Performance evaluation of radiologists with artificial neural network for differential diagnosis of intra-axial cerebral tumors on MR images

Artificial Intelligence and Deep Learning in Revolutionizing Brain Tumor Diagnosis and Treatment: A Narrative Review

The emergence of artificial intelligence (AI) in the medical field holds promise in improving medical management, particularly in personalized strategies for the diagnosis and treatment of brain tumors. However, integrating AI into clinical practice has proven to be a challenge. Deep learning (DL) is very convenient for extracting relevant information from large amounts of data that has increased in medical history and imaging records, which shortens diagnosis time, that would otherwise overwhelm manual methods. In addition, DL aids in automated tumor segmentation, classification, and diagnosis. DL models such as the Brain Tumor Classification Model and the Inception-Resnet V2, or hybrid techniques that enhance these functions and combine DL networks with support vector machine and k-nearest neighbors, identify tumor phenotypes and brain metastases, allowing real-time decision-making and enhancing preoperative planning. AI algorithms and DL development facilitate radiological diagnostics such as computed tomography, positron emission tomography scans, and magnetic resonance imaging (MRI) by integrating two-dimensional and three-dimensional MRI using DenseNet and 3D convolutional neural network architectures, which enable precise tumor delineation. DL offers benefits in neuro-interventional procedures, and the shift toward computer-assisted interventions acknowledges the need for more accurate and efficient image analysis methods. Further research is needed to realize the potential impact of DL in improving these outcomes.

Artificial neural network identified a 20-gene panel in predicting immunotherapy response and survival benefits after anti-PD1/PD-L1 treatment in glioblastoma patients

BACKGROUND

Immune checkpoint inhibitors (ICIs) are a promising immunotherapy approach, but glioblastoma clinical trials have not yielded satisfactory results.

OBJECTIVE

To screen glioblastoma patients who may benefit from immunotherapy.

METHODS

Eighty-one patients receiving anti-PD1/PD-L1 treatment from a large-scale clinical trial and 364 patients without immunotherapy from The Cancer Genome Atlas (TCGA) were included. Patients in the ICI-treated cohort were divided into responders and nonresponders according to overall survival (OS), and the most critical responder-relevant features were screened using random forest (RF). We constructed an artificial neural network (ANN) model and verified its predictive value with immunotherapy response and OS.

RESULTS

We defined two groups of ICI-treated glioblastoma patients with large differences in survival benefits as nonresponders (OS ≤6 months, n = 18) and responders (OS ≥17 months, n = 8). No differentially mutated genes were observed between responders and nonresponders. We performed RF analysis to select the most critical responder-relevant features and developed an ANN with 20 input variables, five hidden neurons and one output neuron. Receiver operating characteristic analysis and the DeLong test demonstrated that the ANN had the best performance in predicting responders, with an AUC of 0.97. Survival analysis indicated that ANN-predicted responders had significantly better OS rates than nonresponders.

CONCLUSION

The 20-gene panel developed by the ANN could be a promising biomarker for predicting immunotherapy response and prognostic benefits in ICI-treated GBM patients and may guide oncologists to accurately select potential responders for the preferential use of ICIs.

Deep learning aided preoperative diagnosis of primary central nervous system lymphoma

The preoperative distinction between glioblastoma (GBM) and primary central nervous system lymphoma (PCNSL) can be difficult, even for experts, but is highly relevant. We aimed to develop an easy-to-use algorithm, based on a convolutional neural network (CNN) to preoperatively discern PCNSL from GBM and systematically compare its performance to experienced neurosurgeons and radiologists. To this end, a CNN-based on DenseNet169 was trained with the magnetic resonance (MR)-imaging data of 68 PCNSL and 69 GBM patients and its performance compared to six trained experts on an external test set of 10 PCNSL and 10 GBM. Our neural network predicted PCNSL with an accuracy of 80% and a negative predictive value (NPV) of 0.8, exceeding the accuracy achieved by clinicians (73%, NPV 0.77). Combining expert rating with automated diagnosis in those cases where experts dissented yielded an accuracy of 95%. Our approach has the potential to significantly augment the preoperative radiological diagnosis of PCNSL.

Systematic reviews of machine learning in healthcare: a literature review

INTRODUCTION

The increasing availability of data and computing power has made machine learning (ML) a viable approach to faster, more efficient healthcare delivery.

METHODS

A systematic literature review (SLR) of published SLRs evaluating ML applications in healthcare settings published between1 January 2010 and 27 March 2023 was conducted.

RESULTS

In total 220 SLRs covering 10,462 ML algorithms were reviewed. The main application of AI in medicine related to the clinical prediction and disease prognosis in oncology and neurology with the use of imaging data. Accuracy, specificity, and sensitivity were provided in 56%, 28%, and 25% SLRs respectively. Internal and external validation was reported in 53% and less than 1% of the cases respectively. The most common modeling approach was neural networks (2,454 ML algorithms), followed by support vector machine and random forest/decision trees (1,578 and 1,522 ML algorithms, respectively).

