Gliomas belong to a group of central nervous system tumors, and consist of various sub-regions. Gold standard labeling of these sub-regions in radiographic imaging is essential for both clinical and ...computational studies, including radiomic and radiogenomic analyses. Towards this end, we release segmentation labels and radiomic features for all pre-operative multimodal magnetic resonance imaging (MRI) (n=243) of the multi-institutional glioma collections of The Cancer Genome Atlas (TCGA), publicly available in The Cancer Imaging Archive (TCIA). Pre-operative scans were identified in both glioblastoma (TCGA-GBM, n=135) and low-grade-glioma (TCGA-LGG, n=108) collections via radiological assessment. The glioma sub-region labels were produced by an automated state-of-the-art method and manually revised by an expert board-certified neuroradiologist. An extensive panel of radiomic features was extracted based on the manually-revised labels. This set of labels and features should enable i) direct utilization of the TCGA/TCIA glioma collections towards repeatable, reproducible and comparative quantitative studies leading to new predictive, prognostic, and diagnostic assessments, as well as ii) performance evaluation of computer-aided segmentation methods, and comparison to our state-of-the-art method.
Purpose
This report presents the methods and results of the Thoracic Auto‐Segmentation Challenge organized at the 2017 Annual Meeting of American Association of Physicists in Medicine. The purpose of ...the challenge was to provide a benchmark dataset and platform for evaluating performance of autosegmentation methods of organs at risk (OARs) in thoracic CT images.
Methods
Sixty thoracic CT scans provided by three different institutions were separated into 36 training, 12 offline testing, and 12 online testing scans. Eleven participants completed the offline challenge, and seven completed the online challenge. The OARs were left and right lungs, heart, esophagus, and spinal cord. Clinical contours used for treatment planning were quality checked and edited to adhere to the RTOG 1106 contouring guidelines. Algorithms were evaluated using the Dice coefficient, Hausdorff distance, and mean surface distance. A consolidated score was computed by normalizing the metrics against interrater variability and averaging over all patients and structures.
Results
The interrater study revealed highest variability in Dice for the esophagus and spinal cord, and in surface distances for lungs and heart. Five out of seven algorithms that participated in the online challenge employed deep‐learning methods. Although the top three participants using deep learning produced the best segmentation for all structures, there was no significant difference in the performance among them. The fourth place participant used a multi‐atlas‐based approach. The highest Dice scores were produced for lungs, with averages ranging from 0.95 to 0.98, while the lowest Dice scores were produced for esophagus, with a range of 0.55–0.72.
Conclusion
The results of the challenge showed that the lungs and heart can be segmented fairly accurately by various algorithms, while deep‐learning methods performed better on the esophagus. Our dataset together with the manual contours for all training cases continues to be available publicly as an ongoing benchmarking resource.
In this paper we report the set-up and results of the Multimodal Brain Tumor Image Segmentation Benchmark (BRATS) organized in conjunction with the MICCAI 2012 and 2013 conferences. Twenty ...state-of-the-art tumor segmentation algorithms were applied to a set of 65 multi-contrast MR scans of low- and high-grade glioma patients - manually annotated by up to four raters - and to 65 comparable scans generated using tumor image simulation software. Quantitative evaluations revealed considerable disagreement between the human raters in segmenting various tumor sub-regions (Dice scores in the range 74%-85%), illustrating the difficulty of this task. We found that different algorithms worked best for different sub-regions (reaching performance comparable to human inter-rater variability), but that no single algorithm ranked in the top for all sub-regions simultaneously. Fusing several good algorithms using a hierarchical majority vote yielded segmentations that consistently ranked above all individual algorithms, indicating remaining opportunities for further methodological improvements. The BRATS image data and manual annotations continue to be publicly available through an online evaluation system as an ongoing benchmarking resource.
