Elsevier

Clinical Imaging

Volume 40, Issue 3, May–June 2016, Pages 386-391
Clinical Imaging

Original Article
Volumetric segmentation of ADC maps and utility of standard deviation as measure of tumor heterogeneity in soft tissue tumors

https://doi.org/10.1016/j.clinimag.2015.11.017Get rights and content

Abstract

Purpose

Determine interobserver concordance of semiautomated three-dimensional volumetric and two-dimensional manual measurements of apparent diffusion coefficient (ADC) values in soft tissue masses (STMs) and explore standard deviation (SD) as a measure of tumor ADC heterogeneity.

Results

Concordance correlation coefficients for mean ADC increased with more extensive sampling. Agreement on the SD of tumor ADC values was better for large regions of interest and multislice methods. Correlation between mean and SD ADC was low, suggesting that these parameters are relatively independent.

Conclusion

Mean ADC of STMs can be determined by volumetric quantification with high interobserver agreement. STM heterogeneity merits further investigation as a potential imaging biomarker that complements other functional magnetic resonance imaging parameters.

Introduction

Although conventional magnetic resonance imaging (MRI) is the primary method of soft tissue tumor imaging, imaging techniques such as diffusion-weighted imaging (DWI) have recently been described and are increasingly being utilized for tumor characterization and posttreatment surveillance [1], [2]. In highly cellular regions within a soft tissue tumor, water movement is restricted, which results in decreased apparent diffusion coefficient (ADC) values [3]. Del Grande and colleagues demonstrated that the addition of functional imaging (including DWI) to conventional contrast-enhanced imaging increased specificity during the evaluation of local recurrence of soft tissue sarcomas from 52% to 97% [2]. In addition, following the treatment of sarcomas, several investigators have shown that ADC values increase when tumor necrosis takes place [4], [5], [6], [7], [8], [9].

While many investigators have used DWI to evaluate tumors and their response to treatment, the methods for quantification and interobserver variability vary widely [2], [4], [10], [11]. Mean ADC values may be derived from subjective small regions of interest (ROI) constructed around only the lowest values on the ADC map, or the ROI may encompass the entire tumor area on a single slice. Focus on mean ADC, to the exclusion of standard deviation (SD), may neglect important histologic features of a soft tissue mass. For example, the high intrinsic T2 signal of low-grade myxomatous tumors elevates mean ADC values above that observed in most cellular sarcomas [12], but intratumoral cystic necrosis can similarly elevate the mean ADC value [13].

Thus, while mean ADC values may help differentiate benign and malignant tumors across a wide range of organ systems and tumor types [14], [15], [16], [17], [18], [19], [20], changes in ADC tumor heterogeneity assessed longitudinally during treatment may provide another method to characterize the viability of tumor and may aid in detection of tumor recurrence even in the absence of intravenous contrast. There are emerging data that indicate utility in quantifying the heterogeneity of tumors with DWI [21], [22], although this has not been fully explored in the evaluation of soft tissue sarcomas. A prerequisite for such an assessment is a uniform repeatable method to analyze sufficient tumor volume so that all parts of the tumor, both necrotic and solid, are encompassed. This study aimed to show the feasibility of volumetric ADC quantification in soft tissue masses and to determine the interobserver agreement of semiautomated three-dimensional (3D) volumetric segmentation compared with two-dimensional (2D) manual quantification methods. Our hypothesis was that 2D multislice and 3D volumetric tumor segmentation on ADC maps would result in good interobserver agreement in the quantification of the ADC SD.

Section snippets

Case selection

This study was approved by the institutional review board with waiver of informed consent. A retrospective database search of extremity MRIs acquired between 2013 and 2014 that included either the terms “msk tumor protocol,” “diffusion-weighted imaging,” or “ADC” in the report. Additionally, a retrospective review of the institutional sarcoma database was performed including cases from 2013 to 2014. Cases were included if MRI was performed with DWI and ADC maps. Additional inclusion criteria

Soft tissue masses

After application of the inclusion and exclusion criteria, 22 soft tissue lesions were included in the study. A total of 22 masses from 22 patients (16 male, 6 female, mean age 50 years, range 18–81) were included in this study. Seventeen out of 18 malignant masses were biopsy-proven malignancies (17 sarcomas, 1 case of biopsy-proven extramedullary myeloma was included), and 5/18 malignant masses were previously untreated at the time of DWI acquisition (Fig. 4). No longitudinal measurements were

Discussion

DWI is a functional technique that has shown promise for several applications in soft tissue tumor imaging, including the characterization of soft tissue masses, assessment of treatment response, and surveillance for potential tumor recurrence. In prior studies, measurements have generally been made by nonstandardized manual methods [13], [24]. In this study, we assessed the interobserver concordance of measuring soft tissue mass mean ADC value and SD about the mean ADC by manual methods as

Conclusions

ADC map quantification of soft tissue tumors using mean and SD parameters is robust, with high interobserver concordance when methods of measurement include large areas of tumor sampling. In quantifying soft tissue tumor ADC heterogeneity, the SD of tumor ADC values may provide a unique imaging biomarker complementary to other functional MRI parameters.

References (30)

  • BH Byun et al.

    Combination of 18F-FDG PET/CT and diffusion-weighted MR imaging as a predictor of histologic response to neoadjuvant chemotherapy: preliminary results in osteosarcoma

    J Nucl Med

    (2013)
  • AA Abdel Razek et al.

    Role of diffusion-weighted MRI in differentiation of masticator space malignancy from infection

    Dentomaxillofac Radiol

    (2013)
  • V Vandecaveye et al.

    Diffusion-weighted magnetic resonance imaging in neck lymph adenopathy

    Cancer Imaging

    (2008)
  • M Maeda et al.

    Soft-tissue tumors evaluated by line-scan diffusion-weighted imaging: influence of myxoid matrix on the apparent diffusion coefficient

    J Magn Reson Imaging

    (2007)
  • TK Subhawong et al.

    Characterization of soft tissue masses: can quantitative diffusion weighted imaging reliably distinguish cysts from solid masses?

    Skeletal Radiol

    (2013)
  • Cited by (15)

    View all citing articles on Scopus
    View full text