Synopsis

A novel clinical protocol for prostate diffusion imaging is presented. The commonly applied strategy of averaging MR signal at a small number of b-values is replaced by the acquisition of a whole range of b-values in conjunction with a model fitting post precessing step. With no increase in acquisition time, our results show that the same image quality can be achieved. As a main benefit, modelling is made more consistent and largely b-value independent for both the simple ADC model, as well as for more complex signal representations such as the kurtosis model.

Introduction

Magnetic resonance diffusion measurements are usually time consuming as the effect of diffusion weighting results in a low signal to noise ratio (SNR). Therefore, a commonly applied strategy for clinical patient examinations in prostate (1,2) is to only probe a few different b-values and to use signal averaging. While this procedure is suitable for generating maps of the apparat diffusion coefficient, the lack of variety in the acquired b-values is a hinder for the application of higher order signal representations, such as the truncated kurtosis model given by S(b)=S₀⋅exp(−bADC_K+b²K·ADC_K²/6) and leads to a bias in ADC values depending on the exact b-values probed (3).


Here we suggest a procedure where 21 linearly spaced b-values over a wider range are probed instead without any averaging applied. Using the same amount of orthogonal directions as in the clinical procedure, the total amount of images stays more or less constant. The lack of averaging leads to a low SNR for a single b-value image (see Fig.1). However, synthesized images at arbitrary b-values can be computed in a post-processing step where the model-driven denoising tool OBSIDIAN (4) is applied. The quality of the synthesized images is comparable to those from the clinical protocol. The main advantage of the suggested procedure is that model parameters, such as the basic ADC and ADC_K, can be determined in a more reproducible way and are not relying on the exact choice probed b-values (3).

Methods

Simulated data was generated for two tissue type models, representing normal and cancerous prostate tissue, based on a biexponential function. Both have a fast diffusion component D₁ of 2.2 μm²/ms, while the slow diffusion parameter D₂ and the fast signal fraction f differ, i.e., D₂ = 0.4 μm²/ms and f = 0.8 in the normal case and D₂ = 0.2 μm² /ms and f = 0.6 for the cancerous case. Simulation data consisted of signals at N = 21 b-values, linearly spaced in the range of 0 to b_max. For both the normal and the cancerous case, 60 000 signal realizations at different SNR levels were generated by adding Gaussian noise with a standard deviation of σ = 1. These profiles were fitted with biexponential, kurtosis and gamma distribution model functions allowing the fitted signal S_fit(b) to be calculated. The averaging effect (AE) for certain b-values was then calculated as AE(b)=(σ/STD(S_fit(b)))², where STD represents the standard deviation over all realizations. A synthetic ADC termed trueADC was computed by means of linear regression for S_fit(b) at b-values of 0, 100, 1000 and 1500 s/mm² as used in the clinical protocol.


Informed consent of the patient was received prior to imaging. Clinical data from the 21 b-value protocol was processed with OBSIDAN (4) employing the kurtosis model. From the resulting parameter maps images at the 4 standard clinical b-values were calculated and used to compute trueADC maps by means of linear regression in each pixel.

Results

The averaging effect across the measured b-value range for different b_max values is shown in Fig.1. In Fig.2 clinical ADC, kurtosis ADC and trueADC are compared for simulated signal. Comparison of clinical data, synthesized images and raw signals is shown in Fig.3. In Fig.4 clinical and computed ADC maps are shown.

Discussion

As shown in Fig.1, a considerable AE results from model fitting of the simulated MR signal, particularly in the middle part of the chosen b-range. As images at b=1000 s/mm² and 1500 s/mm² are of particular importance in clinical practise, a b_max of 2000 s/mm² appears to be a good choice as the AE is high for the clinically relevant b-values. Higher b_max values generally produce a higher AE, however, at the cost of a longer echo time and an increased signal bias (see Fig.3 in reference (4)). For simulations, trueADC is equivalent to clinical ADC, while ADC_K is larger and more noisy, which is not surprising due to the higher complexity of the model.


The quality of synthesized images is on-par with the images obtained from the clinical protocol. Furthermore, the improvement in image quality by model fitting with respect to the raw data is apparent. With respect to the ADC maps, ADC_K is more noisy in comparison to the clinical and the synthesized ADC map (trueADC) in agreement with the simulations. The trueADC map is of equal quality as the clinical ADC map and can be computed for arbitrary b-value choices.

Conclusion

A novel approach for clinical acquisition and processing of prostate is presented where a large range of b-values is probed without the need for averaging. Preliminary results show that image quality is on-par with the six fold averaged clinical images without an increase in acquisition time. In principle, the dense coverage of a whole b-range, gives more reproducible trueADC values, since b-value settings can be modified after the acquisition. Furthermore, the new approach can account for more complex signal representations, such as the kurtosis model. Further work is under way to support these claims and which in turn might lead to a paradigm shift in prostate diffusion acquisition.

Acknowledgements

Barncancerfonden; the Swedish Cancer Society; the Swedish Research Council; The Swedish state under an agreement between the Swedish government and the country councils (the ALF agreement).

References

1. Weinreb JC, Barentsz JO, Choyke PL, Cornud F, Haider MA, Macura KJ, et al. PI-RADS Prostate Imaging – Reporting and Data System: 2015, Version 2. Eur Urol. 2016;69(1):16–40.

2. Rosenkrantz AB, Padhani AR, Chenevert TL, Koh D-M, Keyzer FD, Taouli B, et al. Body diffusion kurtosis imaging: Basic principles, applications, and considerations for clinical practice. J Magn Reson Imaging. 2015;42(5):1190–202.

3. Maier SE, Wallström J, Langkilde F, Johansson J, Kuczera S, Hugosson J, et al. Prostate Cancer Diffusion-Weighted Magnetic Resonance Imaging: Does the Choice of Diffusion-Weighting Level Matter? J Magn Reson Imaging. 2021;

4. Kuczera S, Alipoor M, Langkilde F, Maier SE. Optimized bias and signal inference in diffusion-weighted image analysis (OBSIDIAN). Magn Reson Med. 2021;86(5):2716–32.

5. Richter PH. Estimating errors in least-squares fitting. Telecommun Data Acquis Rep. 1995; Available from: https://ntrs.nasa.gov/citations/19960009481