Recent advancements in machine learning for bone marrow cell morphology analysis

As machine learning progresses, techniques such as neural networks, decision trees, and support vector machines are being increasingly applied in the medical domain, especially for tasks involving large datasets, such as cell detection, recognition, classification, and visualization. Within the domain of bone marrow cell morphology analysis, deep learning offers substantial benefits due to its robustness, ability for automatic feature learning, and strong image characterization capabilities. Deep neural networks are a machine learning paradigm specifically tailored for image processing applications. Artificial intelligence serves as a potent tool in supporting the diagnostic process of clinical bone marrow cell morphology. Despite the potential of artificial intelligence to augment clinical diagnostics in this domain, manual analysis of bone marrow cell morphology remains the gold standard and an indispensable tool for identifying, diagnosing, and assessing the efficacy of hematologic disorders. However, the traditional manual approach is not without limitations and shortcomings, necessitating, the exploration of automated solutions for examining and analyzing bone marrow cytomorphology. This review provides a multidimensional account of six bone marrow cell morphology processes: automated bone marrow cell morphology detection, automated bone marrow cell morphology segmentation, automated bone marrow cell morphology identification, automated bone marrow cell morphology classification, automated bone marrow cell morphology enumeration, and automated bone marrow cell morphology diagnosis. Highlighting the attractiveness and potential of machine learning systems based on bone marrow cell morphology, the review synthesizes current research and recent advances in the application of machine learning in this field. The objective of this review is to offer recommendations to hematologists for selecting the most suitable machine learning algorithms to automate bone marrow cell morphology examinations, enabling swift and precise analysis of bone marrow cytopathic trends for early disease identification and diagnosis. Furthermore, the review endeavors to delineate potential future research avenues for machine learning-based applications in bone marrow cell morphology analysis.

Quantitative interpretation of bone marrow biopsies in MPN-What's the point in a molecular age?

The diagnosis of myeloproliferative neoplasms (MPN) requires the integration of clinical, morphological, genetic and immunophenotypic findings. Recently, there has been a transformation in our understanding of the cellular and molecular mechanisms underlying disease initiation and progression in MPN. This has been accompanied by the widespread application of high-resolution quantitative molecular techniques. By contrast, microscopic interpretation of bone marrow biopsies by haematologists/haematopathologists remains subjective and qualitative. However, advances in tissue image analysis and artificial intelligence (AI) promise to transform haematopathology. Pioneering studies in bone marrow image analysis offer to refine our understanding of the boundaries between reactive samples and MPN subtypes and better capture the morphological correlates of high-risk disease. They also demonstrate potential to improve the evaluation of current and novel therapeutics for MPN and other blood cancers. With increased therapeutic targeting of diverse molecular, cellular and extra-cellular components of the marrow, these approaches can address the unmet need for improved objective and quantitative measures of disease modification in the context of clinical trials. This review focuses on the state-of-the-art in image analysis/AI of bone marrow tissue, with an emphasis on its potential to complement and inform future clinical studies and research in MPN.

Analysis of cellularity in H&E-stained rat bone marrow tissue via deep learning

Our objective was to develop an automated deep-learning-based method to evaluate cellularity in rat bone marrow hematoxylin and eosin whole slide images for preclinical safety assessment. We trained a shallow CNN for segmenting marrow, 2 Mask R-CNN models for segmenting megakaryocytes (MKCs), and small hematopoietic cells (SHCs), and a SegNet model for segmenting red blood cells. We incorporated the models into a pipeline that identifies and counts MKCs and SHCs in rat bone marrow. We compared cell segmentation and counts that our method generated to those that pathologists generated on 10 slides with a range of cell depletion levels from 10 studies. For SHCs, we compared cell counts that our method generated to counts generated by Cellpose and Stardist. The median Dice and object Dice scores for MKCs using our method vs pathologist consensus and the inter- and intra-pathologist variation were comparable, with overlapping first-third quartile ranges. For SHCs, the median scores were close, with first-third quartile ranges partially overlapping intra-pathologist variation. For SHCs, in comparison to Cellpose and Stardist, counts from our method were closer to pathologist counts, with a smaller 95% limits of agreement range. The performance of the bone marrow analysis pipeline supports its incorporation into routine use as an aid for hematotoxicity assessment by pathologists. The pipeline could help expedite hematotoxicity assessment in preclinical studies and consequently could expedite drug development. The method may enable meta-analysis of rat bone marrow characteristics from future and historical whole slide images and may generate new biological insights from cross-study comparisons.

