bims-arihec 2020-04-12 papers

bims-arihec

Biomed News

on Artificial intelligence in healthcare

Issue of 2020‒04‒12
nineteen papers selected by
Céline Bélanger
Cogniges Inc.

Connecting Data to Insight: A Pan-Canadian Study on AI in Healthcare.
Machine intelligence in healthcare-perspectives on trustworthiness, explainability, usability, and transparency.
Artificial Intelligence and Liver Transplant: Predicting Survival of Individual Grafts.
Artificial Intelligence in Gastrointestinal Endoscopy.
Radiomics: A primer for the radiation oncologist.
A study of aortic dissection screening method based on multiple machine learning models.
Breast Cancer Detection and Classification using Traditional Computer Vision Techniques: A Comprehensive Review.
The Impact of Deep Convoluted Neural Network based Artificial Intelligence on Colonoscopy Outcomes: A Systematic Review with Meta-analysis.
Impact of Image Analysis and Artificial Intelligence in Thyroid Pathology, with particular reference to cytological aspects.
Opening the black box of artificial intelligence for clinical decision support: A study predicting stroke outcome.
Development and Validation of the Automated Imaging Differentiation in Parkinsonism (AID-P): A Multi-Site Machine Learning Study.
Artificial Intelligence System Approaching Neuroradiologist-level Differential Diagnosis Accuracy at Brain MRI.
Topics and trends in artificial intelligence assisted human brain research.
Efficacy for Differentiating Nonglaucomatous versus Glaucomatous Optic Neuropathy Using Deep Learning Systems.
Artificial Intelligence in Ophthalmology: A Meta-Analysis of Deep Learning Models for Retinal Vessels Segmentation.
Multiphase CT-based prediction of Child-Pugh classification: a machine learning approach.
Machine Learning and Deep Neural Network Applications in the Thorax: Pulmonary Embolism, Chronic Thromboembolic Pulmonary Hypertension, Aorta, and Chronic Obstructive Pulmonary Disease.
Development of Machine Learning Algorithms to Predict Clinically Meaningful Improvement for the Patient-Reported Health State After Total Hip Arthroplasty.
Machine Learning in Dermatology: Current Applications, Opportunities, and Limitations.

Healthc Q. 2020 Apr;pii: hcq.2020.26144. [Epub ahead of print]23(1): 13-19

Connecting Data to Insight: A Pan-Canadian Study on AI in Healthcare.

Zaki Hakim, Terra Ierasts, Irfan Hakim, Andre D'Penha, Kevin P D Smith, Michael C W Caesar.

Across Canada, healthcare leaders are exploring the potential of artificial intelligence and advanced analytics to transform the healthcare system. This report shares a summary of the current state of healthcare analytics across major hospitals and public healthcare agencies in Canada. We present information on the current level of investment, data governance maturity, analytics talent and tools and models being leveraged across the nation. The findings point to an opportunity for enhanced collaboration in advanced analytics and the adoption of nascent artificial intelligence technologies in healthcare. The recommendations will help drive adoption in Canada, ultimately improving the patient experience and promoting better health outcomes for Canadians.

DOI: https://doi.org/10.12927/hcq.2020.26144
NPJ Digit Med. 2020 ;3 47

Machine intelligence in healthcare-perspectives on trustworthiness, explainability, usability, and transparency.

Christine M Cutillo, Karlie R Sharma, Luca Foschini, Shinjini Kundu, Maxine Mackintosh, Kenneth D Mandl, .

  Machine Intelligence (MI) is rapidly becoming an important approach across biomedical discovery, clinical research, medical diagnostics/devices, and precision medicine. Such tools can uncover new possibilities for researchers, physicians, and patients, allowing them to make more informed decisions and achieve better outcomes. When deployed in healthcare settings, these approaches have the potential to enhance efficiency and effectiveness of the health research and care ecosystem, and ultimately improve quality of patient care. In response to the increased use of MI in healthcare, and issues associated when applying such approaches to clinical care settings, the National Institutes of Health (NIH) and National Center for Advancing Translational Sciences (NCATS) co-hosted a Machine Intelligence in Healthcare workshop with the National Cancer Institute (NCI) and the National Institute of Biomedical Imaging and Bioengineering (NIBIB) on 12 July 2019. Speakers and attendees included researchers, clinicians and patients/ patient advocates, with representation from industry, academia, and federal agencies. A number of issues were addressed, including: data quality and quantity; access and use of electronic health records (EHRs); transparency and explainability of the system in contrast to the entire clinical workflow; and the impact of bias on system outputs, among other topics. This whitepaper reports on key issues associated with MI specific to applications in the healthcare field, identifies areas of improvement for MI systems in the context of healthcare, and proposes avenues and solutions for these issues, with the aim of surfacing key areas that, if appropriately addressed, could accelerate progress in the field effectively, transparently, and ethically.