The National Institutes of Health have placed significant emphasis on sharing of research data to support secondary research. Investigators have been encouraged to publish their clinical and imaging ...data as part of fulfilling their grant obligations. Realizing it was not sufficient to merely ask investigators to publish their collection of imaging and clinical data, the National Cancer Institute (NCI) created the open source National Biomedical Image Archive software package as a mechanism for centralized hosting of cancer related imaging. NCI has contracted with Washington University in Saint Louis to create The Cancer Imaging Archive (TCIA)—an open-source, open-access information resource to support research, development, and educational initiatives utilizing advanced medical imaging of cancer. In its first year of operation, TCIA accumulated 23 collections (3.3 million images). Operating and maintaining a high-availability image archive is a complex challenge involving varied archive-specific resources and driven by the needs of both image submitters and image consumers. Quality archives of any type (traditional library, PubMed, refereed journals) require management and customer service. This paper describes the management tasks and user support model for TCIA.
Grand challenges stimulate advances within the medical imaging research community; within a competitive yet friendly environment, they allow for a direct comparison of algorithms through a ...well-defined, centralized infrastructure. The tasks of the two-part PROSTATEx Challenges (the PROSTATEx Challenge and the PROSTATEx-2 Challenge) are (1) the computerized classification of clinically significant prostate lesions and (2) the computerized determination of Gleason Grade Group in prostate cancer, both based on multiparametric magnetic resonance images. The challenges incorporate well-vetted cases for training and testing, a centralized performance assessment process to evaluate results, and an established infrastructure for case dissemination, communication, and result submission. In the PROSTATEx Challenge, 32 groups apply their computerized methods (71 methods total) to 208 prostate lesions in the test set. The area under the receiver operating characteristic curve for these methods in the task of differentiating between lesions that are and are not clinically significant ranged from 0.45 to 0.87; statistically significant differences in performance among the top-performing methods, however, are not observed. In the PROSTATEx-2 Challenge, 21 groups apply their computerized methods (43 methods total) to 70 prostate lesions in the test set. When compared with the reference standard, the quadratic-weighted kappa values for these methods in the task of assigning a five-point Gleason Grade Group to each lesion range from
to 0.27; superiority to random guessing can be established for only two methods. When approached with a sense of commitment and scientific rigor, challenges foster interest in the designated task and encourage innovation in the field.
Purpose:
The development of computer-aided diagnostic (CAD) methods for lung nodule detection, classification, and quantitative assessment can be facilitated through a well-characterized repository ...of computed tomography (CT) scans. The Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI) completed such a database, establishing a publicly available reference for the medical imaging research community. Initiated by the National Cancer Institute (NCI), further advanced by the Foundation for the National Institutes of Health (FNIH), and accompanied by the Food and Drug Administration (FDA) through active participation, this public-private partnership demonstrates the success of a consortium founded on a consensus-based process.
Methods:
Seven academic centers and eight medical imaging companies collaborated to identify, address, and resolve challenging organizational, technical, and clinical issues to provide a solid foundation for a robust database. The LIDC/IDRI Database contains 1018 cases, each of which includes images from a clinical thoracic CT scan and an associated XML file that records the results of a two-phase image annotation process performed by four experienced thoracic radiologists. In the initial blinded-read phase, each radiologist independently reviewed each CT scan and marked lesions belonging to one of three categories (“
nodule
≥
3
mm
,” “
nodule
<
3
mm
,” and “non-
nodule
≥
3
mm
”). In the subsequent unblinded-read phase, each radiologist independently reviewed their own marks along with the anonymized marks of the three other radiologists to render a final opinion. The goal of this process was to identify as completely as possible all lung nodules in each CT scan without requiring forced consensus.
Results:
The Database contains 7371 lesions marked “nodule” by at least one radiologist. 2669 of these lesions were marked “
nodule
≥
3
mm
” by at least one radiologist, of which 928 (34.7%) received such marks from all four radiologists. These 2669 lesions include nodule outlines and subjective nodule characteristic ratings.
Conclusions:
The LIDC/IDRI Database is expected to provide an essential medical imaging research resource to spur CAD development, validation, and dissemination in clinical practice.
To conduct a comprehensive analysis of radiologist-made assessments of glioblastoma (GBM) tumor size and composition by using a community-developed controlled terminology of magnetic resonance (MR) ...imaging visual features as they relate to genetic alterations, gene expression class, and patient survival.