Feature-guided attention network for medical image segmentation

BACKGROUND

U-Net and its variations have achieved remarkable performances in medical image segmentation. However, they have two limitations. First, the shallow layer feature of the encoder always contains background noise. Second, semantic gaps exist between the features of the encoder and the decoder. Skip-connections directly connect the encoder to the decoder, which will lead to the fusion of semantically dissimilar feature maps.

PURPOSE

To overcome these two limitations, this paper proposes a novel medical image segmentation algorithm, called feature-guided attention network, which consists of U-Net, the cross-level attention filtering module (CAFM), and the attention-guided upsampling module (AUM).

METHODS

In the proposed method, the AUM and the CAFM were introduced into the U-Net, where the AUM learns to filter the background noise in the low-level feature map of the encoder and the CAFM tries to eliminate the semantic gap between the encoder and the decoder. Specifically, the AUM adopts a top-down pathway to use the high-level feature map so as to filter the background noise in the low-level feature map of the encoder. The AUM uses the encoder features to guide the upsampling of the corresponding decoder features, thus eliminating the semantic gap between them. Four medical image segmentation tasks, including coronary atherosclerotic plaque segmentation (Dataset A), retinal vessel segmentation (Dataset B), skin lesion segmentation (Dataset C), and multiclass retinal edema lesions segmentation (Dataset D), were used to validate the proposed method.

RESULTS

For Dataset A, the proposed method achieved higher Intersection over Union (IoU) (67.91 ± 3.82 % $67.91\pm 3.82\%$ ), dice (79.39 ± 3.37 % $79.39\pm 3.37\%$ ), accuracy (98.39 ± 0.34 % $98.39\pm 0.34\%$ ), and sensitivity (85.10 ± 3.74 % $85.10\pm 3.74\%$ ) than the previous best method: CA-Net. For Dataset B, the proposed method achieved higher sensitivity (83.50%) and accuracy (97.55%) than the previous best method: SCS-Net. For Dataset C, the proposed method had highest IoU (83.47 ± 0.41 % $83.47\pm 0.41\%$ ) and dice (90.81 ± 0.34 % $90.81\pm 0.34\%$ ) than those of all compared previous methods. For Dataset D, the proposed method had highest dice (average: 81.53%; retina edema area [REA]: 83.78%; pigment epithelial detachment [PED] 77.13%), sensitivity (REA: 89.01%; SRF: 85.50%), specificity (REA: 99.35%; PED: 100.00), and accuracy (98.73%) among all compared previous networks. In addition, the number of parameters of the proposed method was 2.43 M, which is less than CA-Net (3.21 M) and CPF-Net (3.07 M).

CONCLUSIONS

The proposed method demonstrated state-of-the-art performance, outperforming other top-notch medical image segmentation algorithms. The CAFM filtered the background noise in the low-level feature map of the encoder, while the AUM eliminated the semantic gap between the encoder and the decoder. Furthermore, the proposed method was of high computational efficiency.

A survey on recent trends in deep learning for nucleus segmentation from histopathology images

Nucleus segmentation is an imperative step in the qualitative study of imaging datasets, considered as an intricate task in histopathology image analysis. Segmenting a nucleus is an important part of diagnosing, staging, and grading cancer, but overlapping regions make it hard to separate and tell apart independent nuclei. Deep Learning is swiftly paving its way in the arena of nucleus segmentation, attracting quite a few researchers with its numerous published research articles indicating its efficacy in the field. This paper presents a systematic survey on nucleus segmentation using deep learning in the last five years (2017-2021), highlighting various segmentation models (U-Net, SCPP-Net, Sharp U-Net, and LiverNet) and exploring their similarities, strengths, datasets utilized, and unfolding research areas.