Keywords:  Diagnosis; Disease prevention; Medical imaging; Public health; Therapeutics

DOI:  https://doi.org/10.1038/s41746-020-0254-2
Liver Transpl. 2020 Apr 09.

Artificial Intelligence and Liver Transplant: Predicting Survival of Individual Grafts.

Laura Wingfield, Carlo Ceresa, Simon Thorogood, Jacques Fleuriot, Simon Knight.

  BACKGROUND: The demand for liver transplantation far outstrips the supply of deceased donor organs, and so listing and allocation decisions aim to maximise utility. Most existing methods for predicting transplant outcomes utilise basic methods such as regression modelling - newer artificial intelligence techniques have the potential to improve predictive accuracy.AIMS: To systematically review studies predicting graft outcomes following deceased liver transplantation using Artificial Intelligence (AI) techniques and comparing these to linear regression and standard predictive modelling (donor risk index, DRI; Model for end-stage liver disease, MELD; survival outcome following liver transplantation, SOFT).
METHODS: A systematic review was performed. PubMed, Cochrane, MEDLINE, Science Direct, Springer Link, Elsevier, and reference lists were analysed for appropriate inclusion.
RESULTS: A total of 52 papers were reviewed for inclusion. Of these papers, 9 met the inclusion criteria, reporting outcomes from 18,771 liver transplants. Artificial neural networks (ANN) were the most commonly utilised methodology, being reported in 7 studies. Only two studies directly compared Machine Learning (ML) techniques to liver scoring modalities (i.e. DRI, SOFT, BAR). Both of these studies showed better prediction of individual organ survival with the optimal ANN model reporting AUC ROC 0.82 compared with BAR: 0.62 and SOFT: 0.57; and the other ANN model showing an AUC ROC: 0.84 compared to DRI: 0.68 and SOFT: 0.64.
CONCLUSION: AI techniques can provide high accuracy in predicting graft survival based on donors and recipient variables. When compared to standard techniques, AI methods are dynamic - able to be trained and validated within every population. However, the high accuracy of AI may come at a cost of losing explainability (to patients and clinicians) on how the technology works.

Keywords:  Machine learning; artificial neural networks; hepatic; organ allocation; patient outcomes

DOI:  https://doi.org/10.1002/lt.25772
Clin Endosc. 2020 Mar;53(2): 132-141

Artificial Intelligence in Gastrointestinal Endoscopy.

Alexander P Abadir, Mohammed Fahad Ali, William Karnes, Jason B Samarasena.

  Artificial intelligence (AI) is rapidly integrating into modern technology and clinical practice. Although in its nascency, AI has become a hot topic of investigation for applications in clinical practice. Multiple fields of medicine have embraced the possibility of a future with AI assisting in diagnosis and pathology applications. In the field of gastroenterology, AI has been studied as a tool to assist in risk stratification, diagnosis, and pathologic identification. Specifically, AI has become of great interest in endoscopy as a technology with substantial potential to revolutionize the practice of a modern gastroenterologist. From cancer screening to automated report generation, AI has touched upon all aspects of modern endoscopy. Here, we review landmark AI developments in endoscopy. Starting with broad definitions to develop understanding, we will summarize the current state of AI research and its potential applications. With innovation developing rapidly, this article touches upon the remarkable advances in AI-assisted endoscopy since its initial evaluation at the turn of the millennium, and the potential impact these AI models may have on the modern clinical practice. As with any discussion of new technology, its limitations must also be understood to apply clinical AI tools successfully.

Keywords:  Artificial intelligence; Colonoscopy; Computer assisted diagnosis; Early detection of cancer; Endoscopy

DOI:  https://doi.org/10.5946/ce.2020.038
Cancer Radiother. 2020 Apr 04. pii: S1278-3218(20)30071-8. [Epub ahead of print]

Radiomics: A primer for the radiation oncologist.

J-E Bibault, L Xing, P Giraud, R El Ayachy, N Giraud, P Decazes, A Burgun, P Giraud.