EXPERT OPINION

The review indicated considerable reporting gaps in terms of the ML's performance, both internal and external validation. Greater accessibility to healthcare data for developers can ensure the faster adoption of ML algorithms into clinical practice.

Advances in artificial intelligence, robotics, augmented and virtual reality in neurosurgery

Neurosurgical practitioners undergo extensive and prolonged training to acquire diverse technical proficiencies, while neurosurgical procedures necessitate a substantial amount of pre-, post-, and intraoperative clinical data acquisition, making decisions, attention, and convalescence. The past decade witnessed an appreciable escalation in the significance of artificial intelligence (AI) in neurosurgery. AI holds significant potential in neurosurgery as it supplements the abilities of neurosurgeons to offer optimal interventional and non-interventional care to patients by improving prognostic and diagnostic outcomes in clinical therapy and assisting neurosurgeons in making decisions while surgical interventions to enhance patient outcomes. Other technologies including augmented reality, robotics, and virtual reality can assist and promote neurosurgical methods as well. Moreover, they play a significant role in generating, processing, as well as storing experimental and clinical data. Also, the usage of these technologies in neurosurgery is able to curtail the number of costs linked with surgical care and extend high-quality health care to a wider populace. This narrative review aims to integrate the results of articles that elucidate the role of the aforementioned technologies in neurosurgery.

Systematic review of the performance evaluation of clinicians with or without the aid of machine learning clinical decision support system

Background

For the adoption of machine learning clinical decision support systems (ML-CDSS) it is critical to understand the performance aid of the ML-CDSS. However, it is not trivial, how the performance aid should be evaluated. To design reliable performance evaluation study, both the knowledge from the practical framework of experimental study design and the understanding of domain specific design factors are required.

Objective

The aim of this review study was to form a practical framework and identify key design factors for experimental design in evaluating the performance of clinicians with or without the aid of ML-CDSS.

Methods

The study was based on published ML-CDSS performance evaluation studies. We systematically searched articles published between January 2016 and December 2022. From the articles we collected a set of design factors. Only the articles comparing the performance of clinicians with or without the aid of ML-CDSS using experimental study methods were considered.

Results

The identified key design factors for the practical framework of ML-CDSS experimental study design were performance measures, user interface, ground truth data and the selection of samples and participants. In addition, we identified the importance of randomization, crossover design and training and practice rounds. Previous studies had shortcomings in the rationale and documentation of choices regarding the number of participants and the duration of the experiment.

Conclusion

The design factors of ML-CDSS experimental study are interdependent and all factors must be considered in individual choices.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12553-023-00763-1.

Fully automated radiomics-based machine learning models for multiclass classification of single brain tumors: Glioblastoma, lymphoma, and metastasis

BACKGROUND AND PURPOSE

To investigate the diagnostic performance of fully automated radiomics-based models for multiclass classification of a single enhancing brain tumor among glioblastoma, central nervous system lymphoma, and metastasis.

MATERIALS AND METHODS

The training and test sets were comprised of 538 cases (300 glioblastomas, 73 lymphomas, and 165 metastases) and 169 cases (101 glioblastomas, 29 lymphomas, and 39 metastases), respectively. After fully automated segmentation, radiomic features were extracted. Three conventional machine learning classifiers, including least absolute shrinkage and selection operator (LASSO), adaptive boosting (Adaboost), and support vector machine with the linear kernel (SVC), combined with one of four feature selection methods, including forward sequential feature selection, F score, mutual information, and LASSO, were trained. Additionally, one ensemble classifier based on the three classifiers was used. The diagnostic performance of the optimized models was tested in the test set using the accuracy, F1-macro score, and the area under the receiver operating characteristic curve (AUCROC).

RESULTS

The best performance was achieved when the LASSO was used as a feature selection method. In the test set, the best performance was achieved by the ensemble classifier, showing an accuracy of 76.3% (95% CI, 70.0-82.7), a F1-macro score of 0.704, and an AUCROC of 0.878.

CONCLUSION

Our fully automated radiomics-based models for multiclass classification might be useful for differential diagnosis of a single enhancing brain tumor with a good diagnostic performance and generalizability.

Machine Learning in Differentiating Gliomas from Primary CNS Lymphomas: A Systematic Review, Reporting Quality, and Risk of Bias Assessment

BACKGROUND

Differentiating gliomas and primary CNS lymphoma represents a diagnostic challenge with important therapeutic ramifications. Biopsy is the preferred method of diagnosis, while MR imaging in conjunction with machine learning has shown promising results in differentiating these tumors.