Because all study patients had been previously deidentified by the Cancer Genome Atlas (TCGA), a publicly available data set that contains no linkage to patient identifiers and that is HIPAA compliant, no institutional review board approval was required. Presurgical MR images of 75 patients with GBM with genetic data in the TCGA portal were rated by three neuroradiologists for size, location, and tumor morphology by using a standardized feature set. Interrater agreements were analyzed by using the Krippendorff α statistic and intraclass correlation coefficient. Associations between survival, tumor size, and morphology were determined by using multivariate Cox regression models; associations between imaging features and genomics were studied by using the Fisher exact test.
Interrater analysis showed significant agreement in terms of contrast material enhancement, nonenhancement, necrosis, edema, and size variables. Contrast-enhanced tumor volume and longest axis length of tumor were strongly associated with poor survival (respectively, hazard ratio: 8.84, P = .0253, and hazard ratio: 1.02, P = .00973), even after adjusting for Karnofsky performance score (P = .0208). Proneural class GBM had significantly lower levels of contrast enhancement (P = .02) than other subtypes, while mesenchymal GBM showed lower levels of nonenhanced tumor (P < .01).
This analysis demonstrates a method for consistent image feature annotation capable of reproducibly characterizing brain tumors; this study shows that radiologists' estimations of macroscopic imaging features can be combined with genetic alterations and gene expression subtypes to provide deeper insight to the underlying biologic properties of GBM subsets.
Individual MRI characteristics (e.g., volume) are routinely used to identify survival-associated phenotypes for glioblastoma (GBM). This study investigated whether combinations of MRI features can ...also stratify survival. Furthermore, the molecular differences between phenotype-induced groups were investigated.
Ninety-two patients with imaging, molecular, and survival data from the TCGA (The Cancer Genome Atlas)-GBM collection were included in this study. For combinatorial phenotype analysis, hierarchical clustering was used. Groups were defined based on a cutpoint obtained via tree-based partitioning. Furthermore, differential expression analysis of microRNA (miRNA) and mRNA expression data was performed using GenePattern Suite. Functional analysis of the resulting genes and miRNAs was performed using Ingenuity Pathway Analysis. Pathway analysis was performed using Gene Set Enrichment Analysis.
Clustering analysis reveals that image-based grouping of the patients is driven by 3 features: volume-class, hemorrhage, and T1/FLAIR-envelope ratio. A combination of these features stratifies survival in a statistically significant manner. A cutpoint analysis yields a significant survival difference in the training set (median survival difference: 12 months, p = 0.004) as well as a validation set (p = 0.0001). Specifically, a low value for any of these 3 features indicates favorable survival characteristics. Differential expression analysis between cutpoint-induced groups suggests that several immune-associated (natural killer cell activity, T-cell lymphocyte differentiation) and metabolism-associated (mitochondrial activity, oxidative phosphorylation) pathways underlie the transition of this phenotype. Integrating data for mRNA and miRNA suggests the roles of several genes regulating proliferation and invasion.
A 3-way combination of MRI phenotypes may be capable of stratifying survival in GBM. Examination of molecular processes associated with groups created by this combinatorial phenotype suggests the role of biological processes associated with growth and invasion characteristics.
Online public repositories for sharing research data allow investigators to validate existing research or perform secondary research without the expense of collecting new data. Patient data made ...publicly available through such repositories may constitute a breach of personally identifiable information if not properly de-identified. Imaging data are especially at risk because some intricacies of the Digital Imaging and Communications in Medicine (DICOM) format are not widely understood by researchers. If imaging data still containing protected health information (PHI) were released through a public repository, a number of different parties could be held liable, including the original researcher who collected and submitted the data, the original researcher's institution, and the organization managing the repository. To minimize these risks through proper de-identification of image data, one must understand what PHI exists and where that PHI resides, and one must have the tools to remove PHI without compromising the scientific integrity of the data. DICOM public elements are defined by the DICOM Standard. Modality vendors use private elements to encode acquisition parameters that are not yet defined by the DICOM Standard, or the vendor may not have updated an existing software product after DICOM defined new public elements. Because private elements are not standardized, a common de-identification practice is to delete all private elements, removing scientifically useful data as well as PHI. Researchers and publishers of imaging data can use the tools and process described in this article to de-identify DICOM images according to current best practices.