Impact of Training Data, Ground Truth and Shape Variability in the Deep Learning-Based Semantic Segmentation of HeLa Cells Observed with Electron Microscopy

This paper investigates the impact of the amount of training data and the shape variability on the segmentation provided by the deep learning architecture U-Net. Further, the correctness of ground truth (GT) was also evaluated. The input data consisted of a three-dimensional set of images of HeLa cells observed with an electron microscope with dimensions 8192×8192×517. From there, a smaller region of interest (ROI) of 2000×2000×300 was cropped and manually delineated to obtain the ground truth necessary for a quantitative evaluation. A qualitative evaluation was performed on the 8192×8192 slices due to the lack of ground truth. Pairs of patches of data and labels for the classes nucleus, nuclear envelope, cell and background were generated to train U-Net architectures from scratch. Several training strategies were followed, and the results were compared against a traditional image processing algorithm. The correctness of GT, that is, the inclusion of one or more nuclei within the region of interest was also evaluated. The impact of the extent of training data was evaluated by comparing results from 36,000 pairs of data and label patches extracted from the odd slices in the central region, to 135,000 patches obtained from every other slice in the set. Then, 135,000 patches from several cells from the 8192×8192 slices were generated automatically using the image processing algorithm. Finally, the two sets of 135,000 pairs were combined to train once more with 270,000 pairs. As would be expected, the accuracy and Jaccard similarity index improved as the number of pairs increased for the ROI. This was also observed qualitatively for the 8192×8192 slices. When the 8192×8192 slices were segmented with U-Nets trained with 135,000 pairs, the architecture trained with automatically generated pairs provided better results than the architecture trained with the pairs from the manually segmented ground truths. This suggests that the pairs that were extracted automatically from many cells provided a better representation of the four classes of the various cells in the 8192×8192 slice than those pairs that were manually segmented from a single cell. Finally, the two sets of 135,000 pairs were combined, and the U-Net trained with these provided the best results.

The Application of Morphogo in the Detection of Megakaryocytes from Bone Marrow Digital Images with Convolutional Neural Networks

The evaluation of megakaryocytes is an important part of the work up on bone marrow smear examination. It has significance in the differential diagnosis, therapeutic efficacy assessment, and predication of prognosis of many hematologic diseases. The process of manual identification of megakaryocytes are tedious and lack of reproducibility; therefore, a reliable method of automated megakaryocytic identification is urgently needed. Three hundred and thirty-three bone marrow aspirate smears were digitized by Morphogo system. Pathologists annotated megakaryocytes on the digital images of marrow smears are applied to construct a large dataset for testing the system's predictive performance. Subsequently, we obtained megakaryocyte count and classification for each sample by different methods (system-automated analysis, system-assisted analysis, and microscopic examination) to study the correlation between different counting and classification methods. Morphogo system localized cells likely to be megakaryocytes on digital smears, which were later annotated by pathologists and the system, respectively. The system showed outstanding performance in identifying megakaryocytes in bone marrow smears with high sensitivity (96.57%) and specificity (89.71%). The overall correlation between the different methods was confirmed the high consistency (r ≥ 0.7218, R² ≥ 0.5211) with microscopic examination in classifying megakaryocytes. Morphogo system was proved as a reliable screen tool for analyzing megakaryocytes. The application of Morphogo system shows promises to advance the automation and standardization of bone marrow smear examination.

RAMRU-CAM: Residual-Atrous MultiResUnet with Channel Attention Mechanism for cell segmentation

The task of cell segmentation in microscope images is difficult and popular. In recent years, deep learning-based techniques have made incredible progress in medical and microscopy image segmentation applications. In this paper, we propose a novel deep learning approach called Residual-Atrous MultiResUnet with Channel Attention Mechanism (RAMRU-CAM) for cell segmentation, which combines MultiResUnet architecture with Channel Attention Mechanism (CAM) and Residual-Atrous connections. The Residual-Atrous path mitigates the semantic gap between the encoder and decoder stages and manages the spatial dimension of feature maps. Furthermore, the Channel Attention Mechanism (CAM) blocks are used in the decoder stages to better maintain the spatial details before concatenating the feature maps from the encoder phases to the decoder phases. We evaluated our proposed model on the PhC-C2DH-U373 and Fluo-N2DH-GOWT1 datasets. The experimental results show that our proposed model outperforms recent variants of the U-Net model and the state-of-the-art approaches. We have demonstrated how our model can segment cells precisely while using fewer parameters and low computational complexity.

