Dataset Information

Machine Learning Decision Tree Models for Differentiation of Posterior Fossa Tumors Using Diffusion Histogram Analysis and Structural MRI Findings.

ABSTRACT: We applied machine learning algorithms for differentiation of posterior fossa tumors using apparent diffusion coefficient (ADC) histogram analysis and structural MRI findings. A total of 256 patients with intra-axial posterior fossa tumors were identified, of whom 248 were included in machine learning analysis, with at least 6 representative subjects per each tumor pathology. The ADC histograms of solid components of tumors, structural MRI findings, and patients' age were applied to construct decision models using Classification and Regression Tree analysis. We also compared different machine learning classification algorithms (i.e., naïve Bayes, random forest, neural networks, support vector machine with linear and polynomial kernel) for dichotomized differentiation of the 5 most common tumors in our cohort: metastasis (n = 65), hemangioblastoma (n = 44), pilocytic astrocytoma (n = 43), ependymoma (n = 27), and medulloblastoma (n = 26). The decision tree model could differentiate seven tumor histopathologies with terminal nodes yielding up to 90% accurate classification rates. In receiver operating characteristics (ROC) analysis, the decision tree model achieved greater area under the curve (AUC) for differentiation of pilocytic astrocytoma (p = 0.020); and atypical teratoid/rhabdoid tumor ATRT (p = 0.001) from other types of neoplasms compared to the official clinical report. However, neuroradiologists' interpretations had greater accuracy in differentiating metastases (p = 0.001). Among different machine learning algorithms, random forest models yielded the highest accuracy in dichotomized classification of the 5 most common tumor types; and in multiclass differentiation of all tumor types random forest yielded an averaged AUC of 0.961 in training datasets, and 0.873 in validation samples. Our study demonstrates the potential application of machine learning algorithms and decision trees for accurate differentiation of brain tumors based on pretreatment MRI. Using easy to apply and understandable imaging metrics, the proposed decision tree model can help radiologists with differentiation of posterior fossa tumors, especially in tumors with similar qualitative imaging characteristics. In particular, our decision tree model provided more accurate differentiation of pilocytic astrocytomas from ATRT than by neuroradiologists in clinical reads.

SUBMITTER: Payabvash S

PROVIDER: S-EPMC7018938 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine Learning Decision Tree Models for Differentiation of Posterior Fossa Tumors Using Diffusion Histogram Analysis and Structural MRI Findings.

Payabvash Seyedmehdi S Aboian Mariam M Tihan Tarik T Cha Soonmee S

Frontiers in oncology 20200207

We applied machine learning algorithms for differentiation of posterior fossa tumors using apparent diffusion coefficient (ADC) histogram analysis and structural MRI findings. A total of 256 patients with intra-axial posterior fossa tumors were identified, of whom 248 were included in machine learning analysis, with at least 6 representative subjects per each tumor pathology. The ADC histograms of solid components of tumors, structural MRI findings, and patients' age were applied to construct de ...[more]

PMID: 32117728

Similar Datasets

Project description:PurposeProspective and longitudinal neuroimaging studies of posterior fossa tumors are scarce. Here we evaluate the early changes in white matter and intellectual outcome up to 3 years after diagnosis.Patients and methodsTwenty-two children with posterior fossa tumors and 24 similarly-aged healthy children participated. Patients included: (a) 12 individuals who received surgery, cranial-spinal radiation (CSR), and focal radiation to the tumor bed (CSR group) and (b) 10 individuals who received local therapy, either surgery only or surgery and focal radiation to the tumor bed (Local group). Diffusion tensor imaging (DTI) and intelligence measures were obtained an average of 3 months after diagnosis and then at 12, 24, and 36 months later. DTI tractography and voxel-wise approaches were employed. The Neurological Predictor Scale was used to summarize the type and amount of treatment for PF tumor patients. Linear mixed modelling was used to evaluate group differences at baseline and changes over time in DTI metrics for both the specific white matter tracts and voxel-wise, as well as for intelligence measures.ResultsBased on tractography, patients treated with CSR had significantly higher Axial and Mean diffusivity in the cortical-spinal tracts (CST) 3 month after diagnosis - particularly on the right side, p < .003, compared to healthy children. Mean diffusivity in right CST decreased over time in this group of patients, p = .001. No differences compared to controls were evident in specific tracts for the Local group, p > .10. Voxel-wise analyses revealed multiple areas of white matter compromise in both patients groups. Notably, both patient groups had lower scores on intelligence measures compared to the Control group: The CSR group displayed lower performance 3 months following diagnosis, ps < 0.001, and their performance remained stable over time ps > 0.10, whereas the Local group displayed no differences at 3 months, ps> 0.10, but their performance declined over time, ps < 0.01. At baseline, higher MD in right CST predicted lower Perceptual Reasoning scores across all participants, p = .001. Furthermore, lower FA in left IFOF at baseline predicted decline in Processing Speed over time, p = .001. In patients, more aggressive treatment protocols and presence of mutism were related to lower performance on intelligence measures at baseline, ps < 0.04.ConclusionsChildren treated with CSR displayed diffuse white matter compromise and poor intellectual outcome shortly after radiation treatment. There was evidence of subsequent growth of white matter structure, but stable intellectual insult. Conversely, in children treated with either surgery only or surgery and focal radiation to the tumor bed we observed less compromise of white matter early following treatment and no intellectual insult compared to healthy children. However, declines in intellectual function were evident for these children, though their performance remained within the average normative range. Overall, results suggest that early intervention is necessary to circumvent these deficits.

Dataset Information

Machine Learning Decision Tree Models for Differentiation of Posterior Fossa Tumors Using Diffusion Histogram Analysis and Structural MRI Findings.

Publications

Machine Learning Decision Tree Models for Differentiation of Posterior Fossa Tumors Using Diffusion Histogram Analysis and Structural MRI Findings.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets