Radiomics Can Distinguish Pediatric Supratentorial Embryonal Tumors, High-Grade Gliomas, and Ependymomas

BACKGROUND AND PURPOSE

Pediatric supratentorial tumors such as embryonal tumors, high-grade gliomas, and ependymomas are difficult to distinguish by histopathology and imaging because of overlapping features. We applied machine learning to uncover MR imaging-based radiomics phenotypes that can differentiate these tumor types.

MATERIALS AND METHODS

Our retrospective cohort of 231 patients from 7 participating institutions had 50 embryonal tumors, 127 high-grade gliomas, and 54 ependymomas. For each tumor volume, we extracted 900 Image Biomarker Standardization Initiative-based PyRadiomics features from T2-weighted and gadolinium-enhanced T1-weighted images. A reduced feature set was obtained by sparse regression analysis and was used as input for 6 candidate classifier models. Training and test sets were randomly allocated from the total cohort in a 75:25 ratio.

RESULTS

The final classifier model for embryonal tumor-versus-high-grade gliomas identified 23 features with an area under the curve of 0.98; the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 0.85, 0.91, 0.79, 0.94, and 0.89, respectively. The classifier for embryonal tumor-versus-ependymomas identified 4 features with an area under the curve of 0.82; the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 0.93, 0.69, 0.76, 0.90, and 0.81, respectively. The classifier for high-grade gliomas-versus-ependymomas identified 35 features with an area under the curve of 0.96; the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 0.82, 0.94, 0.82, 0.94, and 0.91, respectively.

CONCLUSIONS

In this multi-institutional study, we identified distinct radiomic phenotypes that distinguish pediatric supratentorial tumors, high-grade gliomas, and ependymomas with high accuracy. Incorporation of this technique in diagnostic algorithms can improve diagnosis, risk stratification, and treatment planning.

Radiomic Phenotypes Distinguish Atypical Teratoid/Rhabdoid Tumors from Medulloblastoma

BACKGROUND AND PURPOSE

Atypical teratoid/rhabdoid tumors and medulloblastomas have similar imaging and histologic features but distinctly different outcomes. We hypothesized that they could be distinguished by MR imaging-based radiomic phenotypes.

MATERIALS AND METHODS

We retrospectively assembled T2-weighted and gadolinium-enhanced T1-weighted images of 48 posterior fossa atypical teratoid/rhabdoid tumors and 96 match-paired medulloblastomas from 7 institutions. Using a holdout test set, we measured the performance of 6 candidate classifier models using 6 imaging features derived by sparse regression of 900 T2WI and 900 T1WI Imaging Biomarker Standardization Initiative-based radiomics features.

RESULTS

From the originally extracted 1800 total Imaging Biomarker Standardization Initiative-based features, sparse regression consistently reduced the feature set to 1 from T1WI and 5 from T2WI. Among classifier models, logistic regression performed with the highest AUC of 0.86, with sensitivity, specificity, accuracy, and F1 scores of 0.80, 0.82, 0.81, and 0.85, respectively. The top 3 important Imaging Biomarker Standardization Initiative features, by decreasing order of relative contribution, included voxel intensity at the 90th percentile, inverse difference moment normalized, and kurtosis-all from T2WI.

CONCLUSIONS

Six quantitative signatures of image intensity, texture, and morphology distinguish atypical teratoid/rhabdoid tumors from medulloblastomas with high prediction performance across different machine learning strategies. Use of this technique for preoperative diagnosis of atypical teratoid/rhabdoid tumors could significantly inform therapeutic strategies and patient care discussions.

Deep Learning for Pediatric Posterior Fossa Tumor Detection and Classification: A Multi-Institutional Study

BACKGROUND AND PURPOSE

Posterior fossa tumors are the most common pediatric brain tumors. MR imaging is key to tumor detection, diagnosis, and therapy guidance. We sought to develop an MR imaging-based deep learning model for posterior fossa tumor detection and tumor pathology classification.

