Improving Noninvasive Classification of Molecular Subtypes of Adult Gliomas With Diffusion‐Weighted MR Imaging: An Externally Validated Machine Learning Algorithm

Abstract

BackgroundGenetic testing for molecular markers of gliomas sometimes is unavailable because of time‐consuming and expensive, even limited tumor specimens or nonsurgery cases.PurposeTo train a three‐class radiomic model classifying three molecular subtypes including isocitrate dehydrogenase (IDH) mutations and 1p/19q‐noncodeleted (IDHmut‐noncodel), IDH wild‐type (IDHwt), IDH‐mutant and 1p/19q‐codeleted (IDHmut‐codel) of adult gliomas and investigate whether radiomic features from diffusion‐weighted imaging (DWI) could bring additive value.Study TypeRetrospective.PopulationA total of 755 patients including 111 IDHmut‐noncodel, 571 IDHwt, and 73 IDHmut‐codel cases were divided into training (n = 480) and internal validation set (n = 275); 139 patients including 21 IDHmut‐noncodel, 104 IDHwt, and 14 IDHmut‐codel cases were utilized as external validation set.Field Strength/SequenceA 1.5 T or 3.0 T/multiparametric MRI, including T1‐weighted (T1), T1‐weighted gadolinium contrast‐enhanced (T1c), T2‐weighted (T2), fluid attenuated inversion recovery (FLAIR), and DWI.AssessmentThe performance of multiparametric radiomic model (random‐forest model) using 22 selected features from T1, T2, FLAIR, T1c images and apparent diffusion coefficient (ADC) maps, and conventional radiomic model using 20 selected features from T1, T2, FLAIR, and T1c images was assessed in internal and external validation sets by comparing probability values and actual incidence.Statistical TestsMann–Whitney U test, Chi‐Squared test, Wilcoxon test, receiver operating curve (ROC), and area under the curve (AUC); DeLong analysis. P < 0.05 was statistically significant.ResultsThe multiparametric radiomic model achieved AUC values for IDHmut‐noncodel, IDHwt, and IDHmut‐codel of 0.8181, 0.8524, and 0.8502 in internal validation set and 0.7571, 0.7779, and 0.7491 in external validation set, respectively. Multiparametric radiomic model showed significantly better diagnostic performance after DeLong analysis, especially in classifying IDHwt and IDHmut‐noncodel subtypes.Data ConclusionRadiomic features from DWI could bring additive value and improve the performance of conventional MRI‐based radiomic model for classifying the molecular subtypes especially IDHmut‐noncodel and IDHwt of adult gliomas.Level of Evidence3.Technical EfficacyStage 2.

Advancing noninvasive glioma classification with diffusion radiomics: Exploring the impact of signal intensity normalization

“…However, when examining the external UCSF dataset, the addition of N4/z-score normalized radiomics for ADC outperformed the other approaches for both the IDH-wt and IDH-mut non-codel groups. These results are in part consistent with the recently published work of Guo et al, 11 which showed a significant performance increase for prediction of IDH-wt and IDH-mut non-codel group in both the internal and external dataset by adding normalization-naïve ADC radiomics using a random forest multiclass model. The fact that in the external validation, in contrast to Guo et al, we only saw a performance increase with normalized, but not naïve ADC radiomics, highlights the importance of intensity normalization of ADC for improving the generalizability of radiomic-based prediction models.…”

Amide proton transfer weighted and diffusion weighted imaging based radiomics classification algorithm for predicting 1p/19q co-deletion status in low grade gliomas

“…There are growing number of studies using machine learning algorithms to predict molecular subtypes such as 1p/19q co-deletion and IDH mutations [ 7 , 21 , 30 – 33 ]. However, most of these studies only used conventional anatomical MR sequences because of the widespread usage.…”

Biologically interpretable multi-task deep learning pipeline predicts molecular alterations, grade, and prognosis in glioma patients

“…Another limitation is the exclusion of functional MRI techniques such as diffusion-and perfusion-weighted imaging from our analysis. These techniques can provide valuable insights into tumor cellularity and vascularity, complementing conventional imaging modalities [38,46,47]. However, their application is not widespread in clinical practice, being more commonly employed for research purposes.…”