Label Cleaning Multiple Instance Learning: Refining Coarse Annotations on Single Whole-Slide Images

Annotating cancerous regions in whole-slide images (WSIs) of pathology samples plays a critical role in clinical diagnosis, biomedical research, and machine learning algorithms development. However, generating exhaustive and accurate annotations is labor-intensive, challenging, and costly. Drawing only coarse and approximate annotations is a much easier task, less costly, and it alleviates pathologists' workload. In this paper, we study the problem of refining these approximate annotations in digital pathology to obtain more accurate ones. Some previous works have explored obtaining machine learning models from these inaccurate annotations, but few of them tackle the refinement problem where the mislabeled regions should be explicitly identified and corrected, and all of them require a - often very large - number of training samples. We present a method, named Label Cleaning Multiple Instance Learning (LC-MIL), to refine coarse annotations on a single WSI without the need for external training data. Patches cropped from a WSI with inaccurate labels are processed jointly within a multiple instance learning framework, mitigating their impact on the predictive model and refining the segmentation. Our experiments on a heterogeneous WSI set with breast cancer lymph node metastasis, liver cancer, and colorectal cancer samples show that LC-MIL significantly refines the coarse annotations, outperforming state-of-the-art alternatives, even while learning from a single slide. Moreover, we demonstrate how real annotations drawn by pathologists can be efficiently refined and improved by the proposed approach. All these results demonstrate that LC-MIL is a promising, lightweight tool to provide fine-grained annotations from coarsely annotated pathology sets.

Deep structured residual encoder-decoder network with a novel loss function for nuclei segmentation of kidney and breast histopathology images

To improve the process of diagnosis and treatment of cancer disease, automatic segmentation of haematoxylin and eosin (H & E) stained cell nuclei from histopathology images is the first step in digital pathology. The proposed deep structured residual encoder-decoder network (DSREDN) focuses on two aspects: first, it effectively utilized residual connections throughout the network and provides a wide and deep encoder-decoder path, which results to capture relevant context and more localized features. Second, vanished boundary of detected nuclei is addressed by proposing an efficient loss function that better train our proposed model and reduces the false prediction which is undesirable especially in healthcare applications. The proposed architecture experimented on three different publicly available H&E stained histopathological datasets namely: (I) Kidney (RCC) (II) Triple Negative Breast Cancer (TNBC) (III) MoNuSeg-2018. We have considered F1-score, Aggregated Jaccard Index (AJI), the total number of parameters, and FLOPs (Floating point operations), which are mostly preferred performance measure metrics for comparison of nuclei segmentation. The evaluated score of nuclei segmentation indicated that the proposed architecture achieved a considerable margin over five state-of-the-art deep learning models on three different histopathology datasets. Visual segmentation results show that the proposed DSREDN model accurately segment the nuclear regions than those of the state-of-the-art methods.

High-resolution deep transferred ASPPU-Net for nuclei segmentation of histopathology images

PURPOSE

Increasing cancer disease incidence worldwide has become a major public health issue. Manual histopathological analysis is a common diagnostic method for cancer detection. Due to the complex structure and wide variability in the texture of histopathology images, it has been challenging for pathologists to diagnose manually those images. Automatic segmentation of histopathology images to diagnose cancer disease is a continuous exploration field in recent times. Segmentation and analysis for diagnosis of histopathology images by using an efficient deep learning algorithm are the purpose of the proposed method.

METHOD

To improve the segmentation performance, we proposed a deep learning framework that consists of a high-resolution encoder path, an atrous spatial pyramid pooling bottleneck module, and a powerful decoder. Compared to the benchmark segmentation models having a deep and thin path, our network is wide and deep that effectively leverages the strength of residual learning as well as encoder-decoder architecture.