MATERIALS AND METHODS

The study cohort comprised 617 children (median age, 92 months; 56% males) from 5 pediatric institutions with posterior fossa tumors: diffuse midline glioma of the pons (n = 122), medulloblastoma (n = 272), pilocytic astrocytoma (n = 135), and ependymoma (n = 88). There were 199 controls. Tumor histology served as ground truth except for diffuse midline glioma of the pons, which was primarily diagnosed by MR imaging. A modified ResNeXt-50-32x4d architecture served as the backbone for a multitask classifier model, using T2-weighted MRIs as input to detect the presence of tumor and predict tumor class. Deep learning model performance was compared against that of 4 radiologists.

RESULTS

Model tumor detection accuracy exceeded an AUROC of 0.99 and was similar to that of 4 radiologists. Model tumor classification accuracy was 92% with an F₁ score of 0.80. The model was most accurate at predicting diffuse midline glioma of the pons, followed by pilocytic astrocytoma and medulloblastoma. Ependymoma prediction was the least accurate. Tumor type classification accuracy and F₁ score were higher than those of 2 of the 4 radiologists.

CONCLUSIONS

We present a multi-institutional deep learning model for pediatric posterior fossa tumor detection and classification with the potential to augment and improve the accuracy of radiologic diagnosis.

Gray Matter Growth Is Accompanied by Increasing Blood Flow and Decreasing Apparent Diffusion Coefficient during Childhood

BACKGROUND AND PURPOSE

Normal values of gray matter volume, cerebral blood flow, and water diffusion have not been established for healthy children. We sought to determine reference values for age-dependent changes of these parameters in healthy children.

MATERIALS AND METHODS

We retrospectively reviewed MR imaging data from 100 healthy children. Using an atlas-based approach, age-related normal values for regional CBF, apparent diffusion coefficient, and volume were determined for the cerebral cortex, hippocampus, thalamus, caudate, putamen, globus pallidus, amygdala, and nucleus accumbens.

RESULTS

All gray matter structures grew rapidly before the age of 10 years and then plateaued or slightly declined thereafter. The ADC of all structures decreased with age, with the most rapid changes occurring prior to the age of 5 years. With the exception of the globus pallidus, CBF increased rather linearly with age.

CONCLUSIONS

Normal brain gray matter is characterized by rapid early volume growth and increasing CBF with concomitantly decreasing ADC. The extracted reference data that combine CBF and ADC parameters during brain growth may provide a useful resource when assessing pathologic changes in children.

Hydrocephalus decreases arterial spin-labeled cerebral perfusion

BACKGROUND AND PURPOSE

Reduced cerebral perfusion has been observed with elevated intracranial pressure. We hypothesized that arterial spin-labeled CBF can be used as a marker for symptomatic hydrocephalus.

MATERIALS AND METHODS

We compared baseline arterial spin-labeled CBF in 19 children (median age, 6.5 years; range, 1-17 years) with new posterior fossa brain tumors and clinical signs of intracranial hypertension with arterial spin-labeled CBF in 16 age-matched controls and 4 patients with posterior fossa tumors without ventriculomegaly or signs of intracranial hypertension. Measurements were recorded in the cerebrum at the vertex, deep gray nuclei, and periventricular white matter and were assessed for a relationship to ventricular size. In 16 symptomatic patients, we compared cerebral perfusion before and after alleviation of hydrocephalus.

RESULTS

Patients with uncompensated hydrocephalus had lower arterial spin-labeled CBF than healthy controls for all brain regions interrogated (P < .001). No perfusion difference was seen between asymptomatic patients with posterior fossa tumors and healthy controls (P = 1.000). The median arterial spin-labeled CBF increased after alleviation of obstructive hydrocephalus (P < .002). The distance between the frontal horns inversely correlated with arterial spin-labeled CBF of the cerebrum (P = .036) but not the putamen (P = .156), thalamus (P = .111), or periventricular white matter (P = .121).

