Diffusion-weighted imaging of prostate cancer: effect of b-value distribution on repeatability and cancer characterization

Merisaari, Harri; Toivonen, Jussi; Pesola, M.; Taimen, Pekka; Boström, Peter J.; Pahikkala, Tapio; Aronen, Hannu J.; Jambor, Ivan

doi:10.1016/j.mri.2015.07.004

Cited by 25 publications

(29 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Patients’ characteristics of the 112 included patients are summarized in Supporting Information Table . Fifty‐one (45%) of the 112 included patients were part of previous studies optimizing b‐value distributions for prostate DWI and evaluating different mathematical models for PCa DWI …”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Repeatability of radiomics and machine learning for DWI: Short‐term repeatability study of 112 patients with prostate cancer

Merisaari

Taimen

Shiradkar

et al. 2019

Magnetic Resonance in Med

Self Cite

View full text Add to dashboard Cite

Purpose To evaluate repeatability of prostate DWI‐derived radiomics and machine learning methods for prostate cancer (PCa) characterization. Methods A total of 112 patients with diagnosed PCa underwent 2 prostate MRI examinations (Scan1 and Scan2) performed on the same day. DWI was performed using 12 b‐values (0–2000 s/mm2), post‐processed using kurtosis function, and PCa areas were annotated using whole mount prostatectomy sections. A total of 1694 radiomic features including Sobel, Kirch, Gradient, Zernike Moments, Gabor, Haralick, CoLIAGe, Haar wavelet coefficients, 3D analogue to Laws features, 2D contours, and corner detectors were calculated. Radiomics and 4 feature pruning methods (area under the receiver operator characteristic curve, maximum relevance minimum redundancy, Spearman’s ρ, Wilcoxon rank‐sum) were evaluated in terms of Scan1‐Scan2 repeatability using intraclass correlation coefficient (ICC)(3,1). Classification performance for clinically significant and insignificant PCa with Gleason grade groups 1 versus >1 was evaluated by area under the receiver operator characteristic curve in unseen random 30% data split. Results The ICC(3,1) values for conventional radiomics and feature pruning methods were in the range of 0.28–0.90. The machine learning classifications varied between Scan1 and Scan2 with % of same class labels between Scan1 and Scan2 in the range of 61–81%. Surface‐to‐volume ratio and corner detector‐based features were among the most represented features with high repeatability, ICC(3,1) >0.75, consistently high ranking using all 4 feature pruning methods, and classification performance with area under the receiver operator characteristic curve >0.70. Conclusion Surface‐to‐volume ratio and corner detectors for prostate DWI led to good classification of unseen data and performed similarly in Scan1 and Scan2 in contrast to multiple conventional radiomic features.

show abstract

Section: Methodsmentioning

confidence: 99%

“…Fifty-one (45%) of the 112 included patients were part of previous studies optimizing b-value distributions for prostate DWI 29 and evaluating different mathematical models for PCa DWI. 30,31,43…”

Section: Methodsmentioning

confidence: 99%

Repeatability of radiomics and machine learning for DWI: Short‐term repeatability study of 112 patients with prostate cancer

Merisaari

Taimen

Shiradkar

et al. 2019

Magnetic Resonance in Med

Self Cite

View full text Add to dashboard Cite

show abstract

“…We recognize that this study was limited by the patient number and heterogeneity; however, it appears adequate for the present exploratory study. Moreover, although the use of more than three b‐values allowed us to improve the parameter repeatability, K/K av and D/D av estimated values could depend on the number and range of used b‐values. Lastly, given that no consensus exists regarding the most appropriate approach for tumor delineation and analyses, our results could have been affected by the signal measurement modality or by the algorithm used to extract the DKI‐derived indices.…”

