Deep Learning for Automatic Differential Diagnosis of Primary Central Nervous System Lymphoma and Glioblastoma: Multi-Parametric Magnetic Resonance Imaging Based Convolutional Neural Network Model

Correlation of functional magnetic resonance imaging features of primary central nervous system lymphoma with vasculogenic mimicry and reticular fibers

Objective

To deepen the imaging-pathological mechanism of primary central nervous system lymphoma (PCNSL) and provide a theoretical basis for clinical diagnosis and treatment, the functional magnetic resonance imaging (fMRI) characteristics of PCNSL were analyzed, and the relationship between the fMRI characteristics and vasculogenic mimicry (VM) and reticular fiber in PCNSL was discussed.

Methods

Ninety-six patients with PCNSL treated in our hospital were divided into three groups according to the pathological examination results, including strong positive group of VM (n = 40), weak positive group of VM (n = 56), strong positive group of reticular fiber (n = 45) and weak positive group of reticular fiber (n = 51). The levels of augmentation index and apparent diffusion coefficient (ADC) were compared among the groups. receiver operator characteristic (ROC) curve analysis was used to analyze the clinical value of ADC value in differential diagnosis of PCNSL.

Results

The levels of augmentation index in the strong positive group of VM were significantly higher than that in the weak positive group of VM, and the ADC value in the strong positive group of VM was significantly lower than that in the weak positive group of VM (P < 0.001). The levels of augmentation index in the strong positive group of reticular fiber were significantly higher than that in the weak positive group of reticular fiber, and ADC value in the strong positive group of reticular fiber was significantly lower than that in reticular fiber weak positive group (P < 0.001). Pearson correlation analysis showed that the levels of augmentation index were positively correlated with VM and reticular fiber (r = 0.529, 0.548, P < 0.001) and the ADC value was negatively correlated with VM and reticular fiber (r = -0.485, -0.513, P < 0.001). There was a significant negative correlation between necrotic lesions and VM (r = -0.185, P < 0.05). The area under the curve (AUC) values of average ADC value, minimum ADC value, and maximum ADC value for individual differential diagnosis of PCNSL were 0.920, 0.901, and 0.702, while the AUC of the combined differential diagnosis was 0.985, with a sensitivity of 95.00 % and a specificity of 92.70 %.

Conclusion

The levels of augmentation index and the ADC value of PCNSL focus are significantly correlated with VM and reticular fiber, and there is a strong negative correlation between necrotic lesions and VM. MRI imaging technology is of great significance in revealing the biological behavior of PCNSL, which can effectively reveal the relationship between VM and reticular fibers and the MRI characteristics in PCNSL, thereby providing a new imaging basis for the clinical diagnosis and treatment of PCNSL.

How does deep learning/machine learning perform in comparison to radiologists in distinguishing glioblastomas (or grade IV astrocytomas) from primary CNS lymphomas?: a meta-analysis and systematic review

BACKGROUND

Several studies have been published comparing deep learning (DL)/machine learning (ML) to radiologists in differentiating PCNSLs from GBMs with equivocal results. We aimed to perform this meta-analysis to evaluate the diagnostic accuracy of ML/DL versus radiologists in classifying PCNSL versus GBM using MRI.

METHODOLOGY

The study was performed in accordance with PRISMA guidelines. Data was extracted and interpreted by two researchers with 12 and 23 years' experience, respectively, and QUADAS-2 tool was used for quality and risk-bias assessment. We constructed contingency tables to derive sensitivity, specificity accuracy, summary receiver operating characteristic (SROC) curve, and the area under the curve (AUC).

RESULTS

Our search identified 11 studies, of which 8 satisfied our inclusion criteria and restricted the analysis in each study to reporting the model showing highest accuracy, with a total sample size of 1159 patients. The random effects model showed a pooled sensitivity of 0.89 [95% CI:0.84-0.92] for ML and 0.82 [95% CI:0.76-0.87] for radiologists. Pooled specificity was 0.88 [95% CI: 0.84-0.91] for ML and 0.90 [95% CI: 0.81-0.95] for radiologists. Pooled accuracy was 0.88 [95% CI: 0.86-0.90] for ML and 0.86 [95% CI: 0.78-0.91] for radiologists. Pooled AUC of ML was 0.94 [95% CI:0.92-0.96]and for radiologists, it was 0.90 [95% CI: 0.84-0.93].

CONCLUSIONS

MRI-based ML/DL techniques can complement radiologists to improve the accuracy of classifying GBMs from PCNSL, possibly reduce the need for a biopsy, and avoid any unwanted neurosurgical resection of a PCNSL.

Artificial Intelligence Imaging for Predicting High-risk Molecular Markers of Gliomas

Gliomas, the most prevalent primary malignant tumors of the central nervous system, present significant challenges in diagnosis and prognosis. The fifth edition of the World Health Organization Classification of Tumors of the Central Nervous System (WHO CNS5) published in 2021, has emphasized the role of high-risk molecular markers in gliomas. These markers are crucial for enhancing glioma grading and influencing survival and prognosis. Noninvasive prediction of these high-risk molecular markers is vital. Genetic testing after biopsy, the current standard for determining molecular type, is invasive and time-consuming. Magnetic resonance imaging (MRI) offers a non-invasive alternative, providing structural and functional insights into gliomas. Advanced MRI methods can potentially reflect the pathological characteristics associated with glioma molecular markers; however, they struggle to fully represent gliomas' high heterogeneity. Artificial intelligence (AI) imaging, capable of processing vast medical image datasets, can extract critical molecular information. AI imaging thus emerges as a noninvasive and efficient method for identifying high-risk molecular markers in gliomas, a recent focus of research. This review presents a comprehensive analysis of AI imaging's role in predicting glioma high-risk molecular markers, highlighting challenges and future directions.