  PURPOSE: Radiomics are a set of methods used to leverage medical imaging and extract quantitative features that can characterize a patient's phenotype. All modalities can be used with several different software packages. Specific informatics methods can then be used to create meaningful predictive models. In this review, we will explain the major steps of a radiomics analysis pipeline and then present the studies published in the context of radiation therapy.METHODS: A literature review was performed on Medline using the search engine PubMed. The search strategy included the search terms "radiotherapy", "radiation oncology" and "radiomics". The search was conducted in July 2019 and reference lists of selected articles were hand searched for relevance to this review.
RESULTS: A typical radiomics workflow always includes five steps: imaging and segmenting, data curation and preparation, feature extraction, exploration and selection and finally modeling. In radiation oncology, radiomics studies have been published to explore different clinical outcome in lung (n=5), head and neck (n=5), esophageal (n=3), rectal (n=3), pancreatic (n=2) cancer and brain metastases (n=2). The quality of these retrospective studies is heterogeneous and their results have not been translated to the clinic.
CONCLUSION: Radiomics has a great potential to predict clinical outcome and better personalize treatment. But the field is still young and constantly evolving. Improvement in bias reduction techniques and multicenter studies will hopefully allow more robust and generalizable models.

Keywords:  Apprentissage profond; Clinical oncology; Clinique; Deep learning; Machine learning; Modeling; Modélisation; Oncologie; Radiation oncology; Radiomics; Radiomique; Radiothérapie

DOI:  https://doi.org/10.1016/j.canrad.2020.01.011
J Thorac Dis. 2020 Mar;12(3): 605-614

A study of aortic dissection screening method based on multiple machine learning models.

Lijue Liu, Caiwang Zhang, Guogang Zhang, Yan Gao, Jingmin Luo, Wei Zhang, Yi Li, Yang Mu.

  Background: The main purpose of the study was to develop an early screening method for aortic dissection (AD) based on machine learning. Due to the rarity of AD and the complexity of symptoms, many doctors have no clinical experience with it. Many patients are not suspected of having AD, which lead to a high rate of misdiagnosis. Here, we report the preliminary study and feasibility of rapid and accurate screening method of AD with machine learning methods.Methods: The dataset analyzed was composed by examination data provided by the Xiangya Hospital Central South University of China which include a total of 60,000 samples, including aortic patients and non-aortic ones. Each sample has 76 features which is consist of routine examinations and other easily accessible information. Since the proportion of people who are affected is usually imbalanced compared to non-diseased people, multiple machine learning models were used, include AdaBoost, SmoteBagging, EasyEnsemble and CalibratedAdaMEC. They used different methods such as ensemble learning, undersampling, oversampling, and cost-sensitivity to solve data imbalance problems.
Results: AdaBoost performed poorly with an average recall of 16.1% and a specificity of 99.8%. SmoteBagging achieved a statistically significant better performance for this problem with an average recall of 78.1% and a specificity of 79.2%. EasyEnsemble reached the values of 77.8% and 79.3% for recall and specificity respectively. CalibratedAdaMEC's recall and specificity are 75.8% and 76%.
Conclusions: It was found that the screening performance of the models evaluated in this paper had a misdiagnosis rate lower than 25% except AdaBoost. The data used in these methods are only routine inspection data. This means that machine learning methods can help us build a fast, cheap, worthwhile and effective early screening approach for AD.

Keywords:  Aortic dissection (AD); class imbalance; machine learning; screening performance

DOI:  https://doi.org/10.21037/jtd.2019.12.119
Curr Med Imaging. 2020 Apr 06.

Breast Cancer Detection and Classification using Traditional Computer Vision Techniques: A Comprehensive Review.

Saliha Zahoor, Ikram Ullah Lali, Muhammad Attique Khan, Kashif Javed, Waqar Mehmood.

  Breast Cancer is a common dangerous disease for women. In the world, many women died due to Breast cancer. However, in the initial stage, the diagnosis of breast cancer can save women's life. To diagnose cancer in the breast tissues there are several techniques and methods. The image processing, machine learning and deep learning methods and techniques are presented in this paper to diagnose the breast cancer. This work will be helpful to adopt better choices and reliable methods to diagnose breast cancer in an initial stage to survive the women's life. To detect the breast masses, microcalcifications, malignant cells the different techniques are used in the Computer-Aided Diagnosis (CAD) systems phases like preprocessing, segmentation, feature extraction, and classification. We have been reported a detailed analysis of different techniques or methods with their usage and performance measurement. From the reported results, it is concluded that for the survival of women's life it is essential to improve the methods or techniques to diagnose breast cancer at an initial stage by improving the results of the Computer-Aided Diagnosis systems. Furthermore, segmentation and classification phases are challenging for researchers for the diagnosis of breast cancer accurately. Therefore, more advanced tools and techniques are still essential for the accurate diagnosis and classification of breast cancer.

Keywords:  Keywords: Cancer; challenges.; classification; features; segmentation

DOI:  https://doi.org/10.2174/1573405616666200406110547
J Gastroenterol Hepatol. 2020 Apr 08.

