistant diagnostic tool for pathologists.

Recognition of acute lymphoblastic leukemia cells in microscopic images using k-means clustering and support vector machine classifier

Acute lymphoblastic leukemia is the most common form of pediatric cancer which is categorized into three L1, L2, and L3 and could be detected through screening of blood and bone marrow smears by pathologists. Due to being time-consuming and tediousness of the procedure, a computer-based system is acquired for convenient detection of Acute lymphoblastic leukemia. Microscopic images are acquired from blood and bone marrow smears of patients with Acute lymphoblastic leukemia and normal cases. After applying image preprocessing, cells nuclei are segmented by k-means algorithm. Then geometric and statistical features are extracted from nuclei and finally these cells are classified to cancerous and noncancerous cells by means of support vector machine classifier with 10-fold cross validation. These cells are also classified into their sub-types by multi-Support vector machine classifier. Classifier is evaluated by these parameters: Sensitivity, specificity, and accuracy which values for cancerous and noncancerous cells 98%, 95%, and 97%, respectively. These parameters are also used for evaluation of cell sub-types which values in mean 84.3%, 97.3%, and 95.6%, respectively. The results show that proposed algorithm could achieve an acceptable performance for the diagnosis of Acute lymphoblastic leukemia and its sub-types and can be used as an assistant diagnostic tool for pathologists.

Value of routine bone marrow examination in pediatric acute myeloid leukemia (AML): a study of the Dutch Childhood Oncology Group (DCOG)

BACKGROUND

The outcome of the treatment of pediatric acute myeloid leukemia (AML) is still disappointing, due to relatively high treatment-related mortality and relapse rates (30-40%). Past treatment protocols have called for routine screening via bone marrow aspiration (BMA) after achievement of first complete remission (CR1) to detect relapse at an early stage. However, supporting evidence for this policy is lacking for non-FAB type-M3 patients.

PROCEDURE

We therefore retrospectively studied the clinical relevance of routine BMA in an unselected cohort of all pediatric AML patients in the Netherlands.

RESULTS

Of 440 patients, data for 349 patients, of whom 148 suffered bone marrow relapse (BM-relapse), could be analyzed. A total of 1,790 BMAs had been performed, 1,648 (92%) routinely, and 142 (8%) on indication when a relapse was suspected. Forty routine BMAs showed BM-relapse (2% of all routine BMAs), while as many as 108 (76%) hematological relapses were confirmed by BMA on indication (P < 0.001). Therefore, 1 in 41 routine BMAs, as opposed to 1 in 1.3 BMAs performed on indication, detected a BM-relapse.

CONCLUSIONS

Routine BMA after CR1 did not significantly contribute to early detection of relapsed AML. These results suggest that BMA after achievement of CR1 should only be performed on indication or within a clinical research setting. Pediatr Blood Cancer 2012; 59: 1239-1244. © 2012 Wiley Periodicals, Inc.

Submicroscopic bone marrow involvement in isolated extramedullary relapses in childhood acute lymphoblastic leukemia: a more precise definition of “isolated” and its possible clinical implications, a collaborative study of the Resistant Disease Committee of the International BFM study group

This study investigates the extent of bone marrow (BM) involvement at diagnosis of apparent isolated extramedullary (AIEM) relapses of childhood acute lymphoblastic leukemia (ALL) and its relation to prognosis. Sixty-four children with first AIEM relapse treated in Germany, Czech Republic, or France were included. Real-time quantitative polymerase chain reaction using T-cell receptor and immunoglobulin gene rearrangements provided a sensitive measure of submicroscopic BM involvement, which was detectable at a level of 10−4 or higher in 46 patients and less than 10−4 in 11 patients, and was nondetectable (sensitivity: 10−4) in 7 patients. In the total cohort, the probability of event-free survival (pEFS) for children with BM involvement of 10−4 or higher was 0.30 (0.09 ± SE) versus 0.60 (± 0.12) for those with less than 10−4 (P = .13). The cumulative incidence of subsequent relapse was 0.24 (± 0.01) for patients with BM involvement less than 10−4 and 0.65 (± 0.01) for those with 10−4 or higher (P = .012). Restricted to central nervous system (CNS) relapses, pEFS was 0.11 (± 0.09) for patients with BM involvement 10−4 or higher and 0.63 (± 0.17) for those with less than 10−4 (P = .053). CNS relapses were associated with a higher (≥ 10−4: 80%) submicroscopic BM involvement than testicular relapses (≥ 10−4: 57%, P = .08). In summary, we show marked heterogeneity of submicroscopic BM involvement at first AIEM relapse diagnosis in children with ALL, and demonstrate its possible prognostic relevance.