PURPOSE

Our aim was to evaluate the quality of reporting and risk of bias, assess data bases with which the machine learning classification algorithms were developed, the algorithms themselves, and their performance.

DATA SOURCES

Ovid EMBASE, Ovid MEDLINE, Cochrane Central Register of Controlled Trials, and the Web of Science Core Collection were searched according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.

STUDY SELECTION

From 11,727 studies, 23 peer-reviewed studies used machine learning to differentiate primary CNS lymphoma from gliomas in 2276 patients.

DATA ANALYSIS

Characteristics of data sets and machine learning algorithms were extracted. A meta-analysis on a subset of studies was performed. Reporting quality and risk of bias were assessed using the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) and Prediction Model Study Risk Of Bias Assessment Tool.

DATA SYNTHESIS

The highest area under the receiver operating characteristic curve (0.961) and accuracy (91.2%) in external validation were achieved by logistic regression and support vector machines models using conventional radiomic features. Meta-analysis of machine learning classifiers using these features yielded a mean area under the receiver operating characteristic curve of 0.944 (95% CI, 0.898-0.99). The median TRIPOD score was 51.7%. The risk of bias was high for 16 studies.

LIMITATIONS

Exclusion of abstracts decreased the sensitivity in evaluating all published studies. Meta-analysis had high heterogeneity.

CONCLUSIONS

Machine learning-based methods of differentiating primary CNS lymphoma from gliomas have shown great potential, but most studies lack large, balanced data sets and external validation. Assessment of the studies identified multiple deficiencies in reporting quality and risk of bias. These factors reduce the generalizability and reproducibility of the findings.

Machine Learning Applications for Differentiation of Glioma from Brain Metastasis-A Systematic Review

Simple Summary

We present a systematic review of published reports on machine learning (ML) applications for the differentiation of gliomas from brain metastases by summarizing study characteristics, strengths, and pitfalls. Based on these findings, we present recommendations for future research in this field.

Abstract

Glioma and brain metastasis can be difficult to distinguish on conventional magnetic resonance imaging (MRI) due to the similarity of imaging features in specific clinical circumstances. Multiple studies have investigated the use of machine learning (ML) models for non-invasive differentiation of glioma from brain metastasis. Many of the studies report promising classification results, however, to date, none have been implemented into clinical practice. After a screening of 12,470 studies, we included 29 eligible studies in our systematic review. From each study, we aggregated data on model design, development, and best classifiers, as well as quality of reporting according to the TRIPOD statement. In a subset of eligible studies, we conducted a meta-analysis of the reported AUC. It was found that data predominantly originated from single-center institutions (n = 25/29) and only two studies performed external validation. The median TRIPOD adherence was 0.48, indicating insufficient quality of reporting among surveyed studies. Our findings illustrate that despite promising classification results, reliable model assessment is limited by poor reporting of study design and lack of algorithm validation and generalizability. Therefore, adherence to quality guidelines and validation on outside datasets is critical for the clinical translation of ML for the differentiation of glioma and brain metastasis.

A Brief History of Machine Learning in Neurosurgery

The history of machine learning in neurosurgery spans three decades and continues to develop at a rapid pace. The earliest applications of machine learning within neurosurgery were first published in the 1990s as researchers began developing artificial neural networks to analyze structured datasets and supervised tasks. By the turn of the millennium, machine learning had evolved beyond proof-of-concept; algorithms had success detecting tumors in unstructured clinical imaging, and unsupervised learning showed promise for tumor segmentation. Throughout the 2000s, the role of machine learning in neurosurgery was further refined. Well-trained models began to consistently best expert clinicians at brain tumor diagnosis. Additionally, the digitization of the healthcare industry provided ample data for analysis, both structured and unstructured. By the 2010s, the use of machine learning within neurosurgery had exploded. The rapid deployment of an exciting new toolset also led to the growing realization that it may offer marginal benefit at best over conventional logistical regression models for analyzing tabular datasets. Additionally, the widespread adoption of machine learning in neurosurgical clinical practice continues to lag until additional validation can ensure generalizability. Many exciting contemporary applications nonetheless continue to demonstrate the unprecedented potential of machine learning to revolutionize neurosurgery when applied to appropriate clinical challenges.