PrecisionLymphoNet: Advancing Malignant Lymphoma Diagnosis via Ensemble Transfer Learning with CNNs

Malignant lymphoma, which impacts the lymphatic system, presents diverse challenges in accurate diagnosis due to its varied subtypes-chronic lymphocytic leukemia (CLL), follicular lymphoma (FL), and mantle cell lymphoma (MCL). Lymphoma is a form of cancer that begins in the lymphatic system, impacting lymphocytes, which are a specific type of white blood cell. This research addresses these challenges by proposing ensemble and non-ensemble transfer learning models employing pre-trained weights from VGG16, VGG19, DenseNet201, InceptionV3, and Xception. For the ensemble technique, this paper adopts a stack-based ensemble approach. It is a two-level classification approach and best suited for accuracy improvement. Testing on a multiclass dataset of CLL, FL, and MCL reveals exceptional diagnostic accuracy, with DenseNet201, InceptionV3, and Xception exceeding 90% accuracy. The proposed ensemble model, leveraging InceptionV3 and Xception, achieves an outstanding 99% accuracy over 300 epochs, surpassing previous prediction methods. This study demonstrates the feasibility and efficiency of the proposed approach, showcasing its potential in real-world medical applications for precise lymphoma diagnosis.

Balanced transformer: efficient classification of glioblastoma and primary central nervous system lymphoma

Objective.Primary central nervous system lymphoma (PCNSL) and glioblastoma (GBM) are malignant primary brain tumors with different biological characteristics. Great differences exist between the treatment strategies of PCNSL and GBM. Thus, accurately distinguishing between PCNSL and GBM before surgery is very important for guiding neurosurgery. At present, the spinal fluid of patients is commonly extracted to find tumor markers for diagnosis. However, this method not only causes secondary injury to patients, but also easily delays treatment. Although diagnosis using radiology images is non-invasive, the morphological features and texture features of the two in magnetic resonance imaging (MRI) are quite similar, making distinction with human eyes and image diagnosis very difficult. In order to solve the problem of insufficient number of samples and sample imbalance, we used data augmentation and balanced sample sampling methods. Conventional Transformer networks use patch segmentation operations to divide images into small patches, but the lack of communication between patches leads to unbalanced data layers.Approach.To address this problem, we propose a balanced patch embedding approach that extracts high-level semantic information by reducing the feature dimensionality and maintaining the geometric variation invariance of the features. This approach balances the interactions between the information and improves the representativeness of the data. To further address the imbalance problem, the balanced patch partition method is proposed to increase the receptive field by sampling the four corners of the sliding window and introducing a linear encoding component without increasing the computational effort, and designed a new balanced loss function.Main results.Benefiting from the overall balance design, we conducted an experiment using Balanced Transformer and obtained an accuracy of 99.89%, sensitivity of 99.74%, specificity of 99.73% and AUC of 99.19%, which is far higher than the previous results (accuracy of 89.6% ∼ 96.8%, sensitivity of 74.3% ∼ 91.3%, specificity of 88.9% ∼ 96.02% and AUC of 87.8% ∼ 94.9%).Significance.This study can accurately distinguish PCNSL and GBM before surgery. Because GBM is a common type of malignant tumor, the 1% improvement in accuracy has saved many patients and reduced treatment times considerably. Thus, it can provide doctors with a good basis for auxiliary diagnosis.

Radiomics for differentiation of gliomas from primary central nervous system lymphomas: a systematic review and meta-analysis

Background and objective

Numerous radiomics-based models have been proposed to discriminate between central nervous system (CNS) gliomas and primary central nervous system lymphomas (PCNSLs). Given the heterogeneity of the existing models, we aimed to define their overall performance and identify the most critical variables to pilot future algorithms.

Methods

A systematic review of the literature and a meta-analysis were conducted, encompassing 12 studies and a total of 1779 patients, focusing on radiomics to differentiate gliomas from PCNSLs. A comprehensive literature search was performed through PubMed, Ovid MEDLINE, Ovid EMBASE, Web of Science, and Scopus databases. Overall sensitivity (SEN) and specificity (SPE) were estimated. Event rates were pooled using a random-effects meta-analysis, and the heterogeneity was assessed using the χ2 test.

Results

The overall SEN and SPE for differentiation between CNS gliomas and PCNSLs were 88% (95% CI = 0.83 - 0.91) and 87% (95% CI = 0.83 - 0.91), respectively. The best-performing features were the ones extracted from the Gray Level Run Length Matrix (GLRLM; ACC 97%), followed by those obtained from the Neighboring Gray Tone Difference Matrix (NGTDM; ACC 93%), and shape-based features (ACC 91%). The 18F-FDG-PET/CT was the best-performing imaging modality (ACC 97%), followed by the MRI CE-T1W (ACC 87% - 95%). Most studies applied a cross-validation analysis (92%).

Conclusion

The current SEN and SPE of radiomics to discriminate CNS gliomas from PCNSLs are high, making radiomics a helpful method to differentiate these tumor types. The best-performing features are the GLRLM, NGTDM, and shape-based features. The 18F-FDG-PET/CT imaging modality is the best-performing, while the MRI CE-T1W is the most used.

Artificial intelligence performance in detecting lymphoma from medical imaging: a systematic review and meta-analysis

BACKGROUND

Accurate diagnosis and early treatment are essential in the fight against lymphatic cancer. The application of artificial intelligence (AI) in the field of medical imaging shows great potential, but the diagnostic accuracy of lymphoma is unclear. This study was done to systematically review and meta-analyse researches concerning the diagnostic performance of AI in detecting lymphoma using medical imaging for the first time.

METHODS

Searches were conducted in Medline, Embase, IEEE and Cochrane up to December 2023. Data extraction and assessment of the included study quality were independently conducted by two investigators. Studies that reported the diagnostic performance of an AI model/s for the early detection of lymphoma using medical imaging were included in the systemic review. We extracted the binary diagnostic accuracy data to obtain the outcomes of interest: sensitivity (SE), specificity (SP), and Area Under the Curve (AUC). The study was registered with the PROSPERO, CRD42022383386.

RESULTS

Thirty studies were included in the systematic review, sixteen of which were meta-analyzed with a pooled sensitivity of 87% (95%CI 83-91%), specificity of 94% (92-96%), and AUC of 97% (95-98%). Satisfactory diagnostic performance was observed in subgroup analyses based on algorithms types (machine learning versus deep learning, and whether transfer learning was applied), sample size (≤ 200 or >  200), clinicians versus AI models and geographical distribution of institutions (Asia versus non-Asia).

