Search

Radiogenomics algorithm

Purpose:Patients with 1p/19q codeleted low-grade glioma (LGG) have longer overall survival and better treatment response than patients with 1p/19q intact tumors. Therefore, it is relevant to know the 1p/19q status. To investigate whether the 1p/19q status can be assessed prior to tumor resection, we developed a machine learning algorithm to predict the 1p/19q status of presumed LGG based on preoperative MRI.Experimental Design:Preoperative brain MR images from 284 patients who had undergone biopsy or resection of presumed LGG were used to train a support vector machine algorithm. The algorithm was trained on the basis of features extracted from post-contrast T1-weighted and T2-weighted MR images and on patients' age and sex. The performance of the algorithm compared with tissue diagnosis was assessed on an external validation dataset of MR images from 129 patients with LGG from The Cancer Imaging Archive (TCIA). Four clinical experts also predicted the 1p/19q status of the TCIA MR images.Results:The algorithm achieved an AUC of 0.72 in the external validation dataset. The algorithm had a higher predictive performance than the average of the neurosurgeons (AUC 0.52) but lower than that of the neuroradiologists (AUC of 0.81). There was a wide variability between clinical experts (AUC 0.45–0.83).Conclusions:Our results suggest that our algorithm can noninvasively predict the 1p/19q status of presumed LGG with a performance that on average outperformed the oncological neurosurgeons. Evaluation on an independent dataset indicates that our algorithm is robust and generalizable.

Overview of importance of imaging and demographic features

Overview of MR settings

Results of mixing EMC/HMC and TCIA datasets

Overview of the radiogenomics pipeline

Excluded patients from TCIA-1p19qDeletion dataset

Purpose: Patients with 1p/19q codeleted low-grade glioma (LGG) have longer overall survival and better treatment response than patients with 1p/19q intact tumors. Therefore, it is relevant to know the 1p/19q status. To investigate whether the 1p/19q status can be assessed prior to tumor resection, we developed a machine learning algorithm to predict the 1p/19q status of presumed LGG based on preoperative MRI. Experimental Design: Preoperative brain MR images from 284 patients who had undergone biopsy or resection of presumed LGG were used to train a support vector machine algorithm. The algorithm was trained on the basis of features extracted from post-contrast T1-weighted and T2-weighted MR images and on patients' age and sex. The performance of the algorithm compared with tissue diagnosis was assessed on an external validation dataset of MR images from 129 patients with LGG from The Cancer Imaging Archive (TCIA). Four clinical experts also predicted the 1p/19q status of the TCIA MR images. Results: The algorithm achieved an AUC of 0.72 in the external validation dataset. The algorithm had a higher predictive performance than the average of the neurosurgeons (AUC 0.52) but lower than that of the neuroradiologists (AUC of 0.81). There was a wide variability between clinical experts (AUC 0.45–0.83). Conclusions: Our results suggest that our algorithm can noninvasively predict the 1p/19q status of presumed LGG with a performance that on average outperformed the oncological neurosurgeons. Evaluation on an independent dataset indicates that our algorithm is robust and generalizable.

To evaluate the use of MRI and CT in the diagnostic work-up of dementia in Dutch memory clinics, and to analyse the rationale for choosing each modality.A digital survey was sent by e-mail to all medical specialists (n=235) working at a memory clinic in the Netherlands.The survey was completed by 64% (151). 85% of the respondents were geriatricians, 13% neurologists and 2% other, working at a total of 69 clinics. 40% variably orders CT or MRI, 37% orders MRI, 19% CT, and 4% CT plus MRI. Primary factors influencing this choice are: MRI contraindications, physical limitations, age, vascular or oncological medical history, and waiting time. With CT, 87% indicates information is lacking: vascular disease/white matter lesions, (hippocampal) atrophy, and specific pathologies (metastases, amyloid angiopathy). Furthermore, respondents prefer MRI because they can assess the images more easily themselves. Only 50% of respondents indicate that CT protocol dictates coronal reconstructions. Additionally, these reconstructions are not provided consistently. Rating-scales are used to describe images in 5%. In 75% assessment is not uniform.MRI is preferred over CT in diagnostic imaging of dementia, in accordance with existing guidelines. However, these guidelines are mostly out-dated and modern multislice CT potential is relatively unknown among geriatricians. In memory clinics, multislice CT could offer a well suitable imaging alternative, but only if multiplanar reconstructions are performed consistently. Furthermore, radiology reports need to be improved by using more standardized assessment.