The Impact of Deep Convoluted Neural Network based Artificial Intelligence on Colonoscopy Outcomes: A Systematic Review with Meta-analysis.

Muhammad Aziz, Rawish Fatima, Charles Dong, Wade Lee-Smith, Ali Nawras.

  BACKGROUND AND STUDY AIMS: The utility of artificial intelligence (AI) in colonoscopy has gained popularity in current times. Recent trials have evaluated the efficacy of deep convoluted neural network (DCNN) based AI system in colonoscopy for improving adenoma detection rate (ADR) and polyp detection rate (PDR). We performed a systematic review and meta-analysis of the available studies to assess the impact of DCNN based AI assisted colonoscopy in improving the ADR and PDR.METHODS: We queried the following database for this study: PubMed, Embase, Cochrane Library, Web of Sciences, and Computers and Applied Sciences. We only included randomized controlled trials (RCT) that compared AI colonoscopy to standard colonoscopy (SC). Our outcomes included ADR and PDR. Risk ratios (RR) with 95 % confidence interval (CI) were calculated using random effects model and DerSimonian-Laird approach for each outcome.
RESULTS: A total of 3 studies with 2815 patients (1415 in SC group and 1400 in AI group) were included. AI colonoscopy resulted in significantly improved ADR (32.9% vs 20.8%, RR: 1.58, 95 % CI 1.39 - 1.80, p = <0.001) and PDR (43.0% vs 27.8%, RR: 1.55, 95 % CI 1.39 - 1.72, p = <0.001) compared to SC.
CONCLUSION: Given the results and limitations, the utility of AI colonoscopy holds promise and should be evaluated in more RCTs across different population, especially in patients solely undergoing colonoscopy for screening purpose as improved ADR will ultimately help in reducing incident colorectal cancer (CRC).

Keywords:  Artificial intelligence; Deep convoluted neural network; colonoscopy; high-definition

DOI:  https://doi.org/10.1111/jgh.15070
Cytopathology. 2020 Apr 05.

Impact of Image Analysis and Artificial Intelligence in Thyroid Pathology, with particular reference to cytological aspects.

Ilaria Girolami, Stefano Marletta, Liron Pantanowitz, Evelin Torresani, Claudio Ghimenton, Mattia Barbareschis, Aldo Scarpa, Matteo Brunelli, Valeria Barresi, Pierpaolo Trimboli, Albino Eccher.

  OBJECTIVE: Thyroid pathology has great potential for automated/artificial intelligence (AI) algorithm application as the incidence of thyroid nodules is increasing and the indeterminate interpretation rate of fine-needle aspiration remains relatively high. The aim of the study is to review the published literature on automated image analysis and AI applications to thyroid pathology with whole-slide imaging (WSI).METHODS: Systematic search was carried out in electronic databases. Studies dealing with thyroid pathology and use of automated algorithms applied to WSI were included. Quality of studies was assessed with a modified QUADAS-2 tool.
RESULTS: Of 919 retrieved articles, 19 were included. The main themes addressed were the comparison of automated assessment of immunohistochemical staining with manual pathologist's assessment, quantification of differences in cellular and nuclear parameters among tumor entities, and discrimination between benign and malignant nodules. Correlation coefficients with manual assessment were higher than 0.76 and diagnostic performance of automated models was comparable with an expert pathologist diagnosis. Computational difficulties were related to the large size of whole-slide images.
CONCLUSIONS: Overall, the results are promising and it is likely that with the resolution of technical issues the application of automated algorithms in thyroid pathology will increase and be adopted following suitable validation studies.

Keywords:  artificial intelligence; digital pathology; image analysis; systematic review; thyroid; whole-slide imaging

DOI:  https://doi.org/10.1111/cyt.12828
PLoS One. 2020 ;15(4): e0231166

Opening the black box of artificial intelligence for clinical decision support: A study predicting stroke outcome.

Esra Zihni, Vince Istvan Madai, Michelle Livne, Ivana Galinovic, Ahmed A Khalil, Jochen B Fiebach, Dietmar Frey.