CONCLUSIONS

Even if possible overestimation and further studies with a better standards for application of AI algorithms in lymphoma detection are needed, we suggest the AI may be useful in lymphoma diagnosis.

Associating Peritoneal Metastasis With T2-Weighted MRI Images in Epithelial Ovarian Cancer Using Deep Learning and Radiomics: A Multicenter Study

BACKGROUND

The preoperative diagnosis of peritoneal metastasis (PM) in epithelial ovarian cancer (EOC) is challenging and can impact clinical decision-making.

PURPOSE

To investigate the performance of T2 -weighted (T2W) MRI-based deep learning (DL) and radiomics methods for PM evaluation in EOC patients.

STUDY TYPE

Retrospective.

POPULATION

Four hundred seventy-nine patients from five centers, including one training set (N = 297 [mean, 54.87 years]), one internal validation set (N = 75 [mean, 56.67 years]), and two external validation sets (N = 53 [mean, 55.58 years] and N = 54 [mean, 58.22 years]).

FIELD STRENGTH/SEQUENCE

1.5 or 3 T/fat-suppression T2W fast or turbo spin-echo sequence.

ASSESSMENT

ResNet-50 was used as the architecture of DL. The largest orthogonal slices of the tumor area, radiomics features, and clinical characteristics were used to construct the DL, radiomics, and clinical models, respectively. The three models were combined using decision-level fusion to create an ensemble model. Diagnostic performances of radiologists and radiology residents with and without model assistance were evaluated.

STATISTICAL TESTS

Receiver operating characteristic analysis was used to assess the performances of models. The McNemar test was used to compare sensitivity and specificity. A two-tailed P < 0.05 was considered significant.

RESULTS

The ensemble model had the best AUCs, outperforming the DL model (0.844 vs. 0.743, internal validation set; 0.859 vs. 0.737, external validation set I) and clinical model (0.872 vs. 0.730, external validation set II). After model assistance, all readers had significantly improved sensitivity, especially for those with less experience (junior radiologist1, from 0.639 to 0.820; junior radiologist2, from 0.689 to 0.803; resident1, from 0.623 to 0.803; resident2, from 0.541 to 0.738). One resident also had significantly improved specificity (from 0.633 to 0.789).

DATA CONCLUSIONS

T2W MRI-based DL and radiomics approaches have the potential to preoperatively predict PM in EOC patients and assist in clinical decision-making.

EVIDENCE LEVEL

4 TECHNICAL EFFICACY: Stage 2.

Comparison of MRI radiomics-based machine learning survival models in predicting prognosis of glioblastoma multiforme

Objective

To compare the performance of radiomics-based machine learning survival models in predicting the prognosis of glioblastoma multiforme (GBM) patients.

Methods

131 GBM patients were included in our study. The traditional Cox proportional-hazards (CoxPH) model and four machine learning models (SurvivalTree, Random survival forest (RSF), DeepSurv, DeepHit) were constructed, and the performance of the five models was evaluated using the C-index.

Results

After the screening, 1792 radiomics features were obtained. Seven radiomics features with the strongest relationship with prognosis were obtained following the application of the least absolute shrinkage and selection operator (LASSO) regression. The CoxPH model demonstrated that age (HR = 1.576, p = 0.037), Karnofsky performance status (KPS) score (HR = 1.890, p = 0.006), radiomics risk score (HR = 3.497, p = 0.001), and radiomics risk level (HR = 1.572, p = 0.043) were associated with poorer prognosis. The DeepSurv model performed the best among the five models, obtaining C-index of 0.882 and 0.732 for the training and test set, respectively. The performances of the other four models were lower: CoxPH (0.663 training set / 0.635 test set), SurvivalTree (0.702/0.655), RSF (0.735/0.667), DeepHit (0.608/0.560).

Conclusion

This study confirmed the superior performance of deep learning algorithms based on radiomics relative to the traditional method in predicting the overall survival of GBM patients; specifically, the DeepSurv model showed the best predictive ability.

Predicting glioblastoma molecular subtypes and prognosis with a multimodal model integrating convolutional neural network, radiomics, and semantics

OBJECTIVE

The aim of this study was to build a convolutional neural network (CNN)-based prediction model of glioblastoma (GBM) molecular subtype diagnosis and prognosis with multimodal features.

METHODS

In total, 222 GBM patients were included in the training set from Sun Yat-sen University Cancer Center (SYSUCC) and 107 GBM patients were included in the validation set from SYSUCC, Xuanwu Hospital Capital Medical University, and the First Hospital of Jilin University. The multimodal model was trained with MR images (pre- and postcontrast T1-weighted images and T2-weighted images), corresponding MRI impression, and clinical patient information. First, the original images were segmented using the Multimodal Brain Tumor Image Segmentation Benchmark toolkit. Convolutional features were extracted using 3D residual deep neural network (ResNet50) and convolutional 3D (C3D). Radiomic features were extracted using pyradiomics. Report texts were converted to word embedding using word2vec. These three types of features were then integrated to train neural networks. Accuracy, precision, recall, and F1-score were used to evaluate the model performance.

RESULTS

The C3D-based model yielded the highest accuracy of 91.11% in the prediction of IDH1 mutation status. Importantly, the addition of semantics improved precision by 11.21% and recall in MGMT promoter methylation status prediction by 14.28%. The areas under the receiver operating characteristic curves of the C3D-based model in the IDH1, ATRX, MGMT, and 1-year prognosis groups were 0.976, 0.953, 0.955, and 0.976, respectively. In external validation, the C3D-based model showed significant improvement in accuracy in the IDH1, ATRX, MGMT, and 1-year prognosis groups, which were 88.30%, 76.67%, 85.71%, and 85.71%, respectively (compared with 3D ResNet50: 83.51%, 66.67%, 82.14%, and 70.79%, respectively).

CONCLUSIONS

The authors propose a novel multimodal model integrating C3D, radiomics, and semantics, which had a great performance in predicting IDH1, ATRX, and MGMT molecular subtypes and the 1-year prognosis of GBM.