Childhood acute lymphoblastic leukemia: are routine end-of-therapy bone marrow and cerebrospinal fluid examinations necessary?

OBJECTIVE

To ascertain the usefulness of bone marrow and cerebrospinal fluid (CSF) examinations in identifying or predicting relapse in children with acute lymphoblastic leukemia (ALL) before discontinuation of chemotherapy.

MATERIAL AND METHODS

We retrospectively reviewed the medical records of 113 children with ALL in first continuous complete remission who had undergone routine end-of-therapy bone marrow aspiration and CSF examinations.

RESULTS

One patient had frank bone marrow relapse at the completion of therapy, which was evident by the presence of blasts in the peripheral blood. None of the other 112 patients had morphologic evidence of bone marrow relapse or positive CSF cytologic findings. The seven subsequent relapses could not have been predicted by the results of end-of-therapy bone marrow or CSF studies.

CONCLUSION

Routine morphologic examination of the bone marrow and CSF at the completion of therapy for ALL has no diagnostic or prognostic value.

Value of routine bone marrow examination for detection of bone marrow relapse in children with standard risk acute lymphoblastic leukemia

The value of routine bone marrow examination (RBME) in children during and after treatment for standard risk acute lymphoblastic leukemia (SR-ALL) was investigated. The clinical symptoms and peripheral blood findings at the time of bone marrow relapse of 28 children were reviewed and compared with those of 28 matched controls in continuous complete remission. Five (45%) children with bone marrow relapse during maintenance therapy and six (35%) after cessation of cytostatic treatment were asymptomatic at the time of relapse. Signs indicative of relapse during treatment were lymphoblast cells in the peripheral blood, thrombocytopenia, hepatomegaly, anemia, or leukopenia in decreasing order of frequency. After cessation of treatment these signs were lymphoblasts in the peripheral blood, hepatomegaly, splenomegaly, thrombocytopenia, or leukocytosis. Except for one case with thrombocytopenia, no signs suspicious for relapse were found in the control groups. When each sign was evaluated separately only the presence of lymphoblasts in peripheral blood and hepatomegaly were significant symptoms for relapse after cessation of treatment. The mean percentage of lymphoblasts in the bone marrow at the time of relapse was significantly lower for patients with an unpredicted relapse (46.8%) than patients with clinical and/or laboratory evidence of relapse (79.5%). When lymphoblasts were present in the peripheral blood the percentage of lymphoblasts in the bone marrow was always more than 40%, both during and after cessation of treatment. These data suggest a relation between clinical and laboratory symptoms and progression of the disease. It is concluded that 46% of relapses are detected by RBME in the absence of clinical or laboratory symptoms. This early detection may have a positive prognostic influence with more effective treatment for relapsed ALL.

Bone marrow relapse occurring as first relapse in children with acute lymphoblastic leukemia

In a retrospective review which covered the whole Dutch childhood population of approximately 3 million children we studied the prognosis in 164 children with acute lymphoblastic leukemia (ALL) who were initially treated between 1973 and 1983, and who had an isolated bone marrow relapse occurring as first relapse. Until their first relapse, the patients were initially treated according to standard protocols, while treatment for relapse was heterogeneous, and not intensive. Second complete remission (CR) was attained by 78% of the patients. The median duration of second CR was 9 months, the median survival 13 months. Multivariate analysis showed that the duration of the first CR was the most significant variable with regard to prognosis. None of the patients who developed their bone marrow relapse during initial treatment, i.e., within 24 months from diagnosis, survived. Among the 73 patients who relapsed after cessation of the initial treatment there were 19 long-term disease-free survivors, 14 of whom had not developed subsequent relapses after 48(+)-125 + months. From this study we conclude that treatment in children with first bone marrow relapse has to be intensified.