State-of-the-art machine learning (ML) artificial intelligence methods are increasingly leveraged in clinical predictive modeling to provide clinical decision support systems to physicians. Modern ML approaches such as artificial neural networks (ANNs) and tree boosting often perform better than more traditional methods like logistic regression. On the other hand, these modern methods yield a limited understanding of the resulting predictions. However, in the medical domain, understanding of applied models is essential, in particular, when informing clinical decision support. Thus, in recent years, interpretability methods for modern ML methods have emerged to potentially allow explainable predictions paired with high performance. To our knowledge, we present in this work the first explainability comparison of two modern ML methods, tree boosting and multilayer perceptrons (MLPs), to traditional logistic regression methods using a stroke outcome prediction paradigm. Here, we used clinical features to predict a dichotomized 90 days post-stroke modified Rankin Scale (mRS) score. For interpretability, we evaluated clinical features' importance with regard to predictions using deep Taylor decomposition for MLP, Shapley values for tree boosting and model coefficients for logistic regression. With regard to performance as measured by Area under the Curve (AUC) values on the test dataset, all models performed comparably: Logistic regression AUCs were 0.83, 0.83, 0.81 for three different regularization schemes; tree boosting AUC was 0.81; MLP AUC was 0.83. Importantly, the interpretability analysis demonstrated consistent results across models by rating age and stroke severity consecutively amongst the most important predictive features. For less important features, some differences were observed between the methods. Our analysis suggests that modern machine learning methods can provide explainability which is compatible with domain knowledge interpretation and traditional method rankings. Future work should focus on replication of these findings in other datasets and further testing of different explainability methods.

DOI: https://doi.org/10.1371/journal.pone.0231166
Lancet Digit Health. 2019 Sep;1(5): e222-e231

Development and Validation of the Automated Imaging Differentiation in Parkinsonism (AID-P): A Multi-Site Machine Learning Study.

Derek B Archer, Justin T Bricker, Winston T Chu, Roxana G Burciu, Johanna L Mccracken, Song Lai, Stephen A Coombes, Ruogu Fang, Angelos Barmpoutis, Daniel M Corcos, Ajay S Kurani, Trina Mitchell, Mieniecia L Black, Ellen Herschel, Tanya Simuni, Todd B Parrish, Cynthia Comella, Tao Xie, Klaus Seppi, Nicolaas I Bohnen, Martijn L T M Müller, Roger L Albin, Florian Krismer, Guangwei Du, Mechelle M Lewis, Xuemei Huang, Hong Li, Ofer Pasternak, Nikolaus R McFarland, Michael S Okun, David E Vaillancourt.

Background: There is a critical need to develop valid, non-invasive biomarkers for Parkinsonian syndromes. The current 17-site, international study assesses whether non-invasive diffusion MRI (dMRI) can distinguish between Parkinsonian syndromes.Methods: We used dMRI from 1002 subjects, along with the Movement Disorders Society Unified Parkinson's Disease Rating Scale part III (MDS-UPDRS III), to develop and validate disease-specific machine learning comparisons using 60 template regions and tracts of interest in Montreal Neurological Institute (MNI) space between Parkinson's disease (PD) and Atypical Parkinsonism (multiple system atrophy - MSA, progressive supranuclear palsy - PSP), as well as between MSA and PSP. For each comparison, models were developed on a training/validation cohort and evaluated in a test cohort by quantifying the area under the curve (AUC) of receiving operating characteristic (ROC) curves.
Findings: In the test cohort for both disease-specific comparisons, AUCs were high in the dMRI + MDS-UPDRS (PD vs. Atypical Parkinsonism: 0·962; MSA vs. PSP: 0·897) and dMRI Only (PD vs. Atypical Parkinsonism: 0·955; MSA vs. PSP: 0·926) models, whereas the MDS-UPDRS III Only models had significantly lower AUCs (PD vs. Atypical Parkinsonism: 0·775; MSA vs. PSP: 0·582).
Interpretations: This study provides an objective, validated, and generalizable imaging approach to distinguish different forms of Parkinsonian syndromes using multi-site dMRI cohorts. The dMRI method does not involve radioactive tracers, is completely automated, and can be collected in less than 12 minutes across 3T scanners worldwide. The use of this test could thus positively impact the clinical care of patients with Parkinson's disease and Parkinsonism as well as reduce the number of misdiagnosed cases in clinical trials.

DOI: https://doi.org/10.1016/s2589-7500(19)30105-0
Radiology. 2020 Apr 07. 190283

Artificial Intelligence System Approaching Neuroradiologist-level Differential Diagnosis Accuracy at Brain MRI.

Andreas M Rauschecker, Jeffrey D Rudie, Long Xie, Jiancong Wang, Michael Tran Duong, Emmanuel J Botzolakis, Asha M Kovalovich, John Egan, Tessa C Cook, R Nick Bryan, Ilya M Nasrallah, Suyash Mohan, James C Gee.