Hybrid Models Based on Fusion Features of a CNN and Handcrafted Features for Accurate Histopathological Image Analysis for Diagnosing Malignant Lymphomas

Malignant lymphoma is one of the most severe types of disease that leads to death as a result of exposure of lymphocytes to malignant tumors. The transformation of cells from indolent B-cell lymphoma to B-cell lymphoma (DBCL) is life-threatening. Biopsies taken from the patient are the gold standard for lymphoma analysis. Glass slides under a microscope are converted into whole slide images (WSI) to be analyzed by AI techniques through biomedical image processing. Because of the multiplicity of types of malignant lymphomas, manual diagnosis by pathologists is difficult, tedious, and subject to disagreement among physicians. The importance of artificial intelligence (AI) in the early diagnosis of malignant lymphoma is significant and has revolutionized the field of oncology. The use of AI in the early diagnosis of malignant lymphoma offers numerous benefits, including improved accuracy, faster diagnosis, and risk stratification. This study developed several strategies based on hybrid systems to analyze histopathological images of malignant lymphomas. For all proposed models, the images and extraction of malignant lymphocytes were optimized by the gradient vector flow (GVF) algorithm. The first strategy for diagnosing malignant lymphoma images relied on a hybrid system between three types of deep learning (DL) networks, XGBoost algorithms, and decision tree (DT) algorithms based on the GVF algorithm. The second strategy for diagnosing malignant lymphoma images was based on fusing the features of the MobileNet-VGG16, VGG16-AlexNet, and MobileNet-AlexNet models and classifying them by XGBoost and DT algorithms based on the ant colony optimization (ACO) algorithm. The color, shape, and texture features, which are called handcrafted features, were extracted by four traditional feature extraction algorithms. Because of the similarity in the biological characteristics of early-stage malignant lymphomas, the features of the fused MobileNet-VGG16, VGG16-AlexNet, and MobileNet-AlexNet models were combined with the handcrafted features and classified by the XGBoost and DT algorithms based on the ACO algorithm. We concluded that the performance of the two networks XGBoost and DT, with fused features between DL networks and handcrafted, achieved the best performance. The XGBoost network based on the fused features of MobileNet-VGG16 and handcrafted features resulted in an AUC of 99.43%, accuracy of 99.8%, precision of 99.77%, sensitivity of 99.7%, and specificity of 99.8%. This highlights the significant role of AI in the early diagnosis of malignant lymphoma, offering improved accuracy, expedited diagnosis, and enhanced risk stratification. This study highlights leveraging AI techniques and biomedical image processing; the analysis of whole slide images (WSI) converted from biopsies allows for improved accuracy, faster diagnosis, and risk stratification. The developed strategies based on hybrid systems, combining deep learning networks, XGBoost and decision tree algorithms, demonstrated promising results in diagnosing malignant lymphoma images. Furthermore, the fusion of handcrafted features with features extracted from DL networks enhanced the performance of the classification models.

Deep learning for image classification between primary central nervous system lymphoma and glioblastoma in corpus callosal tumors

Objectives

It can be challenging in some situations to distinguish primary central nervous system lymphoma (PCNSL) from glioblastoma (GBM) based on magnetic resonance imaging (MRI) scans, especially those involving the corpus callosum. The objective of this study was to assess the diagnostic performance of deep learning (DL) models between PCNSLs and GBMs in corpus callosal tumors.

Materials and Methods

The axial T1-weighted gadolinium-enhanced MRI scans of 274 individuals with pathologically confirmed PCNSL (n = 94) and GBM (n = 180) were examined. After image pooling, pre-operative MRI scans were randomly split with an 80/20 procedure into a training dataset (n = 709) and a testing dataset (n = 177) for DL model development. Therefore, the DL model was deployed as a web application and validated with the unseen images (n = 114) and area under the receiver operating characteristic curve (AUC); other outcomes were calculated to assess the discrimination performance.

Results

The first baseline DL model had an AUC of 0.77 for PCNSL when evaluated with unseen images. The 2nd model with ridge regression regularization and the 3rd model with drop-out regularization increased an AUC of 0.83 and 0.84. In addition, the last model with data augmentation yielded an AUC of 0.57.

Conclusion

DL with regularization may provide useful diagnostic information to help doctors distinguish PCNSL from GBM.

Plasma Exosome Analysis for Protein Mutation Identification Using a Combination of Raman Spectroscopy and Deep Learning

Protein mutation detection using liquid biopsy can be simply performed periodically, making it easy to detect the occurrence of newly emerging mutations rapidly. However, it has low diagnostic accuracy since there are more normal proteins than mutated proteins in body fluids. To increase the diagnostic accuracy, we analyzed plasma exosomes using nanoplasmonic spectra and deep learning. Exosomes, a promising biomarker, are abundant in plasma and stably carry intact proteins originating from mother cells. However, the mutated exosomal proteins cannot be detected sensitively because of the subtle changes in their structure. Therefore, we obtained Raman spectra that provide molecular information about structural changes in mutated proteins. To extract the unique features of the protein from complex Raman spectra, we developed a deep-learning classification algorithm with two deep-learning models. Consequently, controls with wild-type proteins and patients with mutated proteins were classified with high accuracy. As a proof of concept, we discriminated the lung cancer patients with mutations in the epidermal growth factor receptor (EGFR), L858R, E19del, L858R + T790M, and E19del + T790M, from controls with an accuracy of 0.93. Moreover, the protein mutation status of the patients with primary (E19del, L858R) and secondary (+T790M) mutations was clearly monitored. Overall, our technique is expected to be applied as a novel method for companion diagnostic and treatment monitoring.

The value of convolutional neural networks-based deep learning model in differential diagnosis of space-occupying brain diseases

Objectives

It is still a challenge to differentiate space-occupying brain lesions such as tumefactive demyelinating lesions (TDLs), tumefactive primary angiitis of the central nervous system (TPACNS), primary central nervous system lymphoma (PCNSL), and brain gliomas. Convolutional neural networks (CNNs) have been used to analyze complex medical data and have proven transformative for image-based applications. It can quickly acquire diseases' radiographic features and correct doctors' diagnostic bias to improve diagnostic efficiency and accuracy. The study aimed to assess the value of CNN-based deep learning model in the differential diagnosis of space-occupying brain diseases on MRI.