RESULTS

We performed careful experimentation and analysis on three publically available datasets namely kidney dataset, Triple Negative Breast Cancer (TNBC) dataset, and MoNuSeg histopathology image dataset. We have used the two most preferred performance metrics called F1 score and aggregated Jaccard index (AJI) to evaluate the performance of the proposed model. The measured values of F1 score and AJI score are (0.9684, 0.9394), (0.8419, 0.7282), and (0.8344, 0.7169) on the kidney dataset, TNBC histopathology dataset, and MoNuSeg dataset, respectively.

CONCLUSION

Our proposed method yields better results as compared to benchmark segmentation methods on three histopathology datasets. Visual segmentation results justify the high value of the F1 score and AJI scores which indicated that it is a very good prediction by our proposed model.

SRPN: similarity-based region proposal networks for nuclei and cells detection in histology images

The detection of nuclei and cells in histology images is of great value in both clinical practice and pathological studies. However, multiple reasons such as morphological variations of nuclei or cells make it a challenging task where conventional object detection methods cannot obtain satisfactory performance in many cases. A detection task consists of two sub-tasks, classification and localization. Under the condition of dense object detection, classification is a key to boost the detection performance. Considering this, we propose similarity based region proposal networks (SRPN) for nuclei and cells detection in histology images. In particular, a customised convolution layer termed as embedding layer is designed for network building. The embedding layer is added into the region proposal networks, enabling the networks to learn discriminative features based on similarity learning. Features obtained by similarity learning can significantly boost the classification performance compared to conventional methods. SRPN can be easily integrated into standard convolutional neural networks architectures such as the Faster R-CNN and RetinaNet. We test the proposed approach on tasks of multi-organ nuclei detection and signet ring cells detection in histological images. Experimental results show that networks applying similarity learning achieved superior performance on both tasks when compared to their counterparts. In particular, the proposed SRPN achieve state-of-the-art performance on the MoNuSeg benchmark for nuclei segmentation and detection while compared to previous methods, and on the signet ring cell detection benchmark when compared with baselines. The sourcecode is publicly available at: https://github.com/sigma10010/nuclei_cells_det.

Volumetric Semantic Instance Segmentation of the Plasma Membrane of HeLa Cells

In this work, an unsupervised volumetric semantic instance segmentation of the plasma membrane of HeLa cells as observed with serial block face scanning electron microscopy is described. The resin background of the images was segmented at different slices of a 3D stack of 518 slices with 8192 × 8192 pixels each. The background was used to create a distance map, which helped identify and rank the cells by their size at each slice. The centroids of the cells detected at different slices were linked to identify them as a single cell that spanned a number of slices. A subset of these cells, i.e., the largest ones and those not close to the edges were selected for further processing. The selected cells were then automatically cropped to smaller regions of interest of 2000 × 2000 × 300 voxels that were treated as cell instances. Then, for each of these volumes, the nucleus was segmented, and the cell was separated from any neighbouring cells through a series of traditional image processing steps that followed the plasma membrane. The segmentation process was repeated for all the regions of interest previously selected. For one cell for which the ground truth was available, the algorithm provided excellent results in Accuracy (AC) and the Jaccard similarity Index (JI): nucleus: JI =0.9665, AC =0.9975, cell including nucleus JI =0.8711, AC =0.9655, cell excluding nucleus JI =0.8094, AC =0.9629. A limitation of the algorithm for the plasma membrane segmentation was the presence of background. In samples with tightly packed cells, this may not be available. When tested for these conditions, the segmentation of the nuclear envelope was still possible. All the code and data were released openly through GitHub, Zenodo and EMPIAR.

CE-Net: Context Encoder Network for 2D Medical Image Segmentation

Medical image segmentation is an important step in medical image analysis. With the rapid development of a convolutional neural network in image processing, deep learning has been used for medical image segmentation, such as optic disc segmentation, blood vessel detection, lung segmentation, cell segmentation, and so on. Previously, U-net based approaches have been proposed. However, the consecutive pooling and strided convolutional operations led to the loss of some spatial information. In this paper, we propose a context encoder network (CE-Net) to capture more high-level information and preserve spatial information for 2D medical image segmentation. CE-Net mainly contains three major components: a feature encoder module, a context extractor, and a feature decoder module. We use the pretrained ResNet block as the fixed feature extractor. The context extractor module is formed by a newly proposed dense atrous convolution block and a residual multi-kernel pooling block. We applied the proposed CE-Net to different 2D medical image segmentation tasks. Comprehensive results show that the proposed method outperforms the original U-Net method and other state-of-the-art methods for optic disc segmentation, vessel detection, lung segmentation, cell contour segmentation, and retinal optical coherence tomography layer segmentation.