Section: Discussionmentioning

confidence: 99%

Diffusional kurtosis imaging in head and neck cancer: On the use of trace‐weighted images to estimate indices of non‐Gaussian water diffusion

et al. 2018

View full text Add to dashboard Cite

Purpose While previous studies have demonstrated the feasibility and potential usefulness of quantitative non‐Gaussian diffusional kurtosis imaging (DKI) of the brain, more recent research has focused on oncological application of DKI in various body regions such as prostate, breast, and head and neck (HN). Given the need to minimize scan time during most routine magnetic resonance imaging (MRI) acquisitions of body regions, diffusion‐weighted imaging (DWI) with only three orthogonal diffusion weighting directions (x, y, z) is usually performed. Moreover, as water diffusion within malignant tumors is generically thought to be almost isotropic, DWI with only three diffusion weighting directions is considered sufficient for oncological application and it represents the de facto standard in body DKI. In this context, since the kurtosis tensor and diffusion tensor cannot be obtained, the averages of the three directional (Kx, Ky, Kz) and (Dx, Dy, Dz) — namely K and D, respectively — represent the best‐possible surrogates of directionless DKI‐derived indices of kurtosis and diffusivity, respectively. This would require fitting the DKI model to the diffusion‐weighted images acquired along each direction (x, y, z) prior to averaging. However, there is a growing tendency to perform only a single fit of the DKI model to the geometric means of the images acquired with diffusion‐sensitizing gradient along (x, y, z), referred to as trace‐weighted (TW) images. To the best of our knowledge, no in vivo studies have evaluated how TW images affect estimates of DKI‐derived indices of K and D. Thus, the aim of this study was to assess the potential bias and error introduced in estimated K and D by fitting the DKI model to the TW images in HN cancer patients. Methods Eighteen patients with histologically proven malignant tumors of the HN were enrolled in the study. They underwent pretreatment 3 T MRI, including DWI (b‐values: 0, 500, 1000, 1500, 2000 s/mm2). Some patients had multiple lesions, and thus a total of 34 lesions were analyzed. DKI‐derived indices were estimated, voxel‐by‐voxel, using single diffusion‐weighted images along (x, y, z) as well as TW images. A comparison between the two estimation methods was performed by calculating the percentage error in D (Derr) and K (Kerr). Also, diffusivity anisotropy (Danis) and diffusional kurtosis anisotropy (Kanis) were estimated. Agreements between the two estimation methods were assessed by Bland–Altman plots. The Spearman rank correlation test was used to study the correlations between Kerr/Derr and Danis/Kanis. Results The median (95% confidence interval) Kerr and Derr were 5.1% (0.8%, 32.6%) and 1.7% (−2.5%, 5.3%), respectively. A significant relationship was observed between Kerr and Danis (correlation coefficient R = 0.694, P < 0.0001), as well as between Kerr and Kanis (R = 0.848, P < 0.0001). Conclusions In HN cancer, the fit of the DKI model to TW images can introduce bias and error in the estimation of K and D, which may be non‐negligible for single lesions, and should h...

show abstract

“…Before the initiation of the trial, a system for reporting bpMRI was developed based on our development work and previous studies . Suspicious lesions in bpMRI were reported by one reader (I.J.)…”

Section: Methodsmentioning

confidence: 99%

Novel biparametric MRI and targeted biopsy improves risk stratification in men with a clinical suspicion of prostate cancer (IMPROD Trial)

Jambor

Boström

Taimen

et al. 2017

Magnetic Resonance Imaging

Self Cite

View full text Add to dashboard Cite

show abstract

Diffusion-weighted imaging of prostate cancer: effect of b-value distribution on repeatability and cancer characterization

Cited by 25 publications

References 36 publications

Repeatability of radiomics and machine learning for DWI: Short‐term repeatability study of 112 patients with prostate cancer

Repeatability of radiomics and machine learning for DWI: Short‐term repeatability study of 112 patients with prostate cancer

Diffusional kurtosis imaging in head and neck cancer: On the use of trace‐weighted images to estimate indices of non‐Gaussian water diffusion

Novel biparametric MRI and targeted biopsy improves risk stratification in men with a clinical suspicion of prostate cancer (IMPROD Trial)

Contact Info

Product

Resources

About