Methods

We retrospectively analyzed clinical and MRI data from 480 patients with TDLs (n = 116), TPACNS (n = 64), PCNSL (n = 150), and brain gliomas (n = 150). The patients were randomly assigned to training (n = 240), testing (n = 73), calibration (n = 96), and validation (n = 71) groups. And a CNN-implemented deep learning model guided by clinical experts was developed to identify the contrast-enhanced T1-weighted sequence lesions of these four diseases. We utilized accuracy, sensitivity, specificity, and area under the curve (AUC) to evaluate the performance of the CNN model. The model's performance was then compared to the neuroradiologists' diagnosis.

Results

The CNN model had a total accuracy of 87% which was higher than senior neuroradiologists (74%), and the AUC of TDLs, PCNSL, TPACNS and gliomas were 0.92, 0.92, 0.89 and 0.88, respectively.

Conclusion

The CNN model can accurately identify specific radiographic features of TDLs, TPACNS, PCNSL, and gliomas. It has the potential to be an effective auxiliary diagnostic tool in the clinic, assisting inexperienced clinicians in reducing diagnostic bias and improving diagnostic efficiency.

The application of machine learning to imaging in hematological oncology: A scoping review

Background

Here, we conducted a scoping review to (i) establish which machine learning (ML) methods have been applied to hematological malignancy imaging; (ii) establish how ML is being applied to hematological cancer radiology; and (iii) identify addressable research gaps.

Methods

The review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Extension for Scoping Reviews guidelines. The inclusion criteria were (i) pediatric and adult patients with suspected or confirmed hematological malignancy undergoing imaging (population); (ii) any study using ML techniques to derive models using radiological images to apply to the clinical management of these patients (concept); and (iii) original research articles conducted in any setting globally (context). Quality Assessment of Diagnostic Accuracy Studies 2 criteria were used to assess diagnostic and segmentation studies, while the Newcastle-Ottawa scale was used to assess the quality of observational studies.

Results

Of 53 eligible studies, 33 applied diverse ML techniques to diagnose hematological malignancies or to differentiate them from other diseases, especially discriminating gliomas from primary central nervous system lymphomas (n=18); 11 applied ML to segmentation tasks, while 9 applied ML to prognostication or predicting therapeutic responses, especially for diffuse large B-cell lymphoma. All studies reported discrimination statistics, but no study calculated calibration statistics. Every diagnostic/segmentation study had a high risk of bias due to their case-control design; many studies failed to provide adequate details of the reference standard; and only a few studies used independent validation.

Conclusion

To deliver validated ML-based models to radiologists managing hematological malignancies, future studies should (i) adhere to standardized, high-quality reporting guidelines such as the Checklist for Artificial Intelligence in Medical Imaging; (ii) validate models in independent cohorts; (ii) standardize volume segmentation methods for segmentation tasks; (iv) establish comprehensive prospective studies that include different tumor grades, comparisons with radiologists, optimal imaging modalities, sequences, and planes; (v) include side-by-side comparisons of different methods; and (vi) include low- and middle-income countries in multicentric studies to enhance generalizability and reduce inequity.

Progress of radiological‑pathological workflows in the differential diagnosis between primary central nervous system lymphoma and high‑grade glioma (Review)

Primary central nervous system lymphoma (PCNSL) and high‑grade glioma (HGG) are distinct entities of the CNS with completely distinct treatments. The treatment of PCNSL is chemotherapy‑based, while surgery is the first choice for HGG. However, the clinical features of the two entities often overlap, and a clear pathological diagnosis is important for subsequent management, especially for the management of PCNSL. Stereotactic biopsy is recognized as one of the minimally invasive alternatives for evaluating the involvement of the CNS. However, in the case of limited tissue materials, the differential diagnosis between the two entities is still difficult. In addition, some patients are too ill to tolerate a needle biopsy. Therefore, combining imaging, histopathology and laboratory examinations is essential in order to make a clear diagnosis as soon as possible. The present study reviews the progress of comparative research on both imaging and laboratory tests based on the pathophysiological changes of the two entities, and proposes an integrative and optimized diagnostic process, with the purpose of building a better understanding for neurologists, hematologists, radiologists and pathologists.

Machine Learning and Deep Learning CT-Based Models for Predicting the Primary Central Nervous System Lymphoma and Glioma Types: A Multicenter Retrospective Study

Purpose and Background

Distinguishing primary central nervous system lymphoma (PCNSL) and glioma on computed tomography (CT) is an important task since treatment options differ vastly from the two diseases. This study aims to explore various machine learning and deep learning methods based on radiomic features extracted from CT scans and end-to-end convolutional neural network (CNN) model to predict PCNSL and glioma types and compare the performance of different models.

Methods

A total of 101 patients from five Chinese medical centers with pathologically confirmed PCNSL and glioma were analyzed retrospectively, including 50 PCNSL and 51 glioma. After manual segmentation of the region of interest (ROI) on CT scans, 293 radiomic features of each patient were extracted. The radiomic features were used as input, and then, we established six machine learning models and one deep learning model and three readers to identify the two types of tumors. We also established a 2D CNN model using raw CT scans as input. The area under the receiver operating characteristic curve (AUC) and accuracy (ACC) were used to evaluate different models.

Results

The cohort was split into a training (70, 70% patients) and validation cohort (31,30% patients) according to the stratified sampling strategy. Among all models, the MLP performed best, with an accuracy of 0.886 and 0.903, sensitivity of 0.914 and 0.867, specificity of 0.857 and 0.937, and AUC of 0.957 and 0.908 in the training and validation cohorts, respectively, which was significantly higher than the three primary physician's diagnoses (ACCs ranged from 0.710 to 0.742, p &lt; 0.001 for all) and comparable with the senior radiologist (ACC 0.839, p = 0.988). Among all the machine learning models, the AUC ranged from 0.605 to 0.821 in the validation cohort. The end-to-end CNN model achieved an AUC of 0.839 and an ACC of 0.840 in the validation cohort, which had no significant difference in accuracy compared to the MLP model (p = 0.472) and the senior radiologist (p = 0.470).

Conclusion

The established PCNSL and glioma prediction model based on deep neural network methods from CT scans or radiomic features are feasible and provided high performance, which shows the potential to assist clinical decision-making.