CONCLUSIONS

Arterial spin-labeled-CBF was reduced in children with uncompensated hydrocephalus and restored after its alleviation. Arterial spin-labeled-CBF perfusion MR imaging may serve a future role in the neurosurgical evaluation of hydrocephalus, as a potential noninvasive method to follow changes of intracranial pressure with time.

Time-dependent structural changes of the dentatothalamic pathway in children treated for posterior fossa tumor

BACKGROUND AND PURPOSE

Injury to the dentatothalamic pathway that originates in the cerebellum has been suggested as a mechanism for neurologic complications in children treated for posterior fossa tumors. We hypothesized that time-dependent changes occur in the dentatothalamic pathway.

MATERIALS AND METHODS

Diffusion tensor evaluation was performed in 14 children (median age, 4.1 years; age range, 1-20 years) who underwent serial MR imaging at 3T as part of routine follow-up after posterior fossa tumor resection with or without adjuvant therapy. Tensor metrics were obtained in the acute (≤1 week), subacute (1 to <6 months), and chronic (≥6 months) periods after surgery. We evaluated the following dentatothalamic constituents: bilateral dentate nuclei, cerebellar white matter, and superior cerebellar peduncles. Serial dentate nuclei volumes were also obtained and compared with the patient's baseline.

RESULTS

The most significant tensor changes to the superior cerebellar peduncles and cerebellar white matter occurred in the subacute period, regardless of the tumor pathology or therapy regimen, with signs of recovery in the chronic period. However, chronic volume loss and reduced mean diffusivity were observed in the dentate nuclei and did not reverse. This atrophy was associated with radiation therapy and symptoms of ataxia.

CONCLUSIONS

Longitudinal diffusion MR imaging in children treated for posterior fossa tumors showed time-dependent tensor changes in components of the dentatothalamic pathway that suggest evolution of structural damage with inflammation and recovery of tissue directionality. However, the dentate nuclei did not show tensor or volumetric recovery, suggesting that the injury may be chronic.

Reduced cerebral arterial spin-labeled perfusion in children with neurofibromatosis type 1

BACKGROUND AND PURPOSE

Neurofibromatosis type 1 is associated with increased risk for stroke, cerebral vasculopathy, and neurocognitive deficits, but underlying hemodynamic changes in asymptomatic children remain poorly understood. We hypothesized that children with neurofibromatosis type 1 have decreased cerebral blood flow.

MATERIALS AND METHODS

Arterial spin-labeled CBF was measured in 14 children with neurofibromatosis type 1 (median age, 9.7 years; mean, 10.2 years; range, 22 months to 18 years) and compared with age-matched control subjects on 3T MR imaging. Three-dimensional pseudocontinuous spin-echo arterial spin-labeled technique was used. Measurements were obtained at cortical gray matter of bilateral cerebral hemispheres and centrum semiovale by use of the ROI method. Comparison by Mann-Whitney test was used, with Bonferroni-adjusted P values ≤.004 judged as significant.

RESULTS

We identified 7 of 12 areas with significantly diminished arterial spin-labeled CBF in patients with neurofibromatosis type 1 compared with control subjects. These areas included the anterior cingulate gyrus (P = .001), medial frontal cortex (P = .004), centrum semiovale (P = .004), temporo-occipital cortex (P = .002), thalamus (P = .001), posterior cingulate gyrus (P = .002), and occipital cortex (P = .001). Among patients with neurofibromatosis type 1, there were no significant differences in these regions on the basis of the presence of neurofibromatosis type 1 spots or neurocognitive deficits.

CONCLUSIONS

Reduced cerebral perfusion was seen in children with neurofibromatosis type 1, particularly in the posterior circulation and the vascular borderzones of the middle and posterior cerebral arteries.

Arterial spin-labeled perfusion of pediatric brain tumors

BACKGROUND AND PURPOSE

Pediatric brain tumors have diverse pathologic features, which poses diagnostic challenges. Although perfusion evaluation of adult tumors is well established, hemodynamic properties are not well characterized in children. Our goal was to apply arterial spin-labeling perfusion for various pathologic types of pediatric brain tumors and evaluate the role of arterial spin-labeling in the prediction of tumor grade.

