Why imaging data alone is not enough: AI-based integration of imaging, omics, and clinical data

Background: The German research networks DescribePSP and ProPSP prospectively collect comprehensive clinical data, imaging data and biomaterials of patients with a clinical diagnosis of progressive supranuclear palsy. Progressive supranuclear palsy is a rare, adult-onset, neurodegenerative disease with striking clinical heterogeneity. Since now, prospective natural history data are largely lacking. Clinical research into treatment strategies has been limited due to delay in clinical diagnosis and lack of natural history data on distinct clinical phenotypes.Methods: The DescribePSP network is organized by the German Center for Neurodegenerative Diseases. DescribePSP is embedded in a larger network with parallel cohorts of other neurodegenerative diseases and healthy controls. The DescribePSP network is directly linked to other Describe cohorts with other primary diagnoses of the neurodegenerative and vascular disease spectrums and also to an autopsy program for clinico-pathological correlation. The ProPSP network is organized by the German Parkinson and Movement Disorders Society. Both networks follow the same core protocol for patient recruitment and collection of data, imaging and biomaterials. Both networks host a web-based data registry and a central biorepository. Inclusion/exclusion criteria follow the 2017 Movement Disorder Society criteria for the clinical diagnosis of progressive supranuclear palsy.Results: Both networks started recruitment of patients by the end of 2015. As of November 2020, N = 354 and 269 patients were recruited into the DescribePSP and the ProPSP studies, respectively, and N = 131 and 87 patients received at least one follow-up visit.Conclusions: The DescribePSP and ProPSP networks are ideal resources for comprehensive natural history data of PSP, including imaging data and biological samples. In contrast to previous natural history studies, DescribePSP and ProPSP include not only patients with Richardson's syndrome, but also variant PSP phenotypes as well as patients at very early disease stages, before a diagnosis of possible or probable PSP can be made. This will allow for identification and evaluation of early biomarkers for diagnosis, prognosis, and progression.

Download Full-text

Challenges in Building of Deep Learning Models for Glioblastoma Segmentation: Evidence from Clinical Data

Studies in Health Technology and Informatics - Public Health and Informatics ◽

10.3233/shti210168 ◽

2021 ◽

Author(s):

Anvar Kurmukov ◽

Aleksandra Dalechina ◽

Talgat Saparov ◽

Mikhail Belyaev ◽

Svetlana Zolotova ◽

...

Keyword(s):

Deep Learning ◽

Clinical Data ◽

High Performance ◽

Heterogeneous Data ◽

High Variability ◽

Clinical Settings ◽

Training Procedure ◽

Imaging Data ◽

Magnetic Resonance Imaging Mri ◽

Segmentation Models

In this article, we compare the performance of a state-of-the-art segmentation network (UNet) on two different glioblastoma (GB) segmentation datasets. Our experiments show that the same training procedure yields almost twice as bad results on the retrospective clinical data compared to the BraTS challenge data (in terms of Dice score). We discuss possible reasons for such an outcome, including inter-rater variability and high variability in magnetic resonance imaging (MRI) scanners and scanner settings. The high performance of segmentation models, demonstrated on preselected imaging data, does not bring the community closer to using these algorithms in clinical settings. We believe that a clinically applicable deep learning architecture requires a shift from unified datasets to heterogeneous data.

Download Full-text

The OncoLifeS data-biobank for oncology: a comprehensive repository of clinical data, biological samples, and the patient’s perspective

Journal of Translational Medicine ◽

10.1186/s12967-019-2122-x ◽

2019 ◽

Vol 17 (1) ◽

Cited By ~ 2

Author(s):

Grigory Sidorenkov ◽

Janny Nagel ◽

Coby Meijer ◽

Jacko J. Duker ◽

Harry J. M. Groen ◽

...