Deep learning-based relapse prediction of neuromyelitis optica spectrum disorder with anti-aquaporin-4 antibody

Objective

We previously identified the independent predictors of recurrent relapse in neuromyelitis optica spectrum disorder (NMOSD) with anti-aquaporin-4 antibody (AQP4-ab) and designed a nomogram to estimate the 1- and 2-year relapse-free probability, using the Cox proportional hazard (Cox-PH) model, assuming that the risk of relapse had a linear correlation with clinical variables. However, whether the linear assumption fits real disease tragedy is unknown. We aimed to employ deep learning and machine learning to develop a novel prediction model of relapse in patients with NMOSD and compare the performance with the conventional Cox-PH model.

Methods

This retrospective cohort study included patients with NMOSD with AQP4-ab in 10 study centers. In this study, 1,135 treatment episodes from 358 patients in Huashan Hospital were employed as the training set while 213 treatment episodes from 92 patients in nine other research centers as the validation set. We compared five models with added variables of gender, AQP4-ab titer, previous attack under the same therapy, EDSS score at treatment initiation, maintenance therapy, age at treatment initiation, disease duration, the phenotype of the most recent attack, and annualized relapse rate (ARR) of the most recent year by concordance index (C-index): conventional Cox-PH, random survival forest (RSF), LogisticHazard, DeepHit, and DeepSurv.

Results

When including all variables, RSF outperformed the C-index in the training set (0.739), followed by DeepHit (0.737), LogisticHazard (0.722), DeepSurv (0.698), and Cox-PH (0.679) models. As for the validation set, the C-index of LogisticHazard outperformed the other models (0.718), followed by DeepHit (0.704), DeepSurv (0.698), RSF (0.685), and Cox-PH (0.651) models. Maintenance therapy was calculated to be the most important variable for relapse prediction.

Conclusion

This study confirmed the superiority of deep learning to design a prediction model of relapse in patients with AQP4-ab-positive NMOSD, with the LogisticHazard model showing the best predictive power in validation.

Differentiation of Benign From Malignant Parotid Gland Tumors Using Conventional MRI Based on Radiomics Nomogram

Objectives

We aimed to develop and validate radiomic nomograms to allow preoperative differentiation between benign- and malignant parotid gland tumors (BPGT and MPGT, respectively), as well as between pleomorphic adenomas (PAs) and Warthin tumors (WTs).

Materials and Methods

This retrospective study enrolled 183 parotid gland tumors (68 PAs, 62 WTs, and 53 MPGTs) and divided them into training (n = 128) and testing (n = 55) cohorts. In total, 2553 radiomics features were extracted from fat-saturated T2-weighted images, apparent diffusion coefficient maps, and contrast-enhanced T1-weighted images to construct single-, double-, and multi-sequence combined radiomics models, respectively. The radiomics score (Rad-score) was calculated using the best radiomics model and clinical features to develop the radiomics nomogram. The receiver operating characteristic curve and area under the curve (AUC) were used to assess these models, and their performances were compared using DeLong’s test. Calibration curves and decision curve analysis were used to assess the clinical usefulness of these models.

Results

The multi-sequence combined radiomics model exhibited better differentiation performance (BPGT vs. MPGT, AUC=0.863; PA vs. MPGT, AUC=0.929; WT vs. MPGT, AUC=0.825; PA vs. WT, AUC=0.927) than the single- and double sequence radiomics models. The nomogram based on the multi-sequence combined radiomics model and clinical features attained an improved classification performance (BPGT vs. MPGT, AUC=0.907; PA vs. MPGT, AUC=0.961; WT vs. MPGT, AUC=0.879; PA vs. WT, AUC=0.967).

Conclusions

Radiomics nomogram yielded excellent diagnostic performance in differentiating BPGT from MPGT, PA from MPGT, and PA from WT.

A Systematic Review of the Current Status and Quality of Radiomics for Glioma Differential Diagnosis

Simple Summary

Gliomas can be difficult to discern clinically and radiologically from other brain lesions (either neoplastic or non-neoplastic) since their clinical manifestations as well as preoperative imaging features often overlap and appear misleading. Radiomics could be extremely helpful for non-invasive glioma differential diagnosis (DDx). However, implementation in clinical practice is still distant and concerns have been raised regarding the methodological quality of radiomic studies. In this context, we aimed to summarize the current status and quality of radiomic studies concerning glioma DDx in a systematic review. In total, 42 studies were selected and examined in our work. Our study revealed that, despite promising and encouraging results, current studies on radiomics for glioma DDx still lack the quality required to allow its introduction into clinical practice. This work could provide new insights and help to reach a consensus on the use of the radiomic approach for glioma DDx.

Abstract

Radiomics is a promising tool that may increase the value of imaging in differential diagnosis (DDx) of glioma. However, implementation in clinical practice is still distant and concerns have been raised regarding the methodological quality of radiomic studies. Therefore, we aimed to systematically review the current status of radiomic studies concerning glioma DDx, also using the radiomics quality score (RQS) to assess the quality of the methodology used in each study. A systematic literature search was performed to identify original articles focused on the use of radiomics for glioma DDx from 2015. Methodological quality was assessed using the RQS tool. Spearman’s correlation (ρ) analysis was performed to explore whether RQS was correlated with journal metrics and the characteristics of the studies. Finally, 42 articles were selected for the systematic qualitative analysis. Selected articles were grouped and summarized in terms of those on DDx between glioma and primary central nervous system lymphoma, those aiming at differentiating glioma from brain metastases, and those based on DDx of glioma and other brain diseases. Median RQS was 8.71 out 36, with a mean RQS of all studies of 24.21%. Our study revealed that, despite promising and encouraging results, current studies on radiomics for glioma DDx still lack the quality required to allow its introduction into clinical practice. This work could provide new insights and help to reach a consensus on the use of the radiomic approach for glioma DDx.

