Skip to main content

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

Editor’s Choice

This 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, medulloblastoma, pilocytic astrocytoma, and ependymoma. 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 MRI as input to detect the presence of tumor and predict tumor class. Model tumor detection accuracy exceeded an AUC 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.

Abstract

BACKGROUND AND PURPOSE

Figure 2 from Quon et al
CAMs depicting the areas of the input slice that the model preferentially emphasizes when predicting tumor subtype on individual scan slices. The upper row of each subpanel shows the T2 slice with tumor areas manually denoted (upper left) and CAM overlay of the most confident prediction of the model (upper right). The lower row of each panel shows less confident predictions. Examples of correct predictions of PA (A) and MB (B) and incorrect predictions of PA (C) and MB (D) are shown.

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.

Read this article: https://bit.ly/2G0eDCx

The post Deep Learning for Pediatric Posterior Fossa Tumor Detection and Classification: A Multi-Institutional Study appeared first on AJNR Blog.



from AJNR Blog https://ift.tt/2RYoXgI

Labels

Labels:

Comments

Post a Comment

Popular posts from this blog

Menopause Symptoms Reduced by Cold Water Swimming

Cold water swimming significantly eases menopausal symptoms. Surveying 1114 women, with 785 experiencing menopause, researchers found improvements in anxiety, mood swings, low mood, and hot flushes among participants. from Neuroscience News https://ift.tt/9AqHsEa
Post a Comment
Read more

UPI: Kids with psych disorders most likely to take dangerous viral challenges

The “choking game” — and other clearly ill-advised and dangerous internet challenges — leave many parents wondering what drives teens to take the bait and participate. Now, a new study suggests that an underlying psychological disorder may be one reason why some kids jump at online dares such as the “Bird Box” challenge, where people walk around blindfolded, and the Tide Pod challenge, daring people to eat laundry detergent. (January 28, 2019) Read the full article here from Brain Health Daily http://bit.ly/2DIWHbD
Post a Comment
Read more

The emerging influential role of microglia in neurology

In her most catchily titled book, The Angel and the Assassin , Donna Jackson Nakazawa highlighed nerve cells which have hitherto been very little acknowledged – microglia . Long ignored as bit players in the big league of the nervous system, Nakazawa colourfully illustrated what many neuroscientists are beginning to realise: the small size of microglia belies their huge influence ; m icroglia are, after all, the defence force of the nervous system, protecting the brain from microbial invaders . In keeping with their small size, their role is to surreptitiously  present the antigens of invading bugs to T cells , the toffs who actually carry out the final hatchet job . It is therefore not surprising that any dysfunction of microglia will come with significant clinical consequences .  By GerryShaw – Own work , CC BY-SA 3.0 , Link The most important clinical fallout of dysfunctional microglia appears to be the emergence of dementia. It is indeed spec...
Post a Comment
Read more