Speed of relapse in children with acute lymphoblastic leukaemia

16 children with acute lymphoblastic leukaemia (ALL) who relapsed on maintenance treatment between September 1978 and December 1981 were studied. Their previous bone marrow aspirates were reviewed to determine the first evidence for marrow relapse and the subsequent rate of evolution of the disease. 60% of children had bone marrow evidence of unsuspected relapse at 8 wk, and 40% had evidence at 12 wk before clinical or peripheral blood relapse occurred. Early detection of relapse will prevent continued use of ineffective maintenance chemotherapy and may also reduce morbidity during subsequent induction therapy. Regular 8-weekly marrow aspirates are therefore recommended during first remission of ALL in children with a histocompatible sibling who would be eligible for bone marrow transplantation in second remission.

Yield of unpredicted bone-marrow relapse diagnosed by routine marrow aspiration in children with acute lymphoblastic leukaemia. A report from the Children's Cancer Study Group

1466 children with acute lymphoblastic leukaemia had routine bone-marrow aspiration at the end of each 84-day cycle of maintenance therapy. Relapses detected by routine bone-marrow aspiration were classified according to whether or not they could be predicted by clinical signs or peripheral blood counts. In the low-risk, moderate-risk, and high-risk leukaemic patients 0.4%, 0.5%, and 0.8%, respectively, of the total routine bone-marrow aspirations yielded an unpredicted bone-marrow relapse. 19.4% of relapses were detected by routine surveillance marrow aspirations before any clinical signs of relapse on physical examination or peripheral blood count. The median survival after relapse in the predicted group was significantly shorter than that in the unpredicted group, but the eventual outcome was the same in both groups, 95% of all patients dying within 24 months of relapse.

Early relapse of acute lymphoblastic leukemia is not predictable by serial biochemical assays of terminal transferase activity in cells from peripheral blood

Serial samples of peripheral blood were collected from 37 children with acute lymphoblastic leukemia (ALL) in remission. Activity of terminal transferase (TdT) was assayed by a biochemical technique. False positive results were obtained infrequently (approx. 1%), in contrast to experience with bone marrow analyses. However, early relapse of disease was not predictable in ALL by repeated measurement of TdT in circulating mononuclear cells during remission.

Detection of circulating 'terminal transferase-positive' cells does not predict relapse in acute lymphoblastic leukemia

Serial samples of peripheral blood were obtained from 35 children with ALL over a period of 18 months. The mononuclear cells were examined for TdT by indirect immunofluorescence using an unpurified anti-calf thymus TdT as the primary antibody. This analysis failed to distinguish those children who were destined to relapse (n = 9) from those who remained in continuous complete remission. Rather, the exhibition of fluorescence was linked to the co-existence of infection, with a negative predictive value of 0.91. Putative 'TdT-positive' cells were concentrated in the T-lymphocyte fraction and the very process of E-rosette formation seemed to contribute to this phenomenon. It appears as if the anti-TdT reagent recognizes not only TdT but also a variety of antigens which are expressed on or in immature and activated lymphocytes.

A comparison of peripheral blood and buffy coat smear examination for the prediction of bone marrow relapse of acute lymphoblastic leukaemia in childhood

In an attempt to see if buffy coat smear examination might be an alternative to bone marrow aspiration for predicting relapse, 98 consecutive bone marrow aspirates from 96 children with acute lymphoblastic leukaemia were examined blind with buffy coat and peripheral blood from the same patients. The 28 bone marrow aspirates from children no longer on treatment were all normal, and routine aspirates would appear unjustified in these patients. Eight of the remaining marrows showed relapse, but only three were not predicted from the peripheral blood and buffy coat. In no case was buffy coat superior to peripheral blood in the detection of bone marrow relapse. Routine bone marrow aspirates are an inefficient way of diagnosing relapse in acute lymphoblastic leukaemia in childhood, despite their precision, and a prospective study is needed to determine their value.