Texture Analysis of High b‐Value Diffusion‐Weighted Imaging for Evaluating Consistency of Pituitary Macroadenomas

Abstract

Background: Preoperative evaluation of the consistency of pituitary macroadenomas is important for neurosurgeons to prepare the surgical plan. Purpose: To evaluate the diagnostic performance of texture analysis (TA) of diffusion-weighted imaging (DWI) at a standard b-value (b = 1000 s/mm 2 ) and a high b-value (b = 2000 s/mm 2 ) for their ability to assess the tumor consistency of pituitary macroadenomas. Study Type: Retrospective. Population/Subjects: Fifty patients with histologically confirmed pituitary macroadenomas were classified as soft (n = 37) or hard (n = 13) types. Field Strength/Sequence: Coronal T 2 -weighted imaging (T 2 WI), Readout Segmentation of Long Variable Echo-trains (RESOLVE) DWI at b = 1000 s/mm 2 and b = 2000 s/mm 2 were acquired with 3.0T MRI. Assessment: The corresponding apparent diffusion coefficient (ADC) maps (ADC 1000 and ADC 2000 ) were registered to T 2 WI. Regions of interest (ROIs) were manually drawn along the solid part of the tumor from the coregistered T 2 WI-ADC images. The texture parameters from T 2 WI, ADC 1000 , and ADC 2000 were acquired. Statistical Tests: The texture parameters were compared between the two types by using unpaired Student's t-test. Receiver operating characteristic (ROC) curves and logistic regression analyses were used to assess their diagnostic performance. Results: Significant differences in TA parameters of ADC 1000 and ADC 2000 were observed between soft and hard types (P < 0.05 for all), whereas the TA of T 2 WI resulted in no significant difference (P > 0.05 for all). TA of ADC 2000 provided a superior diagnostic performance compared with that of ADC 1000 (P = 0.038). A combination of mean value and entropy of ADC 2000 yielded an AUC, a sensitivity, and a specificity of 0.911, 78.4% and 92.3%, respectively. Data Conclusion: TA of ADC values were useful for assessing the tumor consistency of pituitary macroadenomas. ADC 2000 may facilitate better type discrimination. Level of Evidence: 3 Technical Efficacy Stage: 2 J. MAGN. RESON. IMAGING 2020;51:1507-1513.View this article online at wileyonlinelibrary.com.

Editorial for “Texture Analysis of High b‐value Diffusion‐Weighted Imaging for Evaluating Consistency of Pituitary Macroadenomas”

“…These histologic features result in greater cellularity for hard adenomas, which helps explain why they can be differentiated from soft adenomas with DWI. The increased cellularity of hard adenomas results in lower apparent diffusion coefficient (ADC) values, which is more pronounced at higher b‐values, as shown by Su et al The lower ADC values seen with high b‐value DWI are believed to result in greater differences between pixel signal intensity values within the regions of interest, thus resulting in more pronounced differences with MRTA. It may also be that hard adenomas are more heterogeneous than soft adenomas, but there is a need to validate with radiologic‐pathologic correlation for pituitary macroadenoma architecture in future studies.…”

“…What drives this process is the belief that medical images contain information about underlying pathophysiology, and that these data can be extracted and used through quantitative image analysis . The article by Su et al is a novel application of magnetic resonance texture analysis (MRTA) as it shows how pituitary macroadenoma consistency can be predicted preoperatively with high b‐value diffusion‐weighted imaging (DWI).…”

“…As emphasized by Su et al, there is considerable importance for accurate preoperative identification of pituitary macroadenoma consistency, since the surgical approach differs for hard and soft adenomas. This importance is further emphasized by the volume of work dedicated to exploring this topic.…”

Radiomic Features on Multiparametric MRI for Preoperative Evaluation of Pituitary Macroadenomas Consistency: Preliminary Findings

“…However, approximately 10% of pituitary macroadenomas (PMA) are fibrous and exhibit hard consistency, and are poorly amenable to suction and curettage 3 . Therefore, preoperative evaluation of the consistency of PMA is important for surgical planning and for the avoidance of multistage surgical procedures 4,5 …”

“…Therefore, the goal of this study was to investigate if the creation of new composite image features based on conventional multiparametric MRI (mpMRI) could effectively assess tumor consistency of PMA. Our approach was different from the previous work, in which image features were extracted from two‐dimensional (2D) tumor region of single‐sequence MRIs (eg, T1WI, 11 T1CE, 7,11 T2WI, 3,7,9 or diffusion‐weighted imaging (DWI) 4,5,12 ). In this study, we focused on 3D radiomic features (shape, histogram, texture, and wavelet) from PMA volume of interest (VOI) on the individual T1WI/T1CE/T2WI and their combinations.…”