Background Although artificial intelligence (AI) shows promise across many aspects of radiology, the use of AI to create differential diagnoses for rare and common diseases at brain MRI has not been demonstrated. Purpose To evaluate an AI system for generation of differential diagnoses at brain MRI compared with radiologists. Materials and Methods This retrospective study tested performance of an AI system for probabilistic diagnosis in patients with 19 common and rare diagnoses at brain MRI acquired between January 2008 and January 2018. The AI system combines data-driven and domain-expertise methodologies, including deep learning and Bayesian networks. First, lesions were detected by using deep learning. Then, 18 quantitative imaging features were extracted by using atlas-based coregistration and segmentation. Third, these image features were combined with five clinical features by using Bayesian inference to develop probability-ranked differential diagnoses. Quantitative feature extraction algorithms and conditional probabilities were fine-tuned on a training set of 86 patients (mean age, 49 years ± 16 [standard deviation]; 53 women). Accuracy was compared with radiology residents, general radiologists, neuroradiology fellows, and academic neuroradiologists by using accuracy of top one, top two, and top three differential diagnoses in 92 independent test set patients (mean age, 47 years ± 18; 52 women). Results For accuracy of top three differential diagnoses, the AI system (91% correct) performed similarly to academic neuroradiologists (86% correct; P = .20), and better than radiology residents (56%; P < .001), general radiologists (57%; P < .001), and neuroradiology fellows (77%; P = .003). The performance of the AI system was not affected by disease prevalence (93% accuracy for common vs 85% for rare diseases; P = .26). Radiologists were more accurate at diagnosing common versus rare diagnoses (78% vs 47% across all radiologists; P < .001). Conclusion An artificial intelligence system for brain MRI approached overall top one, top two, and top three differential diagnoses accuracy of neuroradiologists and exceeded that of less-specialized radiologists. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Zaharchuk in this issue.

DOI: https://doi.org/10.1148/radiol.2020190283
PLoS One. 2020 ;15(4): e0231192

Topics and trends in artificial intelligence assisted human brain research.

Xieling Chen, Juan Chen, Gary Cheng, Tao Gong.

Artificial intelligence (AI) assisted human brain research is a dynamic interdisciplinary field with great interest, rich literature, and huge diversity. The diversity in research topics and technologies keeps increasing along with the tremendous growth in application scope of AI-assisted human brain research. A comprehensive understanding of this field is necessary to assess research efficacy, (re)allocate research resources, and conduct collaborations. This paper combines the structural topic modeling (STM) with the bibliometric analysis to automatically identify prominent research topics from the large-scale, unstructured text of AI-assisted human brain research publications in the past decade. Analyses on topical trends, correlations, and clusters reveal distinct developmental trends of these topics, promising research orientations, and diverse topical distributions in influential countries/regions and research institutes. These findings help better understand scientific and technological AI-assisted human brain research, provide insightful guidance for resource (re)allocation, and promote effective international collaborations.

DOI: https://doi.org/10.1371/journal.pone.0231192
Am J Ophthalmol. 2020 Apr 02. pii: S0002-9394(20)30146-X. [Epub ahead of print]

Efficacy for Differentiating Nonglaucomatous versus Glaucomatous Optic Neuropathy Using Deep Learning Systems.

Hee Kyung Yang, Young Jae Kim, Jae Yun Sung, Dong Hyun Kim, Kwang Gi Kim, Jeong-Min Hwang.

PURPOSE: To assess the performance of deep learning approaches for differentiating nonglaucomatous optic neuropathy versus glaucomatous optic neuropathy (GON) on color fundus photographs by the use of image recognition.DESIGN: Development of an Artificial Intelligence Classification algorithm METHODS: Setting: Institutional.
SUBJECTS: An analysis including 3,815 fundus images from the PACS system of Seoul National University Bundang Hospital consisting of 2,883 normal optic disc images, 446 nonglaucomatous optic neuropathy with optic disc pallor (NGON) and 486 GON.
OBSERVATIONS: The presence of NGON and GON was interpreted by two expert neuro-ophthalmologists and had corroborate evidence on visual field testing and optical coherence tomography. Images were preprocessed in size and color enhancement before input. We applied the convolutional neural network (CNN) of ResNet-50 architecture. The area under the Precision-Recall curve (average precision, AP) was evaluated for the efficacy of deep learning algorithms to assess the performance of classifying nonglaucomatous optic disc pallor and GON.
RESULTS: The diagnostic accuracy of the ResNet-50 model to detect GON among NGON images showed a sensitivity of 93.4% and specificity of 81.8%. The area under the Precision-Recall curve for differentiating NGON vs GON showed an AP value of 0.874. False positive cases were found with extensive areas of peripapillary atrophy and tilted optic discs.
CONCLUSION: Artificial intelligence-based deep learning algorithms for detecting optic disc diseases showed excellent performance in differentiating nonglaucomatous and glaucomatous optic neuropathy on color fundus photographs, necessitating further research for clinical application.