Neurosurgery and artificial intelligence

Neurosurgeons receive extensive and lengthy training to equip themselves with various technical skills, and neurosurgery require a great deal of pre-, intra- and postoperative clinical data collection, decision making, care and recovery. The last decade has seen a significant increase in the importance of artificial intelligence (AI) in neurosurgery. AI can provide a great promise in neurosurgery by complementing neurosurgeons' skills to provide the best possible interventional and noninterventional care for patients by enhancing diagnostic and prognostic outcomes in clinical treatment and help neurosurgeons with decision making during surgical interventions to improve patient outcomes. Furthermore, AI is playing a pivotal role in the production, processing and storage of clinical and experimental data. AI usage in neurosurgery can also reduce the costs associated with surgical care and provide high-quality healthcare to a broader population. Additionally, AI and neurosurgery can build a symbiotic relationship where AI helps to push the boundaries of neurosurgery, and neurosurgery can help AI to develop better and more robust algorithms. This review explores the role of AI in interventional and noninterventional aspects of neurosurgery during pre-, intra- and postoperative care, such as diagnosis, clinical decision making, surgical operation, prognosis, data acquisition, and research within the neurosurgical arena.

Differentiating Glioblastomas from Solitary Brain Metastases: An Update on the Current Literature of Advanced Imaging Modalities

Differentiating between glioblastomas and solitary brain metastases proves to be a challenging diagnosis for neuroradiologists, as both present with imaging patterns consisting of peritumoral hyperintensities with similar intratumoral texture on traditional magnetic resonance imaging sequences. Early diagnosis is paramount, as each pathology has completely different methods of clinical assessment. In the past decade, recent developments in advanced imaging modalities enabled providers to acquire a more accurate diagnosis earlier in the patient's clinical assessment, thus optimizing clinical outcome. Dynamic susceptibility contrast has been optimized for detecting relative cerebral blood flow and relative cerebral blood volume. Diffusion tensor imaging can be used to detect changes in mean diffusivity. Neurite orientation dispersion and density imaging is an innovative modality detecting changes in intracellular volume fraction, isotropic volume fraction, and extracellular volume fraction. Magnetic resonance spectroscopy is able to assist by providing a metabolic descriptor while detecting variable ratios of choline/N-acetylaspartate, choline/creatine, and N-acetylaspartate/creatine. Finally, radiomics and machine learning algorithms have been devised to assist in improving diagnostic accuracy while often utilizing more than one advanced imaging protocol per patient. In this review, we provide an update on all the current evidence regarding the identification and differentiation of glioblastomas from solitary brain metastases.

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

BACKGROUND

Differential diagnosis of primary central nervous system lymphoma (PCNSL) and glioblastoma (GBM) is useful to guide treatment strategies.

PURPOSE

To investigate the use of a convolutional neural network (CNN) model for differentiation of PCNSL and GBM without tumor delineation.

STUDY TYPE

Retrospective.

POPULATION

A total of 289 patients with PCNSL (136) or GBM (153) were included, the average age of the cohort was 54 years, and there were 173 men and 116 women.

FIELD STRENGTH/SEQUENCE

3.0 T Axial contrast-enhanced T₁ -weighted spin-echo inversion recovery sequence (CE-T₁ WI), T₂ -weighted fluid-attenuation inversion recovery sequence (FLAIR), and diffusion weighted imaging (DWI, b = 0 second/mm² , 1000 seconds/mm² ).

ASSESSMENT

A single-parametric CNN model was built using CE-T₁ WI, FLAIR, and the apparent diffusion coefficient (ADC) map derived from DWI, respectively. A decision-level fusion based multi-parametric CNN model (DF-CNN) was built by combining the predictions of single-parametric CNN models through logistic regression. An image-level fusion based multi-parametric CNN model (IF-CNN) was built using the integrated multi-parametric MR images. The radiomics models were developed. The diagnoses by three radiologists with 6 years (junior radiologist Y.Y.), 11 years (intermediate-level radiologist Y.T.), and 21 years (senior radiologist Y.L.) of experience were obtained.

STATISTICAL ANALYSIS

The 5-fold cross validation was used for model evaluation. The Pearson's chi-squared test was used to compare the accuracies. U-test and Fisher's exact test were used to compare clinical characteristics.

RESULTS

The CE-T₁ WI, FLAIR, and ADC based single-parametric CNN model had accuracy of 0.884, 0.782, and 0.700, respectively. The DF-CNN model had an accuracy of 0.899 which was higher than the IF-CNN model (0.830, P = 0.021), but had no significant difference in accuracy compared to the radiomics model (0.865, P = 0.255), and the senior radiologist (0.906, P = 0.886).

DATA CONCLUSION

A CNN model can differentiate PCNSL from GBM without tumor delineation, and comparable to the radiomics models and radiologists.