MATERIALS AND METHODS

Arterial spin-labeling perfusion of 54 children (mean age, 7.5 years; 33 boys and 21 girls) with treatment-naive brain tumors was retrospectively evaluated. The 3D pseudocontinuous spin-echo arterial spin-labeling technique was acquired at 3T MR imaging. Maximal relative tumor blood flow was obtained by use of the ROI method and was compared with tumor histologic features and grade.

RESULTS

Tumors consisted of astrocytic (20), embryonal (11), ependymal (3), mixed neuronal-glial (8), choroid plexus (5), craniopharyngioma (4), and other pathologic types (3). The maximal relative tumor blood flow of high-grade tumors (grades III and IV) was significantly higher than that of low-grade tumors (grades I and II) (P < .001). There was a wider relative tumor blood flow range among high-grade tumors (2.14 ± 1.78) compared with low-grade tumors (0.60 ± 0.29) (P < .001). Across the cohort, relative tumor blood flow did not distinguish individual histology; however, among posterior fossa tumors, relative tumor blood flow was significantly higher for medulloblastoma compared with pilocytic astrocytoma (P = .014).

CONCLUSIONS

Characteristic arterial spin-labeling perfusion patterns were seen among diverse pathologic types of brain tumors in children. Arterial spin-labeling perfusion can be used to distinguish high-grade and low-grade tumors.

Low grade gliomas of childhood: the actual management

Low grade gliomas (LGG) are the most common brain tumors of childhood and adolescence, consisting of a mixed group of grade I and grade II neoplasms. In general, their location and resectability are the most important predictors of outcome. Surgery is curative, usually for superficial tumors of the cerebral or cerebellar hemispheres, but has a more controversial role for deep-seated midline tumors. Where gross total resection is not feasible, LGG becomes a chronic disease of childhood, and adjuvant therapies must be tailored for each individual based on age and tumor location. Radiotherapy (RT) is effective but associated with neurological, cognitive, and endocrinological morbidity, prompting the use of chemotherapy regimens aimed at delaying RT, especially in younger children. Long-term surveillance imaging for up to five years after treatment is warranted even after gross total resection, and lifelong surveillance is warranted after RT because of iatrogenic effects. Despite a favorable prognosis for survival, LGG are associated with disability, decreased quality of life, and late effects of treatment, all requiring long-term specialty care through a multidisciplinary approach.

Diffusion-weighted MRI: distinction of skull base chordoma from chondrosarcoma

BACKGROUND AND PURPOSE

Chordoma and chondrosarcoma of the skull base are rare tumors with overlapping presentations and anatomic imaging features but different prognoses. We hypothesized that these tumors might be distinguished by using diffusion-weighted MR imaging.

MATERIALS AND METHODS

We retrospectively reviewed 19 patients with pathologically confirmed chordoma or chondrosarcoma who underwent both conventional and diffusion-weighted MR imaging. Differences in distributions of ADC were assessed by the Kruskal-Wallis test. Associations between histopathologic diagnosis and conventional MR imaging features (T2 signal intensity, contrast enhancement, and tumor location) were assessed with the Fisher exact test.

RESULTS

Chondrosarcoma was associated with the highest mean ADC value (2051 ± 261 × 10(-6) mm(2)/s) and was significantly different from classic chordoma (1474 ± 117 × 10(-6) mm(2)/s) and poorly differentiated chordoma (875 ± 100 × 10(-6) mm(2)/s) (P < .001). Poorly differentiated chordoma was characterized by low T2 signal intensity (P = .001), but other conventional MR imaging features of enhancement and/or lesion location did not reliably distinguish these tumor types.

CONCLUSIONS

Diffusion-weighted MR imaging may be useful in assessing clival tumors, particularly in differentiating chordoma from chondrosarcoma. A prospective study of a larger cohort will be required to determine the value of ADC in predicting histopathologic diagnosis.