Construction of an Artificial Intelligence Writing Model for English Based on Fusion Neural Network Model

This paper presents an in-depth study and analysis of the model of English writing using artificial intelligence algorithms of neural networks. Based on word vectors, the unsupervised disambiguation, and clustering of multimedia contexts extracted from massive online videos, the disambiguation accuracy reaches over 0.7, and the resulting small-scale multimedia context set can cover up to 90% of vocabulary learning tasks; user experiments show that the multimedia context learning system based on this method can improve the effectiveness and experience of ESL vocabulary learning, as well as the long-term word sense memory of learners. The results are 30% better. Based on the dependency grammatical relations and semantic metrics of collocations on a large-scale professional corpus, we established a collocation intention description and retrieval method in line with users' linguistic cognition and doubled the usage rate of collocation retrieval on the actual deployment system after half a year, becoming a user "sticky" ESL writing aid, and further defined style. Dictionaries only provide basic lexical definitions, and, even if supported by example sentences, they still cannot meet the needs of ESL authors in terms of expressive accuracy and richness. However, the current machine translation is based on the black box deep neural network construction, and its translation process is not understandable and interactive. Among the three algorithmic models constructed in this paper, the multitask learning model outperforms the conditional random field model and the LSTM-CRF model because the multitask learning model with auxiliary tasks solves the problem of sparse data to a certain extent, allowing the model to be trained more adequately in the case of uneven label distribution, and thus performs better than other models in the task of grammatical error detection.

Neuroplasticity of Glioma Patients: Brain Structure and Topological Network

Glioma is the most common primary malignant brain tumor in adults. It accounts for about 75% of such tumors and occurs more commonly in men. The incidence rate has been increasing in the past 30 years. Moreover, the 5-year overall survival rate of glioma patients is < 35%. Different locations, grades, and molecular characteristics of gliomas can lead to different behavioral deficits and prognosis, which are closely related to patients' quality of life and associated with neuroplasticity. Some advanced magnetic resonance imaging (MRI) technologies can explore the neuroplasticity of structural, topological, biochemical metabolism, and related mechanisms, which may contribute to the improvement of prognosis and function in glioma patients. In this review, we summarized the studies conducted on structural and topological plasticity of glioma patients through different MRI technologies and discussed future research directions. Previous studies have found that glioma itself and related functional impairments can lead to structural and topological plasticity using multimodal MRI. However, neuroplasticity caused by highly heterogeneous gliomas is not fully understood, and should be further explored through multimodal MRI. In addition, the individualized prediction of functional prognosis of glioma patients from the functional level based on machine learning (ML) is promising. These approaches and the introduction of ML can further shed light on the neuroplasticity and related mechanism of the brain, which will be helpful for management of glioma patients.

Clinical explainable differential diagnosis of polypoidal choroidal vasculopathy and age-related macular degeneration using deep learning

BACKGROUND

This study aims to achieve an automatic differential diagnosis between two types of retinal pathologies with similar pathological features - Polypoidal choroidal vasculopathy (PCV) and wet age-related macular degeneration (AMD) from volumetric optical coherence tomography (OCT) images, and identify clinically-relevant pathological features, using an explainable deep-learning-based framework.

METHODS

This is a retrospective study with data from a cross-sectional cohort. The OCT volume of 73 eyes from 59 patients was included in this study. Disease differentiation was achieved through single-B-scan-based classification followed by a volumetric probability prediction aggregation step. We compared different labeling strategies with and without identifying pathological B-scans within each OCT volume. Clinical interpretability was achieved through normalized aggregation of B-scan-based saliency maps followed by maximum-intensity-projection onto the en face plane. We derived the PCV score from the proposed differential diagnosis framework with different labeling strategies. The en face projection of saliency map was validated with the pathologies identified in Indocyanine green angiography (ICGA).

RESULTS

Model trained with both labeling strategies achieved similar level differentiation power (>90%), with good correspondence between pathological features detected from the projected en face saliency map and ICGA.

CONCLUSIONS

This study demonstrated the potential clinical application of non-invasive differential diagnosis using AI-driven OCT-based analysis, with minimal requirement of labeling efforts, along with clinical explainability achieved through automatically detected disease-related pathologies.

A Deep Learning Model for Preoperative Differentiation of Glioblastoma, Brain Metastasis and Primary Central Nervous System Lymphoma: A Pilot Study

Background

Neuroimaging differentiation of glioblastoma, primary central nervous system lymphoma (PCNSL) and solitary brain metastasis (BM) remains challenging in specific cases showing similar appearances or atypical features. Overall, advanced MRI protocols have high diagnostic reliability, but their limited worldwide availability, coupled with the overlapping of specific neuroimaging features among tumor subgroups, represent significant drawbacks and entail disparities in the planning and management of these oncological patients.

Objective

To evaluate the classification performance metrics of a deep learning algorithm trained on T1-weighted gadolinium-enhanced (T1Gd) MRI scans of glioblastomas, atypical PCNSLs and BMs.

Materials and Methods

We enrolled 121 patients (glioblastoma: n=47; PCNSL: n=37; BM: n=37) who had undergone preoperative T1Gd-MRI and histopathological confirmation. Each lesion was segmented, and all ROIs were exported in a DICOM dataset. The patient cohort was then split in a training and hold-out test sets following a 70/30 ratio. A Resnet101 model, a deep neural network (DNN), was trained on the training set and validated on the hold-out test set to differentiate glioblastomas, PCNSLs and BMs on T1Gd-MRI scans.

Results

The DNN achieved optimal classification performance in distinguishing PCNSLs (AUC: 0.98; 95%CI: 0.95 - 1.00) and glioblastomas (AUC: 0.90; 95%CI: 0.81 - 0.97) and moderate ability in differentiating BMs (AUC: 0.81; 95%CI: 0.70 - 0.95). This performance may allow clinicians to correctly identify patients eligible for lesion biopsy or surgical resection.

Conclusion

We trained and internally validated a deep learning model able to reliably differentiate ambiguous cases of PCNSLs, glioblastoma and BMs by means of T1Gd-MRI. The proposed predictive model may provide a low-cost, easily-accessible and high-speed decision-making support for eligibility to diagnostic brain biopsy or maximal tumor resection in atypical cases.

Beyond Glioma: The Utility of Radiomic Analysis for Non-Glial Intracranial Tumors

Simple Summary

Tumor qualities, such as growth rate, firmness, and intrusion into healthy tissue, can be very important for operation planning and further treatment. Radiomics is a promising new method that allows the determination of some of these qualities on images performed before surgery. In this article, we provide a review of the use of radiomics in various tumors of the central nervous system, such as metastases, lymphoma, meningioma, medulloblastoma, and pituitary tumors.