LEVEL OF EVIDENCE

4 TECHNICAL EFFICACY: Stage 2.

Promises and Perils of Artificial Intelligence in Neurosurgery

Artificial intelligence (AI)-facilitated clinical automation is expected to become increasingly prevalent in the near future. AI techniques may permit rapid and detailed analysis of the large quantities of clinical data generated in modern healthcare settings, at a level that is otherwise impossible by humans. Subsequently, AI may enhance clinical practice by pushing the limits of diagnostics, clinical decision making, and prognostication. Moreover, if combined with surgical robotics and other surgical adjuncts such as image guidance, AI may find its way into the operating room and permit more accurate interventions, with fewer errors. Despite the considerable hype surrounding the impending medical AI revolution, little has been written about potential downsides to increasing clinical automation. These may include both direct and indirect consequences. Directly, faulty, inadequately trained, or poorly understood algorithms may produce erroneous results, which may have wide-scale impact. Indirectly, increasing use of automation may exacerbate de-skilling of human physicians due to over-reliance, poor understanding, overconfidence, and lack of necessary vigilance of an automated clinical workflow. Many of these negative phenomena have already been witnessed in other industries that have already undergone, or are undergoing "automation revolutions," namely commercial aviation and the automotive industry. This narrative review explores the potential benefits and consequences of the anticipated medical AI revolution from a neurosurgical perspective.

Machine learning applications for the differentiation of primary central nervous system lymphoma from glioblastoma on imaging: a systematic review and meta-analysis

OBJECTIVEGlioblastoma (GBM) and primary central nervous system lymphoma (PCNSL) are common intracranial pathologies encountered by neurosurgeons. They often may have similar radiological findings, making diagnosis difficult without surgical biopsy; however, management is quite different between these two entities. Recently, predictive analytics, including machine learning (ML), have garnered attention for their potential to aid in the diagnostic assessment of a variety of pathologies. Several ML algorithms have recently been designed to differentiate GBM from PCNSL radiologically with a high sensitivity and specificity. The objective of this systematic review and meta-analysis was to evaluate the implementation of ML algorithms in differentiating GBM and PCNSL.METHODSThe authors performed a systematic review of the literature using PubMed in accordance with PRISMA guidelines to select and evaluate studies that included themes of ML and brain tumors. These studies were further narrowed down to focus on works published between January 2008 and May 2018 addressing the use of ML in training models to distinguish between GBM and PCNSL on radiological imaging. Outcomes assessed were test characteristics such as accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC).RESULTSEight studies were identified addressing use of ML in training classifiers to distinguish between GBM and PCNSL on radiological imaging. ML performed well with the lowest reported AUC being 0.878. In studies in which ML was directly compared with radiologists, ML performed better than or as well as the radiologists. However, when ML was applied to an external data set, it performed more poorly.CONCLUSIONSFew studies have applied ML to solve the problem of differentiating GBM from PCNSL using imaging alone. Of the currently published studies, ML algorithms have demonstrated promising results and certainly have the potential to aid radiologists with difficult cases, which could expedite the neurosurgical decision-making process. It is likely that ML algorithms will help to optimize neurosurgical patient outcomes as well as the cost-effectiveness of neurosurgical care if the problem of overfitting can be overcome.

Use of artificial neural network for pretreatment verification of intensity modulation radiation therapy fields

OBJECTIVE

The accuracy of dose delivery for intensity modulated radiotherapy (IMRT) treatments should be determined by an accurate quality assurance procedure. In this work, we used artificial neural networks (ANNs) as an application for the pre-treatment dose verification of IMRT fields based two-dimensional-fluence maps acquired by an electronic portal imaging device (EPID).

METHODS

The ANN must be trained and validated before use for the pretreatment dose verification. Hence, 60 EPID fluence maps of the anteroposterior prostate and nasopharynx IMRT fields were used as an input for the ANN (feed forward type), and a dose map of those fluence maps that were acquired by two-dimensional Array Seven29TM as an output for the ANN.

RESULTS

After the training and validation of the neural network, the analysis of 20 IMRT anteroposterior fields showed excellent agreement between the ANN output and the dose map predicted by the treatment planning system. The average overall global and local γ field pass rate was greater than 90% for the prostate and nasopharynx fields, with the 2 mm/3% criteria.

CONCLUSION

The results indicated that the ANN can be used as a fast and powerful tool for pretreatment dose verification, based on an EPID fluence map.

ADVANCES IN KNOWLEDGE

In this study, ANN is proposed for EPID based dose validation of IMRT fields. The proposed method has good accuracy and high speed in response to problems. Neural network show to be low price and precise method for IMRT fields verification.

Natural and Artificial Intelligence in Neurosurgery: A Systematic Review

BACKGROUND

Machine learning (ML) is a domain of artificial intelligence that allows computer algorithms to learn from experience without being explicitly programmed.

OBJECTIVE

To summarize neurosurgical applications of ML where it has been compared to clinical expertise, here referred to as "natural intelligence."

METHODS

A systematic search was performed in the PubMed and Embase databases as of August 2016 to review all studies comparing the performance of various ML approaches with that of clinical experts in neurosurgical literature.