Keyword(s):

Quality Of Life ◽

Clinical Data ◽

Clinical Care ◽

Participation Rate ◽

Quality Data ◽

Imaging Data ◽

Life Study ◽

Cancer Heterogeneity ◽

Living Well

Abstract Background Understanding cancer heterogeneity, its temporal evolution over time, and the outcomes of guided treatment depend on accurate data collection in a context of routine clinical care. We have developed a hospital-based data-biobank for oncology, entitled OncoLifeS (Oncological Life Study: Living well as a cancer survivor), that links routine clinical data with preserved biological specimens and quality of life assessments. The aim of this study is to describe the organization and development of a data-biobank for cancer research. Results We have enrolled 3704 patients aged ≥ 18 years diagnosed with cancer, of which 45 with hereditary breast-ovarian cancer (70% participation rate) as of October 24th, 2019. The average age is 63.6 ± 14.2 years and 1892 (51.1%) are female. The following data are collected: clinical and treatment details, comorbidities, lifestyle, radiological and pathological findings, and long-term outcomes. We also collect and store various biomaterials of patients as well as information from quality of life assessments. Conclusion Embedding a data-biobank in clinical care can ensure the collection of high-quality data. Moreover, the inclusion of longitudinal quality of life data allows us to incorporate patients’ perspectives and inclusion of imaging data provides an opportunity for analyzing raw imaging data using artificial intelligence (AI) methods, thus adding new dimensions to the collected data.

Download Full-text

Multidisciplinary Evaluation in Patients with Lung Disease Associated with Connective Tissue Disease

Seminars in Respiratory and Critical Care Medicine ◽

10.1055/s-0039-1684020 ◽

2019 ◽

Vol 40 (02) ◽

pp. 184-193 ◽

Cited By ~ 5

Author(s):

Athol Wells ◽

Anand Devaraj ◽

Elizabetta A. Renzoni ◽

Christopher P. Denton

Keyword(s):

Connective Tissue ◽

Lung Disease ◽

Connective Tissue Disease ◽

Clinical Data ◽

Pulmonary Function Tests ◽

Systemic Disease ◽

Evidence Base ◽

Pulmonary Vascular Disease ◽

Imaging Data ◽

Treatment Needs

AbstractMultidisciplinary diagnosis is now viewed as the diagnostic reference standard in interstitial lung disease (ILD). This process consists of the integration of the evidence base with clinical reasoning in the formulation of a diagnosis and requires input from clinicians, radiologists, and, in selected cases, histopathologists. In ILD associated with connective tissue disease (CTD-ILD), multidisciplinary evaluation is especially helpful when CTD is suspected but cannot be diagnosed using strict criteria. In this context, the integration of systemic clinical data, serologic information, and computed tomography and biopsy findings may allow CTD-ILD to be diagnosed. However, the value of multidisciplinary evaluation in CTD-ILD is not confined to diagnosis. The frequent coexistence of pulmonary processes other than ILD, including pulmonary vascular disease, extrapulmonic restriction, and airways disease, often has a major impact on symptoms and pulmonary function tests (PFTs). In this review, we highlight the value of multidisciplinary discussion (MDD) in reconciling clinical data, PFT, and imaging data in the accurate staging of disease severity, baseline prognostic evaluation, and the identification of progression of ILD. MDD also provides a means to combine the views of respiratory physicians and rheumatologists in formulating a treatment strategy. It is often possible to reach a robust view as to whether management should be driven by systemic disease, pulmonary disease, or both. When treatment needs to be introduced or modified for both systemic and pulmonary reasons, face-to-face discussion facilities the selection of therapeutic agents that are likely to be efficacious for both systemic and pulmonary diseases.

Download Full-text

Deep Learning Models for Predicting Severe Progression in COVID-19-Infected Patients: Retrospective Study

JMIR Medical Informatics ◽

10.2196/24973 ◽

2021 ◽

Vol 9 (1) ◽

pp. e24973

Author(s):

Thao Thi Ho ◽

Jongmin Park ◽

Taewoo Kim ◽

Byunggeon Park ◽

Jaehee Lee ◽

...

Keyword(s):