DOI: https://doi.org/10.1016/j.ajo.2020.03.035
J Clin Med. 2020 Apr 03. pii: E1018. [Epub ahead of print]9(4):

Artificial Intelligence in Ophthalmology: A Meta-Analysis of Deep Learning Models for Retinal Vessels Segmentation.

Md Mohaimenul Islam, Tahmina Nasrin Poly, Bruno Andreas Walther, Hsuan Chia Yang, Yu-Chuan Jack Li.

  BACKGROUND AND OBJECTIVE: Accurate retinal vessel segmentation is often considered to be a reliable biomarker of diagnosis and screening of various diseases, including cardiovascular diseases, diabetic, and ophthalmologic diseases. Recently, deep learning (DL) algorithms have demonstrated high performance in segmenting retinal images that may enable fast and lifesaving diagnoses. To our knowledge, there is no systematic review of the current work in this research area. Therefore, we performed a systematic review with a meta-analysis of relevant studies to quantify the performance of the DL algorithms in retinal vessel segmentation.METHODS: A systematic search on EMBASE, PubMed, Google Scholar, Scopus, and Web of Science was conducted for studies that were published between 1 January 2000 and 15 January 2020. We followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) procedure. The DL-based study design was mandatory for a study's inclusion. Two authors independently screened all titles and abstracts against predefined inclusion and exclusion criteria. We used the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool for assessing the risk of bias and applicability.
RESULTS: Thirty-one studies were included in the systematic review; however, only 23 studies met the inclusion criteria for the meta-analysis. DL showed high performance for four publicly available databases, achieving an average area under the ROC of 0.96, 0.97, 0.96, and 0.94 on the DRIVE, STARE, CHASE_DB1, and HRF databases, respectively. The pooled sensitivity for the DRIVE, STARE, CHASE_DB1, and HRF databases was 0.77, 0.79, 0.78, and 0.81, respectively. Moreover, the pooled specificity of the DRIVE, STARE, CHASE_DB1, and HRF databases was 0.97, 0.97, 0.97, and 0.92, respectively.
CONCLUSION: The findings of our study showed the DL algorithms had high sensitivity and specificity for segmenting the retinal vessels from digital fundus images. The future role of DL algorithms in retinal vessel segmentation is promising, especially for those countries with limited access to healthcare. More compressive studies and global efforts are mandatory for evaluating the cost-effectiveness of DL-based tools for retinal disease screening worldwide.

Keywords:  artificial intelligence; convolutional neural network; deep learning; diabetes mellitus; retinal vessel

DOI:  https://doi.org/10.3390/jcm9041018
Eur Radiol Exp. 2020 Apr 06. 4(1): 20

Multiphase CT-based prediction of Child-Pugh classification: a machine learning approach.

Johannes Thüring, Oliver Rippel, Christoph Haarburger, Dorit Merhof, Philipp Schad, Philipp Bruners, Christiane K Kuhl, Daniel Truhn.

  BACKGROUND: To evaluate whether machine learning algorithms allow the prediction of Child-Pugh classification on clinical multiphase computed tomography (CT).METHODS: A total of 259 patients who underwent diagnostic abdominal CT (unenhanced, contrast-enhanced arterial, and venous phases) were included in this retrospective study. Child-Pugh scores were determined based on laboratory and clinical parameters. Linear regression (LR), Random Forest (RF), and convolutional neural network (CNN) algorithms were used to predict the Child-Pugh class. Their performances were compared to the prediction of experienced radiologists (ERs). Spearman correlation coefficients and accuracy were assessed for all predictive models. Additionally, a binary classification in low disease severity (Child-Pugh class A) and advanced disease severity (Child-Pugh class ≥ B) was performed.
RESULTS: Eleven imaging features exhibited a significant correlation when adjusted for multiple comparisons with Child-Pugh class. Significant correlations between predicted and measured Child-Pugh classes were observed (ρLA = 0.35, ρRF = 0.32, ρCNN = 0.51, ρERs = 0.60; p < 0.001). Significantly better accuracies for the prediction of Child-Pugh classes versus no-information rate were found for CNN and ERs (p ≤ 0.034), not for LR and RF (p ≥ 0.384). For binary severity classification, the area under the curve at receiver operating characteristic analysis was significantly lower (p ≤ 0.042) for LR (0.71) and RF (0.69) than for CNN (0.80) and ERs (0.76), without significant differences between CNN and ERs (p = 0.144).
CONCLUSIONS: The performance of a CNN in assessing Child-Pugh class based on multiphase abdominal CT images is comparable to that of ERs.