RESULTS

Twenty-three studies were identified that used ML algorithms for diagnosis, presurgical planning, or outcome prediction in neurosurgical patients. Compared to clinical experts, ML models demonstrated a median absolute improvement in accuracy and area under the receiver operating curve of 13% (interquartile range 4-21%) and 0.14 (interquartile range 0.07-0.21), respectively. In 29 (58%) of the 50 outcome measures for which a P-value was provided or calculated, ML models outperformed clinical experts (P < .05). In 18 of 50 (36%), no difference was seen between ML and expert performance (P > .05), while in 3 of 50 (6%) clinical experts outperformed ML models (P < .05). All 4 studies that compared clinicians assisted by ML models vs clinicians alone demonstrated a better performance in the first group.

CONCLUSION

We conclude that ML models have the potential to augment the decision-making capacity of clinicians in neurosurgical applications; however, significant hurdles remain associated with creating, validating, and deploying ML models in the clinical setting. Shifting from the preconceptions of a human-vs-machine to a human-and-machine paradigm could be essential to overcome these hurdles.

Machine learning for semi-automated classification of glioblastoma, brain metastasis and central nervous system lymphoma using magnetic resonance advanced imaging

Background

Differentiating glioblastoma, brain metastasis, and central nervous system lymphoma (CNSL) on conventional magnetic resonance imaging (MRI) can present a diagnostic dilemma due to the potential for overlapping imaging features. We investigate whether machine learning evaluation of multimodal MRI can reliably differentiate these entities.

Methods

Preoperative brain MRI including diffusion weighted imaging (DWI), dynamic contrast enhanced (DCE), and dynamic susceptibility contrast (DSC) perfusion in patients with glioblastoma, lymphoma, or metastasis were retrospectively reviewed. Perfusion maps (rCBV, rCBF), permeability maps (K-trans, Kep, Vp, Ve), ADC, T1C+ and T2/FLAIR images were coregistered and two separate volumes of interest (VOIs) were obtained from the enhancing tumor and non-enhancing T2 hyperintense (NET2) regions. The tumor volumes obtained from these VOIs were utilized for supervised training of support vector classifier (SVC) and multilayer perceptron (MLP) models. Validation of the trained models was performed on unlabeled cases using the leave-one-subject-out method. Head-to-head and multiclass models were created. Accuracies of the multiclass models were compared against two human interpreters reviewing conventional and diffusion-weighted MR images.

Results

Twenty-six patients enrolled with histopathologically-proven glioblastoma (n=9), metastasis (n=9), and CNS lymphoma (n=8) were included. The trained multiclass ML models discriminated the three pathologic classes with a maximum accuracy of 69.2% accuracy (18 out of 26; kappa 0.540, P=0.01) using an MLP trained with the VpNET2 tumor volumes. Human readers achieved 65.4% (17 out of 26) and 80.8% (21 out of 26) accuracies, respectively. Using the MLP VpNET2 model as a computer-aided diagnosis (CADx) for cases in which the human reviewers disagreed with each other on the diagnosis resulted in correct diagnoses in 5 (19.2%) additional cases.

Conclusions

Our trained multiclass MLP using VpNET2 can differentiate glioblastoma, brain metastasis, and CNS lymphoma with modest diagnostic accuracy and provides approximately 19% increase in diagnostic yield when added to routine human interpretation.

2D Dose Reconstruction by Artificial Neural Network for Pretreatment Verification of IMRT Fields

The use of intensity-modulated radiation therapy (IMRT) is developing rapidly in clinical routines. Because of the high complexity and uniqueness of IMRT treatment plans, patient-specific pretreatment quality assurance is generally considered a necessary prerequisite for patient treatment. In this work, we proposed a modified methodology of electronic portal imaging device (EPID)-based dose validation for pretreatment verification of IMRT fields by applying artificial neural networks (ANNs). The ANN must be trained and validated before use for pretreatment dose verification. For this purpose, 20 EPID fluence maps of IMRT prostate anterior-posterior fields were used as an input for ANN (feed forward type) and a dose map of those fluence maps that were predicted by treatment planning system as an output for ANN. After the training and validation of the neural network, the analysis of 10 IMRT prostate anterior-posterior fields showed excellent agreement between ANN output and dose map predicted by the treatment planning system. The average overall fields pass rate was 96.0% ± 0.1% with 3 mm/3% criteria. The results indicated that the ANN can be used as a low-cost, fast, and powerful tool for pretreatment dose verification, based on an EPID fluence map.