Neural Network ◽

Deep Learning ◽

High Risk ◽

Convolutional Neural Network ◽

Clinical Data ◽

Area Under The Curve ◽

Classification Performance ◽

Imaging Data ◽

Learning Models ◽

Predictive Tool

Background Many COVID-19 patients rapidly progress to respiratory failure with a broad range of severities. Identification of high-risk cases is critical for early intervention. Objective The aim of this study is to develop deep learning models that can rapidly identify high-risk COVID-19 patients based on computed tomography (CT) images and clinical data. Methods We analyzed 297 COVID-19 patients from five hospitals in Daegu, South Korea. A mixed artificial convolutional neural network (ACNN) model, combining an artificial neural network for clinical data and a convolutional neural network for 3D CT imaging data, was developed to classify these cases as either high risk of severe progression (ie, event) or low risk (ie, event-free). Results Using the mixed ACNN model, we were able to obtain high classification performance using novel coronavirus pneumonia lesion images (ie, 93.9% accuracy, 80.8% sensitivity, 96.9% specificity, and 0.916 area under the curve [AUC] score) and lung segmentation images (ie, 94.3% accuracy, 74.7% sensitivity, 95.9% specificity, and 0.928 AUC score) for event versus event-free groups. Conclusions Our study successfully differentiated high-risk cases among COVID-19 patients using imaging and clinical features. The developed model can be used as a predictive tool for interventions in aggressive therapies.

Download Full-text

Imaging-AMARETTO: An Imaging Genomics Software Tool to Interrogate Multiomics Networks for Relevance to Radiography and Histopathology Imaging Biomarkers of Clinical Outcomes

JCO Clinical Cancer Informatics ◽

10.1200/cci.19.00125 ◽

2020 ◽

pp. 421-435 ◽

Cited By ~ 1

Author(s):

Olivier Gevaert ◽

Mohsen Nabian ◽

Shaimaa Bakr ◽

Celine Everaert ◽

Jayendra Shinde ◽

...

Keyword(s):

Clinical Outcomes ◽

Clinical Data ◽

Complex Diseases ◽

Imaging Biomarkers ◽

Low Grade ◽

Imaging Features ◽

Imaging Data ◽

Imaging Genomics ◽

Genomics Tools ◽

Patient Studies

PURPOSE The availability of increasing volumes of multiomics, imaging, and clinical data in complex diseases such as cancer opens opportunities for the formulation and development of computational imaging genomics methods that can link multiomics, imaging, and clinical data. METHODS Here, we present the Imaging-AMARETTO algorithms and software tools to systematically interrogate regulatory networks derived from multiomics data within and across related patient studies for their relevance to radiography and histopathology imaging features predicting clinical outcomes. RESULTS To demonstrate its utility, we applied Imaging-AMARETTO to integrate three patient studies of brain tumors, specifically, multiomics with radiography imaging data from The Cancer Genome Atlas (TCGA) glioblastoma multiforme (GBM) and low-grade glioma (LGG) cohorts and transcriptomics with histopathology imaging data from the Ivy Glioblastoma Atlas Project (IvyGAP) GBM cohort. Our results show that Imaging-AMARETTO recapitulates known key drivers of tumor-associated microglia and macrophage mechanisms, mediated by STAT3, AHR, and CCR2, and neurodevelopmental and stemness mechanisms, mediated by OLIG2. Imaging-AMARETTO provides interpretation of their underlying molecular mechanisms in light of imaging biomarkers of clinical outcomes and uncovers novel master drivers, THBS1 and MAP2, that establish relationships across these distinct mechanisms. CONCLUSION Our network-based imaging genomics tools serve as hypothesis generators that facilitate the interrogation of known and uncovering of novel hypotheses for follow-up with experimental validation studies. We anticipate that our Imaging-AMARETTO imaging genomics tools will be useful to the community of biomedical researchers for applications to similar studies of cancer and other complex diseases with available multiomics, imaging, and clinical data.

Download Full-text

Folded Free Vascularized Fibular Bone Graft for Segmental Femoral Bone Defect: a Retrospective Report

10.21203/rs.3.rs-606412/v1 ◽

2021 ◽

Author(s):

Yifan Wu ◽

Chao Jian ◽

Baiwen Qi ◽

Zonghuan Li ◽

Aixi Yu

Keyword(s):