Keywords:  Artificial intelligence; Liver cirrhosis; Machine learning; Neural networks (computer); Tomography (x-ray computed)

DOI:  https://doi.org/10.1186/s41747-020-00148-3
J Thorac Imaging. 2020 Apr 07.

Machine Learning and Deep Neural Network Applications in the Thorax: Pulmonary Embolism, Chronic Thromboembolic Pulmonary Hypertension, Aorta, and Chronic Obstructive Pulmonary Disease.

Martine Remy-Jardin, Jean-Baptiste Faivre, Rainer Kaergel, Antoine Hutt, Paul Felloni, Suonita Khung, Anne-Laure Lejeune, Jessica Giordano, Jacques Remy.

The radiologic community is rapidly integrating a revolution that has not fully entered daily practice. It necessitates a close collaboration between computer scientists and radiologists to move from concepts to practical applications. This article reviews the current littérature on machine learning and deep neural network applications in the field of pulmonary embolism, chronic thromboembolic pulmonary hypertension, aorta, and chronic obstructive pulmonary disease.

DOI: https://doi.org/10.1097/RTI.0000000000000492
J Arthroplasty. 2020 Mar 18. pii: S0883-5403(20)30267-9. [Epub ahead of print]

Development of Machine Learning Algorithms to Predict Clinically Meaningful Improvement for the Patient-Reported Health State After Total Hip Arthroplasty.

Kyle N Kunze, Aditya V Karhade, Alex J Sadauskas, Joseph H Schwab, Brett R Levine.

  BACKGROUND: Failure to achieve clinically significant outcome (CSO) improvement after total hip arthroplasty (THA) imposes a potential cost-to-risk imbalance in the context of bundle payment models. Patient perception of their health state is one component of such risk. The purpose of the current study is to develop machine learning algorithms to predict CSO for the patient-reported health state (PRHS) and build a clinical decision-making tool based on risk factors.METHODS: A retrospective review of primary THA patients between 2014 and 2017 was performed. Variables considered for prediction included demographics, medical history, preoperative PRHS, and modified Harris Hip Score. The minimal clinically important difference (MCID) for the PRHS was calculated using a distribution-based method. Five supervised machine learning algorithms were developed and assessed by discrimination, calibration, Brier score, and decision curve analysis.
RESULTS: Of 616 patients, a total of 407 (69.2%) achieved the MCID for the PRHS. The random forest algorithm achieved the best performance in the independent testing set not used for algorithm development (c-statistic 0.97, calibration intercept -0.05, calibration slope 1.45, Brier score 0.054). The most important factors for achieving the MCID were preoperative PRHS, preoperative opioid use, age, and body mass index. Individual patient-level explanations were provided for the algorithm predictions and the algorithms were incorporated into an open access digital application available here: https://sorg-apps.shinyapps.io/THA_PRHS_mcid/.
CONCLUSION: The current study created a clinical decision-making tool based on partially modifiable risk factors for predicting CSO after THA. The tool demonstrates excellent discriminative capacity for identifying those at greatest risk for failing to achieve CSO in their current health state and may allow for preoperative health optimization.

Keywords:  MCID; THA; clinical outcomes; clinically significant outcome; machine learning; total hip arthroplasty

DOI:  https://doi.org/10.1016/j.arth.2020.03.019
Dermatol Ther (Heidelb). 2020 Apr 06.

Machine Learning in Dermatology: Current Applications, Opportunities, and Limitations.

Stephanie Chan, Vidhatha Reddy, Bridget Myers, Quinn Thibodeaux, Nicholas Brownstone, Wilson Liao.

  Machine learning (ML) has the potential to improve the dermatologist's practice from diagnosis to personalized treatment. Recent advancements in access to large datasets (e.g., electronic medical records, image databases, omics), faster computing, and cheaper data storage have encouraged the development of ML algorithms with human-like intelligence in dermatology. This article is an overview of the basics of ML, current applications of ML, and potential limitations and considerations for further development of ML. We have identified five current areas of applications for ML in dermatology: (1) disease classification using clinical images; (2) disease classification using dermatopathology images; (3) assessment of skin diseases using mobile applications and personal monitoring devices; (4) facilitating large-scale epidemiology research; and (5) precision medicine. The purpose of this review is to provide a guide for dermatologists to help demystify the fundamentals of ML and its wide range of applications in order to better evaluate its potential opportunities and challenges.

Keywords:  Artificial intelligence; Convolutional neural network; Deep learning; Dermatology; Image classification; Machine learning; Mobile applications; Personal monitoring devices; Precision medicine

DOI:  https://doi.org/10.1007/s13555-020-00372-0