Computerized Classification of Pneumoconiosis on Digital Chest Radiography Artificial Neural Network with Three Stages

It is difficult for radiologists to classify pneumoconiosis from category 0 to category 3 on chest radiographs. Therefore, we have developed a computer-aided diagnosis (CAD) system based on a three-stage artificial neural network (ANN) method for classification based on four texture features. The image database consists of 36 chest radiographs classified as category 0 to category 3. Regions of interest (ROIs) with a matrix size of 32 × 32 were selected from chest radiographs. We obtained a gray-level histogram, histogram of gray-level difference, gray-level run-length matrix (GLRLM) feature image, and gray-level co-occurrence matrix (GLCOM) feature image in each ROI. For ROI-based classification, the first ANN was trained with each texture feature. Next, the second ANN was trained with output patterns obtained from the first ANN. Finally, we obtained a case-based classification for distinguishing among four categories with the third ANN method. We determined the performance of the third ANN by receiver operating characteristic (ROC) analysis. The areas under the ROC curve (AUC) of the highest category (severe pneumoconiosis) case and the lowest category (early pneumoconiosis) case were 0.89 ± 0.09 and 0.84 ± 0.12, respectively. The three-stage ANN with four texture features showed the highest performance for classification among the four categories. Our CAD system would be useful for assisting radiologists in classification of pneumoconiosis from category 0 to category 3.

Evaluation of apparent diffusion coefficient thresholds for diagnosis of medulloblastoma using diffusion-weighted imaging

We assess a diffusion-weighted imaging (DWI) analysis technique as a potential basis for computer-aided diagnosis (CAD) of pediatric posterior fossa tumors. A retrospective medical record search identified 103 children (mean age: 87 months) with posterior fossa tumors having a total of 126 preoperative MR scans with DWI. The minimum ADC (ADCmin) and normalized ADC (nADC) values [ratio of ADCmin values in tumor compared to normal tissue] were measured by a single observer blinded to diagnosis. Receiver operating characteristic (ROC) curves were generated to determine the optimal threshold for which the nADC and ADCmin values would predict tumor histology. Inter-rater reliability for predicting tumor type was evaluated using values measured by two additional observers. At histology, ten tumor types were identified, with astrocytoma (n=50), medulloblastoma (n=33), and ependymoma (n=9) accounting for 89%. Mean ADCmin (0.54 × 10(-3) mm(2)/s) and nADC (0.70) were lowest for medulloblastoma. Mean ADCmin (1.28 × 10(-3) mm(2)/s) and nADC (1.64) were highest for astrocytoma. For the ROC analysis, the area under the curve when discriminating medulloblastoma from other tumors using nADC was 0.939 and 0.965 when using ADCmin. The optimal ADCmin threshold was 0.66 × 10(-3) mm(2)/s, which yielded an 86% positive predictive value, 97% negative predictive value, and 93% accuracy. Inter-observer variability was very low, with near perfect agreement among all observers in predicting medulloblastoma. Our data indicate that both ADCmin and nADC could serve as the basis for a CAD program to distinguish medulloblastoma from other posterior fossa tumors with a high degree of accuracy.

Classification of cerebral lymphomas and glioblastomas featuring luminance distribution analysis

Differentiating lymphomas and glioblastomas is important for proper treatment planning. A number of works have been proposed but there are still some problems. For example, many works depend on thresholding a single feature value, which is susceptible to noise. In other cases, experienced observers are required to extract the feature values or to provide some interactions with the system. Even if experts are involved, interobserver variance becomes another problem. In addition, most of the works use only one or a few slice(s) because 3D tumor segmentation is time consuming. In this paper, we propose a tumor classification system that analyzes the luminance distribution of the whole tumor region. Typical cases are classified by the luminance range thresholding and the apparent diffusion coefficients (ADC) thresholding. Nontypical cases are classified by a support vector machine (SVM). Most of the processing elements are semiautomatic. Therefore, even novice users can use the system easily and get the same results as experts. The experiments were conducted using 40 MRI datasets. The classification accuracy of the proposed method was 91.1% without the ADC thresholding and 95.4% with the ADC thresholding. On the other hand, the baseline method, the conventional ADC thresholding, yielded only 67.5% accuracy.

Toward IMRT 2D dose modeling using artificial neural networks: a feasibility study

PURPOSE

To investigate the feasibility of artificial neural networks (ANN) to reconstruct dose maps for intensity modulated radiation treatment (IMRT) fields compared with those of the treatment planning system (TPS).

METHODS

An artificial feed forward neural network and the back-propagation learning algorithm have been used to replicate dose calculations of IMRT fields obtained from PINNACLE(3) v9.0. The ANN was trained with fluence and dose maps of IMRT fields for 6 MV x-rays, which were obtained from the amorphous silicon (a-Si) electronic portal imaging device of Novalis TX. Those fluence distributions were imported to the TPS and the dose maps were calculated on the horizontal midpoint plane of a water equivalent homogeneous cylindrical virtual phantom. Each exported 2D dose distribution from the TPS was classified into two clusters of high and low dose regions, respectively, based on the K-means algorithm and the Euclidian metric in the fluence-dose domain. The data of each cluster were divided into two sets for the training and validation phase of the ANN, respectively. After the completion of the ANN training phase, 2D dose maps were reconstructed by the ANN and isodose distributions were created. The dose maps reconstructed by ANN were evaluated and compared with the TPS, where the mean absolute deviation of the dose and the γ-index were used.