Bone Graft ◽

Bone Defect ◽

Clinical Data ◽

External Fixators ◽

Imaging Data ◽

Femoral Bone ◽

Fixation Failure ◽

Time Operation ◽

Segmental Bone

Abstract ObjectiveVascularized fibular bone graft is an efficient method for various segmental bone defects. The objective of this report is to introduce our experience of folded free vascularized fibular bone graft for segmental femoral bone defect.Patients and methodsClinical data collected by surgeons and Hospital Information System (HIS) system were screened respectively. Cases with segmental femoral bone defect repaired by folded free vascularized fibular bone graft were collected. Clinical data including demographic characteristics, defect size, coinfection, perioperative treatment and imaging data during follow up were all collected for analysis.ResultsTwelve patients (10 males and 2 females), aged from 6 to 58, were included in this report. The defect range was 3 to 10 cm, with an average of 6.2 cm. Three cases were complicated with infection, the others were not. Folded free vascularized fibular bone graft were harvested for the reconstruction of segmental femoral bone defect. The grafts were fixed with plates in 9 cases and external fixators in 3 cases. All grafts healed uneventfully with an average healing time of 5.2 months (range 4~8 months). Internal fixation failure occurred in one case. The follow up time ranged from 15 to 130 months (average 58.3 months).ConclusionFolded free fibula graft is one of the optional methods for segmental bone defect of femur. Through this method, patients can achieve one-time operation to reconstruct the bone defect of the affected limb.

Download Full-text

Folded Free Vascularized Fibular Bone Graft for Segmental Femoral Bone Defect: a Retrospective Report

10.21203/rs.3.rs-606412/v2 ◽

2021 ◽

Author(s):

Yifan Wu ◽

Chao Jian ◽

Baiwen Qi ◽

Zonghuan Li ◽

Aixi Yu

Keyword(s):

Bone Graft ◽

Bone Defect ◽

Clinical Data ◽

External Fixators ◽

Imaging Data ◽

Femoral Bone ◽

Fixation Failure ◽

Time Operation ◽

Segmental Bone

Abstract ObjectiveVascularized fibular bone graft is an efficient method for various segmental bone defects. The objective of this report is to introduce our experience of folded free vascularized fibular bone graft for segmental femoral bone defect.Patients and methodsClinical data collected by surgeons and Hospital Information System (HIS) system were screened respectively. Cases with segmental femoral bone defect repaired by folded free vascularized fibular bone graft were collected. Clinical data including demographic characteristics, defect size, coinfection, perioperative treatment and imaging data during follow up were all collected for analysis.ResultsTwelve patients (10 males and 2 females), aged from 6 to 58, were included in this report. The defect range was 3 to 10 cm, with an average of 6.2 cm. Three cases were complicated with infection, the others were not. Folded free vascularized fibular bone graft were harvested for the reconstruction of segmental femoral bone defect. The grafts were fixed with plates in 9 cases and external fixators in 3 cases. All grafts healed uneventfully with an average healing time of 5.2 months (range 4~8 months). Internal fixation failure occurred in one case. The follow up time ranged from 15 to 130 months (average 58.3 months).ConclusionFolded free fibula graft is one of the optional methods for segmental bone defect of femur. Through this method, patients can achieve one-time operation to reconstruct the bone defect of the affected limb.

Download Full-text

Deep learning models for predicting severe progression in COVID-19-infected patients (Preprint)

10.2196/preprints.24973 ◽

2020 ◽

Author(s):

Sanghun Choi ◽

Jae-Kwang Lim ◽

Thao Thi Ho ◽

Jongmin Park ◽

Taewoo Kim ◽

...

Keyword(s):

Neural Network ◽

Deep Learning ◽

High Risk ◽

Clinical Data ◽

Mixed Model ◽

Classification Performance ◽

Imaging Data ◽

Learning Models ◽

Active Therapy ◽

Novel Coronavirus

BACKGROUND Many COVID-19 patients rapidly progress into respiratory failure with a broad range of severity. Identification of the high-risk cases is critical for early intervention. OBJECTIVE The aim of this study is to develop deep learning models that can rapidly diagnose high-risk COVID-19 patients based on computed tomography (CT) images and clinical data. METHODS We analyzed 297 COVID-19 patients from five hospitals in Daegu, South Korea. A mixed model (ACNN) including an artificial neural network for clinical data and a convolution-neural network for 3D CT imaging data is developed to classify high-risk cases with a severe progression (event) from low-risk COVID-19 patients (event-free). RESULTS By using the mixed ACNN model, we could obtain high classification performance using novel coronavirus pneumonia (NCP) lesion images (93.9% accuracy, 80.8% sensitivity, 96.9% specificity, and 0.916 AUC) and using lung segmentation images (94.3% accuracy, 74.7% sensitivity, 95.9% specificity, and 0.928 AUC) for event vs. event-free groups. CONCLUSIONS Our study has successfully differentiated high-risk cases among COVID-19 patients using the imaging and clinical features of COVID-19 patients. The developed model is potentially utilized as a prediction tool for intervening active therapy.

Download Full-text