Abstract

The field of radiomics is rapidly expanding and gaining a valuable role in neuro-oncology. The possibilities related to the use of radiomic analysis, such as distinguishing types of malignancies, predicting tumor grade, determining the presence of particular molecular markers, consistency, therapy response, and prognosis, can considerably influence decision-making in medicine in the near future. Even though the main focus of radiomic analyses has been on glial CNS tumors, studies on other intracranial tumors have shown encouraging results. Therefore, as the main focus of this review, we performed an analysis of publications on PubMed and Web of Science databases, focusing on radiomics in CNS metastases, lymphoma, meningioma, medulloblastoma, and pituitary tumors.

Classification of glioblastoma versus primary central nervous system lymphoma using convolutional neural networks

A subset of primary central nervous system lymphomas (PCNSL) are difficult to distinguish from glioblastoma multiforme (GBM) on magnetic resonance imaging (MRI). We developed a convolutional neural network (CNN) to distinguish these tumors on contrast-enhanced T₁-weighted images. Preoperative brain tumor MRIs were retrospectively collected among 320 patients with either GBM (n = 160) and PCNSL (n = 160) from two academic institutions. The individual images from these MRIs consisted of a training set (n = 1894 GBM and 1245 PCNSL), a validation set (n = 339 GBM; 202 PCNSL), and a testing set (99 GBM and 108 PCNSL). Three CNNs using the EfficientNetB4 architecture were evaluated. To increase the size of the training set and minimize overfitting, random flips and changes to color were performed on the training set. Our transfer learning approach (with image augmentation and 292 epochs) yielded an AUC of 0.94 (95% CI: 0.91–0.97) for GBM and an AUC of 0.95 (95% CI: 0.92–0.98) for PCNL. In the second case (not augmented and 137 epochs), the images were augmented prior to training. The area under the curve for GBM was 0.92 (95% CI: 0.88–0.96) for GBM and an AUC of 0.94 (95% CI: 0.91–0.97) for PCNSL. For the last case (augmented, Gaussian noise and 238 epochs) the AUC for GBM was 0.93 (95% CI: 0.89–0.96) and an AUC 0.93 (95% CI = 0.89–0.96) for PCNSL. Even with a relatively small dataset, our transfer learning approach demonstrated CNNs may provide accurate diagnostic information to assist radiologists in distinguishing PCNSL and GBM.

Radiomic Based Machine Learning Performance for a Three Class Problem in Neuro-Oncology: Time to Test the Waters?

Simple Summary

Prior radiomic studies have addressed a two-class tumor classification problem (glioblastoma (GBM) versus primary CNS lymphoma (PCNSL) or GBM versus metastasis). However, this approach is prone to bias and excludes other common brain tumor types. We addressed a real-life clinical problem by including the three most common brain tumor types (GBM, PCNSL, and metastasis). We investigated two key issues using different MRI sequence combinations: performance variation based on tumor subregions (necrotic, enhancing, edema and combined enhancing, and necrotic masks), and performance metrics based on the chosen classifier model/feature selection combination. Our study provides evidence that radiomics-based three-class tumor differentiation is feasible, and that embedded models perform better than those with a priori feature selection. We found that T1 contrast enhanced is the single best sequence with comparable performance to that of multiparametric MRI, and model performance varies based on tumor subregion and the combination of model/feature selection methods.

Abstract

Prior radiomics studies have focused on two-class brain tumor classification, which limits generalizability. The performance of radiomics in differentiating the three most common malignant brain tumors (glioblastoma (GBM), primary central nervous system lymphoma (PCNSL), and metastatic disease) is assessed; factors affecting the model performance and usefulness of a single sequence versus multiparametric MRI (MP-MRI) remain largely unaddressed. This retrospective study included 253 patients (120 metastatic (lung and brain), 40 PCNSL, and 93 GBM). Radiomic features were extracted for whole a tumor mask (enhancing plus necrotic) and an edema mask (first pipeline), as well as for separate enhancing and necrotic and edema masks (second pipeline). Model performance was evaluated using MP-MRI, individual sequences, and the T1 contrast enhanced (T1-CE) sequence without the edema mask across 45 model/feature selection combinations. The second pipeline showed significantly high performance across all combinations (Brier score: 0.311–0.325). GBRM fit using the full feature set from the T1-CE sequence was the best model. The majority of the top models were built using a full feature set and inbuilt feature selection. No significant difference was seen between the top-performing models for MP-MRI (AUC 0.910) and T1-CE sequence with (AUC 0.908) and without edema masks (AUC 0.894). T1-CE is the single best sequence with comparable performance to that of multiparametric MRI (MP-MRI). Model performance varies based on tumor subregion and the combination of model/feature selection methods.

Radiomics-Based Differentiation between Glioblastoma, CNS Lymphoma, and Brain Metastases: Comparing Performance across MRI Sequences and Machine Learning Models

Prior radiomics studies have focused on two-class brain tumor classification, which limits generalizability. The performance of radiomics in differentiating the three most common malignant brain tumors (glioblastoma (GBM), primary central nervous system lymphoma (PCNSL), and metastatic disease) is assessed; factors affecting the model performance and usefulness of a single sequence versus multiparametric MRI (MP-MRI) remain largely unaddressed. This retrospective study included 253 patients (120 metastatic (lung and brain), 40 PCNSL, and 93 GBM). Radiomic features were extracted for whole a tumor mask (enhancing plus necrotic) and an edema mask (first pipeline), as well as for separate enhancing and necrotic and edema masks (second pipeline). Model performance was evaluated using MP-MRI, individual sequences, and the T1 contrast enhanced (T1-CE) sequence without the edema mask across 45 model/feature selection combinations. The second pipeline showed significantly high performance across all combinations (Brier score: 0.311–0.325). GBRM fit using the full feature set from the T1-CE sequence was the best model. The majority of the top models were built using a full feature set and inbuilt feature selection. No significant difference was seen between the top-performing models for MP-MRI (AUC 0.910) and T1-CE sequence with (AUC 0.908) and without edema masks (AUC 0.894). T1-CE is the single best sequence with comparable performance to that of multiparametric MRI (MP-MRI). Model performance varies based on tumor subregion and the combination of model/feature selection methods.