[A fast adaptive active contour model based on local gray difference for parotid duct]

OBJECTIVE

To establish a fast adaptive active contour model based on local gray difference for parotid duct image segmentation.

METHODS

On the basis of the LBF model, we added the mean difference of the local gray scale inside and outside the contour as the energy term of the driving evolution curve, and the local gray-scale variance difference was used to replaceλ₁ and λ₂ as the control term of the energy parameter value. Two local similarity factors of different neighborhood sizes were introduced to correct the effects of image gray unevenness and boundary blur to improve the segmentation efficiency.

RESULTS

During image segmentation, this algorithm allowed for adaptive adjustment of the evolution direction, velocity and the energy weight of the internal and external regions according to the difference of gray mean and variance between the internal and external regions. This algorithm was also capable of detecting the actual boundary in a complex gradient boundary region, thus enabling the evolution curve to approach the target boundary quickly and accurately.

CONCLUSIONS

The proposed algorithm is superior to the existing segmentation algorithms and allows fast and accurate segmentation of the parotid duct with well-preserved image details.

Graph-Based Rate Control in Pathology Imaging With Lossless Region of Interest Coding

The increasing availability of digital pathology images has motivated the design of tools to foster multidisciplinary collaboration among researchers, pathologists, and computer scientists. Telepathology plays an important role in the development of collaborative tools, as it facilitates the transmission and access to pathology images by multiple users. However, the huge file size associated with pathology images usually prevents full exploitation of the collaborative telepathology system potential. Within this context, rate control (RC) is an important tool that allows meeting storage and bandwidth requirements by controlling the bit rate of the coded image. In this paper, we propose a novel graph-based RC algorithm with lossless region of interest (RoI) coding for pathology images. The algorithm, which is designed for block-based predictive transform coding methods, compresses the non-RoI in a lossy manner according to a target bit rate and the RoI in a lossless manner. It employs a graph where each node represents a constituent block of the image to be coded. By incorporating information about the coding cost similarities of blocks into the graph, a graph kernel is used to distribute a target bit budget among the non-RoI blocks. In order to increase RC accuracy, the algorithm uses a rate-lambda (R- ) model to approximate the slope of the rate-distortion curve of the non-RoI, and a graph-based approach to guarantee that the target bit rate is accurately attained. The algorithm is implemented in the High-Efficiency Video Coding standard and tested over a wide range of pathology images with multiple RoIs. Evaluation results show that it outperforms the other state-of-the-art methods designed for single images by very accurately attaining the target bit rate of the non-RoI.

ImaEdge: a platform for the quantitative analysis of cortical proteins spatiotemporal dynamics during cell polarization

Cell polarity involves the compartmentalization of the cell cortex. The establishment of cortical compartments arises from the spatial bias in the activity and concentration of cortical proteins. The mechanistic dissection of cell polarity requires the accurate detection of dynamic changes in cortical proteins, but the fluctuations of cell shape and the inhomogeneous distributions of cortical proteins greatly complicate the quantitative extraction of their global and local changes during cell polarization. To address these problems, we introduce an open-source software package, ImaEdge, which automates the segmentation of the cortex from time-lapse movies, and enables quantitative extraction of cortical protein intensities. We demonstrate that ImaEdge enables efficient and rigorous analysis of the dynamic evolution of cortical PAR proteins during C. elegans embryogenesis. It is also capable of accurate tracking of varying levels of transgene expression and discontinuous signals of the actomyosin cytoskeleton during multiple rounds of cell division. ImaEdge provides a unique resource for the quantitative studies of cortical polarization, with the potential for application to many types of polarized cells.