RESULTS

A good agreement between the doses calculated from the TPS and the trained ANN was achieved. In particular, an average relative dosimetric difference of 4.6% and an average γ-index passing rate of 93% were obtained for low dose regions, and a dosimetric difference of 2.3% and an average γ-index passing rate of 97% for high dose region.

CONCLUSIONS

An artificial neural network has been developed to convert fluence maps to corresponding dose maps. The feasibility and potential of an artificial neural network to replicate complex convolution kernels in the TPS for IMRT dose calculations have been demonstrated.

Esophageal stenosis associated with tumor regression in radiotherapy for esophageal cancer: frequency and prediction

PURPOSE

To determine clinical factors for predicting the frequency and severity of esophageal stenosis associated with tumor regression in radiotherapy for esophageal cancer.

METHODS AND MATERIALS

The study group consisted of 109 patients with esophageal cancer of T1-4 and Stage I-III who were treated with definitive radiotherapy and achieved a complete response of their primary lesion at Kyushu University Hospital between January 1998 and December 2007. Esophageal stenosis was evaluated using esophagographic images within 3 months after completion of radiotherapy. We investigated the correlation between esophageal stenosis after radiotherapy and each of the clinical factors with regard to tumors and therapy. For validation of the correlative factors for esophageal stenosis, an artificial neural network was used to predict the esophageal stenotic ratio.

RESULTS

Esophageal stenosis tended to be more severe and more frequent in T3-4 cases than in T1-2 cases. Esophageal stenosis in cases with full circumference involvement tended to be more severe and more frequent than that in cases without full circumference involvement. Increases in wall thickness tended to be associated with increases in esophageal stenosis severity and frequency. In the multivariate analysis, T stage, extent of involved circumference, and wall thickness of the tumor region were significantly correlated to esophageal stenosis (p = 0.031, p < 0.0001, and p = 0.0011, respectively). The esophageal stenotic ratio predicted by the artificial neural network, which learned these three factors, was significantly correlated to the actual observed stenotic ratio, with a correlation coefficient of 0.864 (p < 0.001).

CONCLUSION

Our study suggested that T stage, extent of involved circumference, and esophageal wall thickness of the tumor region were useful to predict the frequency and severity of esophageal stenosis associated with tumor regression in radiotherapy for esophageal cancer.

Medical image analysis with artificial neural networks

Given that neural networks have been widely reported in the research community of medical imaging, we provide a focused literature survey on recent neural network developments in computer-aided diagnosis, medical image segmentation and edge detection towards visual content analysis, and medical image registration for its pre-processing and post-processing, with the aims of increasing awareness of how neural networks can be applied to these areas and to provide a foundation for further research and practical development. Representative techniques and algorithms are explained in detail to provide inspiring examples illustrating: (i) how a known neural network with fixed structure and training procedure could be applied to resolve a medical imaging problem; (ii) how medical images could be analysed, processed, and characterised by neural networks; and (iii) how neural networks could be expanded further to resolve problems relevant to medical imaging. In the concluding section, a highlight of comparisons among many neural network applications is included to provide a global view on computational intelligence with neural networks in medical imaging.

Artificial neural networks for prediction of response to chemoradiation in HT29 xenografts

PURPOSE

To evaluate the feasibility of using neural networks for predicting treatment response by using longitudinal measurements of apparent diffusion coefficient (ADC) obtained from diffusion-weighted magnetic resonance imaging (DWMRI).

METHODS AND MATERIALS

Mice bearing HT29 xenografts were allocated to six treatment groups receiving different combinations of daily chemotherapy and/or radiation therapy for 2 weeks. T(2)-weighted and DWMR images were acquired before treatment, twice during fractionated chemoradiation (at days 4 and 11), and four times after treatment ended (at days 18, 25, 32, and 46). A tumor doubling growth delay (T(delay)) value was found for individual xenografts. ADC values and treatment groups (1-6) were used as input to a back propagation neural network (BPNN) to predict T(delay).

RESULTS

When treatment group and ADC values from days 0, 4, 11, 18, 25, 32, and 46 were used as inputs to the BPNN, a strong correlation between measured and predicted T(delay) values was found (R = 0.731, p < 0.01). When ADC values from days 0, 4, and 11, and the treatment group were used as inputs, the correlation between predicted and measured T(delay) was 0.693 (p < 0.01).

CONCLUSIONS

BPNN was successfully used to predict T(delay) from tumor ADC values obtained from HT29 xenografts undergoing fractionated chemoradiation therapy.